JP2010074645A - Defective pixel determination system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a defective pixel determination system capable of accurately determining whether or not a radiation detection element is a defective pixel and especially turning a radiograph detector to a state capable of accurately detecting the lesion of microcalcification in mammography. <P>SOLUTION: The defective pixel determination system 30 includes the radiograph detector 1 provided with the plurality of radiation detection elements (x, y), image data acquisition means 4, 5 and 6 and a communication means 3, a radiation generator 34, and a defective pixel determination device 31. When a signal instructing determination start is inputted, the defective pixel determination device 31 irradiates the radiograph detector 1 with radiation from the radiation generator 34 in the state without the presence of an object, makes respective actual image data F(x, y) outputted from the respective radiation detection elements (x, y) of the radiograph detector 1 be transmitted, and when the number of times that the actual image data F(x, y) exceeds a preset threshold V'th becomes equal to or more than the preset number of times, determines that the radiation detection element (x, y) as the defective pixel. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a defective pixel determination system.

  In a radiation image detector in which a solid-state imaging device called a so-called flat panel detector (FPD) is two-dimensionally arranged, a photoconductive substance such as a-Se (amorphous selenium) is used as the detection device. Radiation energy is directly converted into electric charge, and this electric charge is read out as an electric signal for each pixel by a signal reading switch element such as TFT (Thin Film Transistor) arranged two-dimensionally. An indirect system that converts light into light by a scintillator or the like, converts the light into electric charge by a photoelectric conversion element such as a two-dimensionally arranged photodiode, and reads it as an electric signal by a TFT or the like is well known.

  In either method, it is necessary to correct the photographed image data by performing gain correction, offset correction, or the like on the photographed image data obtained by detecting the radiation transmitted through the subject by the radiation image detector. It is known.

In general, in the correction of actual image data, as shown in the following equation (1), the actual image data output from each radiation detection element (the coordinates in the sensor panel unit are (x, y)) of the radiation image detector. By subtracting the offset correction value O (x, y) from F (x, y) and multiplying the difference by the gain correction value G (x, y), the final image data F _O (x, y) is obtained. Correction is performed in such a way as to obtain. Note that the image data F _O (x, y) generated based on the actual image data F (x, y) output from one radiation detection element of the radiation image detector is a captured image (that is, a radiation image). Since this corresponds to image data for one pixel, the radiation detection element may be referred to as a pixel hereinafter.
F _O (x, y) = (F (x, y) −O (x, y)) × G (x, y) (1)

  As described above, in the correction of the actual image data, since it is necessary to obtain the offset correction value O (x, y) and the gain correction value G (x, y), the radiological image detector is periodically calibrated. In general, the offset correction value O (x, y) and the gain correction value G (x, y) whose characteristics may change with time are updated.

However, in this method, the temperature of each element in the radiographic image detector when calibration is performed matches the temperature of each element in the radiographic image detector when radiographic imaging is performed. As a precondition, if the temperature of each element in the radiation image detector at the time of calibration differs from the temperature of each element in the radiation image detector at the time of radiographic imaging, it is temperature dependent. There is a problem that the offset correction value O (x, y) having a high characteristic is deviated from an appropriate value and the SN ratio of the final image data F _O (x, y) is deteriorated.

  In order to solve this problem, an output value (hereinafter referred to as a dark read value D (x, y) from each radiation detection element in a state in which radiation is not irradiated immediately before or immediately after imaging for each radiographic imaging. The dark read value may be referred to as a dark image.) Is detected, and the offset correction value O (x, y) in the radiographic image capturing may be calculated. This is a process for obtaining the offset correction value O (x, y) under the same temperature conditions as possible as the temperature characteristics of the radiation detection element at the time when the actual image data F (x, y) is obtained by radiographic imaging. is there.

  However, as in the case of acquiring the actual image data F (x, y), when acquiring the dark read value D (x, y), various electric noises, that is, the dark current noise of the photodiode, the TFT transient, etc. Noise, TFT thermal noise, TFT leak noise, thermal noise caused by parasitic capacitance of the data line that reads charges from the TFT, amplifier noise inside the readout circuit, quantization noise caused by A / D conversion, etc. are affected. Even if the dark read value D (x, y) is read under temperature conditions, the dark read value D (x, y) has fluctuations (variations) in signal values due to these electrical noises.

  Therefore, even if the dark reading value D (x, y) is read immediately before or after radiographic imaging, the read dark reading value D (x, y) is offset, that is, offset correction in imaging conditions such as the temperature condition. It cannot necessarily be said that it is the true value of the value O (x, y). Therefore, dark reading is performed a plurality of times immediately before and after radiographic imaging to calculate the average value of each dark reading value D (x, y), and the average value is adopted as the offset correction value O (x, y). Is often performed (see, for example, Patent Documents 1 to 3).

This is to calculate the average value of the dark read values D (x, y) read out a plurality of times under the same temperature conditions as the temperature characteristics of the radiation detection element or the like in the radiographic imaging immediately before or after the radiographic imaging. For example, since the fluctuation of the dark reading value D (x, y) is reduced or offset, the average value is equal to or at least equal to the true value of the offset correction value O (x, y) under the photographing conditions. This is based on the idea of close values. Then, if the photographed image data F (x, y) is corrected using the offset correction value O (x, y) that is the average value, the SN of the final image data F _O (x, y) after correction is corrected. The ratio can be good.

  On the other hand, image processing such as correction processing such as offset correction has been conventionally performed by a processing device such as an image processor or a console that is different from an imaging device such as a radiation image detector (for example, , See Patent Document 4). In recent years, portable radiographic image detectors have been developed that have built-in batteries and that transmit and receive live-action image data F (x, y) and the like with an external processing device or the like wirelessly without using a cable. (For example, see Patent Document 5).

  In addition, in the radiation image detector, defective pixels that cannot output normal pixel values may occur during the manufacturing process. For example, as an example of a defective pixel, a pixel that constantly outputs a large pixel value (for example, a pixel whose signal value always reaches a saturation level), or a pixel that constantly outputs only a small pixel value (for example, There are pixels that always output only a zero level signal value), and pixels that always output only a constant signal value (a pixel whose output pixel value does not change even when irradiated with radiation).

  These defective pixels may occur during the TFT manufacturing process, or may occur when a photoelectric conversion element is formed on the TFT. The radiation image detector is required to appropriately change the pixel value in proportion to the irradiated radiation dose, but in these defective pixels, a normal pixel value change corresponding to the irradiated radiation dose cannot be expected.

If such a defective pixel occurs in the diagnostic image, it may cause a misdiagnosis or interfere with the diagnostic action. If there is a defective pixel, the defective pixel is interpolated using surrounding pixel values. It is required to perform processing so that defective pixels cannot be visually recognized in the final image data F _O (x, y).

  In order to perform such correction processing, the pixel position of the defective pixel must be known in advance. Such defective pixels are detected by performing post-manufacturing shipping inspections or inspections involving radiation irradiation during regular calibration, and the pixel positions of the detected defective pixels are registered in the defective pixel map ( For example, see Patent Document 6). This defective pixel map is constructed for each radiographic image detector, and thereafter operated and managed in association with individual radiographic image detectors.

When performing the above-described offset correction and gain correction, the defective pixel map is also referred to and managed so that the pixel value of the defective pixel is not erroneously used for offset correction or gain correction.
US Pat. No. 5,452,338 US Pat. No. 6,222,901 U.S. Pat. No. 7,041,955 JP-A-11-113889 JP-A-7-140255 Japanese Patent Laid-Open No. 2001-8198

  As an example of a defective pixel, a pixel that constantly outputs a large pixel value (for example, a pixel whose signal value always reaches a saturation level), or a pixel that constantly outputs only a small pixel value (for example, a signal value that always remains) Is a pixel that outputs only zero level values), and pixels that always output only a constant signal value (pixels whose output pixel values do not change even when irradiated with radiation) are increased, but these defective pixels are very It can be easily detected.

However, although not belonging to any of the above, there is a pixel (hereinafter referred to as an abnormal pixel) that outputs an abnormal signal value with a certain probability. Since these pixels are not registered in the defective pixel map, there is a risk that a pixel having an abnormal value is generated in the final image data F _O (x, y) with a certain probability, which hinders diagnosis. It was.

  Further, in the defective pixel determination method so far, when the defective pixel is determined, the defective pixel is detected using the image information irradiated with radiation. There has been a problem that the number of images used for determining defective pixels is limited to several to at most ten. Therefore, even if a defective pixel that outputs an abnormal value on a regular basis can be detected, an abnormal pixel that occurs with a certain probability cannot be captured, or when a defective pixel is determined using statistical values such as standard deviation, It was difficult to obtain good statistics. For this reason, if the number of times of radiation irradiation is increased, the productivity is lowered and the production cost is increased.

  Also, such defective pixels and abnormal pixels tend to increase over time, so it is not safe to register defective pixels at the time of shipment from the factory. When a new defective pixel or abnormal pixel is found after calibration, the pixel position is preferably registered in the defective pixel map or appropriately processed as an abnormal pixel.

  On the other hand, by displaying image data based on the pixel output value for pixels that have returned from an abnormal state to a normal state, a higher-definition image can be provided compared to the interpolated image data of surrounding pixels.

  However, the determination of whether or not the pixel is an abnormal pixel every time imaging is performed requires a processing time, and it takes a long time to display image data for final diagnosis.

  In particular, regarding the detection of microcalcification lesions in mammography, the size of the lesion is close to the pixel level. Therefore, when mammography is performed, the value of the actual image data output from the radiation detection element and the surrounding radiation detection elements Even if the value of the actual image data that is output is significantly different, it is determined whether it is a result of photographing an actual microcalcification lesion existing in the lesion or just an abnormal value being output. This is difficult, and it has been difficult to determine whether or not the radiation detection element can be regarded as an abnormal pixel based on actual image data obtained by mammography.

  The present invention has been made to solve the above-described problems, and accurately determines whether or not a radiation detection element is a defective pixel. In particular, a radiographic image detector detects a microcalcification lesion in mammography. It is an object of the present invention to provide a defective pixel determination system that can be accurately detected.

In order to solve the above problem, the defective pixel determination system of the present invention includes:
A plurality of radiation detection elements arranged two-dimensionally;
Image data acquisition means for acquiring real image data based on radiation irradiation from the plurality of radiation detection elements;
A communication means for transmitting the photographed image data;
A radiation image detector comprising:
A radiation generator for irradiating the radiation image detector with radiation;
When a signal instructing the start of determination is input, radiation is emitted from the radiation generation device to the radiation image detector in a state where no subject is present, and is output from each radiation detection element of the radiation image detector. A defective pixel determination device that transmits each of the actual image data, and determines whether or not each of the radiation detection elements of the radiological image detector is a defective pixel based on each of the actual image data;
With
In the defective pixel determination device, the number of times that the actual image data output from the radiation detection element of the radiation image detector becomes an abnormal value exceeding a preset threshold is equal to or greater than the preset number. In this case, the radiation detection element is determined as a defective pixel.

  According to the defective pixel determination system of the method of the present invention, when the number of times determined to be abnormal in the past has become a predetermined number or more, the abnormal pixel is registered as a defective pixel in the defective pixel map, and thereafter Since it is not used for offset correction value generation or final image data generation related to the pixel, the time required for processing can be shortened. In mammography or the like, all pixels that have been determined to be abnormal pixels even once in the past are regarded as defective pixels (threshold is set once).

  When the lesion is a lesion such as microcalcification that is approximately the same size as the pixel size of the radiation image detector as in mammography, the part information included in the imaging order information includes defective pixel discrimination before imaging. (The threshold is automatically set once), the engineer can determine whether the radiological image detector can be used for imaging the lesion, and if there is no alternative detector, It is possible to maintain diagnostic accuracy by devising the arrangement of radiation image detectors at the time of imaging such that defective pixels are outside the region of interest.

  In live-action image data in which a microcalcification lesion of approximately the same size as the pixel size obtained by mammography is taken, if there is a pixel whose data value is significantly different from the surrounding pixels, it exists in the lesion Although it is difficult to distinguish whether an actual micro lime image is taken or an abnormal value is output, in the defective pixel determination system of the system as in the present invention, a subject is detected with respect to a radiological image detector. In a state in which no radiation exists, the radiation image detector is uniformly irradiated with radiation, and actual image data is output from each radiation detection element. In such a situation, if there is no abnormal pixel or defective pixel, the actual image data with almost the same signal value should be output from each radiation detection element. In such a situation, the surrounding pixels are the data values. If there is a pixel that greatly differs, it can be accurately determined that it is an abnormal pixel.

  Therefore, if the part to be imaged is mammography for microcalcification diagnosis, the abnormal pixel determination method of the present invention is performed prior to imaging, and the presence or absence of abnormal pixels is confirmed. Can safely determine the presence / absence (diagnosis) of a microcalcified lesion based on the captured image. On the other hand, if there is an abnormal pixel, the engineer sends the abnormal pixel to the radiation image detector that is the region of interest. The diagnostic accuracy can be maintained by adjusting the position of the radiological image detector so that there is no image or selecting a new radiological image detector without using the radiological image detector.

  Hereinafter, an embodiment of a defective pixel determination system according to the present invention will be described with reference to the drawings. However, the present invention is not limited to the following illustrated examples.

[Basic configuration of radiation image detector]
First, a basic configuration of a radiation image detector used in the defective pixel determination system according to the present invention will be described.

As shown in FIG. 1, the radiation image detector (flat panel detector) 1 includes a housing 2 that protects the inside, and radiates irradiated radiation to the inside of the radiation incident surface X of the housing 2. A scintillator layer (not shown) for conversion into (1) is formed. As the scintillator layer, for example, a layer formed using a phosphor in which a luminescent center substance is activated in a matrix such as CsI: Tl, Gd ₂ O ₂ S: Tb, ZnS: Ag, or the like can be used.

  As shown in the equivalent circuit diagram of FIG. 2, on the surface side opposite to the surface on which the radiation of the scintillator layer is incident, there are a plurality of elements that convert light output from the scintillator layer into electrical signals as radiation detection elements. The sensor panel section 4 is provided in which the photodiodes 14 are two-dimensionally arranged. The charge (signal value) output from one photodiode 14 forms one pixel. Further, as will be described in detail later, each photodiode 14 is connected to a TFT 15 which is a signal reading switch element.

  In the following, the case where the so-called indirect radiation image detector 1 that converts radiation into light by the scintillator layer and detects it by a photoelectric conversion element such as a photodiode as described above will be described. In addition to this, it is also possible to use a so-called direct-type radiation image detector that directly converts the radiation incident on the detection element without passing through the scintillator layer into an electrical signal. The invention can be applied.

  Hereinafter, detection elements used in these types of radiation image detectors are collectively referred to as radiation detection elements. That is, for example, in the indirect radiation image detector 1 as in the present embodiment, the radiation detection element includes one photodiode 14, a TFT 15 connected thereto, and a portion corresponding to the photodiode 14 in the scintillator layer. For example, a direct radiation image detector includes a detection element and a switch element such as a TFT connected thereto.

  The radiation image detector 1 includes a battery 21 (see FIG. 2). As shown in FIG. 1, in this embodiment, an antenna device 3 that is a wireless communication unit is embedded in the battery replacement lid member 10 provided on the side surface portion of the housing 2 of the radiation image detector 1. Is provided. Furthermore, a power switch 11 of the radiation image detector 1 and indicators 12 for displaying various operation statuses are provided on the side surface of the housing 2.

  As shown in FIG. 2, in the vicinity of the sensor panel unit 4, a reading unit 5 that reads an output value of each radiation detection element of the sensor panel unit 4 is provided. The reading unit 5 includes a control unit 6 including a microcomputer, a storage unit 7 including a ROM (Read Only Memory) and a RAM (Random Access Memory), a flash memory, a scan driving circuit 8, a reading circuit 9, and the like. ing.

  As shown in FIG. 3, the plurality of radiation detection elements arranged two-dimensionally on the sensor panel unit 4 respectively have a position x in the row direction and a position y in the column direction of the radiation detection elements in the sensor panel unit 4. The coordinates (x, y) for each component are assigned in advance as the number (x, y) of the radiation detection element. Hereinafter, when individual radiation detection elements are specified, they are referred to as radiation detection elements (x, y).

  In FIG. 3, 8 × 16 radiation detection elements (x, y) are illustrated, but this is a simplified representation, and actually more radiation detection elements (x, y). ) Are two-dimensionally arranged and assigned numbers. In addition, the coordinates (x, y) (that is, the radiation detection element number (x, y)) may be referred to as a pixel number (x, y) or a pixel position (xy).

  The configurations of the sensor panel unit 4 and the reading unit 5 will be further described. As shown in the equivalent circuit diagram of FIG. 2, one electrode of each radiation detection element (x, y) of the sensor panel unit 4 is used for signal readout. The source electrode of the TFT 15 which is the switch element is connected. In addition, a bias line Lb is connected to the other electrode of each radiation detection element (x, y), and the bias line Lb is connected to a bias power supply 16, and each radiation detection element (x, y) is connected from the bias power supply 16. A bias voltage is applied to y).

  The gate electrode of each TFT 15 is connected to a scanning line Ll extending from the scanning drive circuit 8, and the drain electrode of each TFT 15 is connected to a signal line Lr. Each signal line Lr is connected to an amplifier circuit 17 in the readout circuit 9, and an output line of each amplifier circuit 17 is connected to an analog multiplexer 19 via a sample hold circuit 18. The analog multiplexer 19 is connected to an A / D converter 20, and the reading circuit 9 is connected to the control means 6 via the A / D converter 20. A storage means 7 is connected to the control means 6.

  In the radiation image detector 1, in radiation image capturing for capturing a subject (not shown), when radiation transmitted through the subject enters the scintillator layer, light is irradiated from the scintillator layer to the sensor panel unit 4, and the amount of light irradiation received Accordingly, electric charges are accumulated in the radiation detection element (x, y).

  When the radiographic image capturing is completed and the actual image data is read from the radiographic image detector 1 as an electrical signal, a read voltage is applied from the scanning line Ll to the gate electrode of the TFT 15 to open the gate of each TFT 15, The charge accumulated from the detection element (x, y) through the TFT 15 is taken out as an electric signal to the signal line Lr. Then, the electric signal is amplified by the amplifier circuit 17 and output from the analog multiplexer 19 to the control means 6 via the A / D converter 20 in sequence.

  The control means 6 associates the amplified electrical signal output from the radiation detection element (x, y) with the number (x, y) of the radiation detection element (ie, pixel) described above, and captures the actual image data F (x, y). It is stored in the storage means 7 as y).

  By sequentially scanning the scanning lines Ll to which the readout voltage is applied to the TFT 15 and performing the above readout processing for each scanning line Ll, electrical signals are respectively read from all the radiation detection elements (x, y) of the sensor panel unit 4, Each electric signal is associated with a pixel number (x, y) and stored in the storage means 7 as each photographed image data F (x, y).

  As described above, the radiation image detector 1 includes the sensor panel unit 4 in which a plurality of radiation detection elements (x, y) are two-dimensionally arranged, the control means 6, the scanning drive circuit 8, the readout circuit 9, and the like. The read unit 5 or the like forms image data acquisition means for acquiring the actual image data F (x, y).

  Further, the image data acquisition means of the radiation image detector 1 not only acquires the actual image data F (x, y) as described above, but also performs dark reading.

  In the dark reading, after resetting a plurality of radiation detection elements (x, y) of the radiation image detector 1 to release charges, the gate of each TFT 15 is closed and the radiation image detector 1 is not irradiated with radiation. Keep on. Then, after a predetermined time has elapsed (usually, after the same time as the charge accumulation time in the radiation detection element (x, y) during radiation irradiation), a read voltage is applied from the scanning line Ll to the gate electrode of the TFT 15 to The gate of the TFT 15 is opened, the charge (dark charge) accumulated in each radiation detection element (x, y) is taken out to the signal line Lr, and the output value is amplified by the amplifier circuit 17 in the same manner as described above. Are sequentially output to the control means 6 via the A / D converter 20.

  In this way, the electrical signal output from each radiation detection element (x, y) that is not exposed to radiation is the dark read value. The control means 6 stores each electrical signal output from each radiation detection element (x, y) in association with each pixel number (x, y) as a dark read value D (x, y) in the storage means 7. It is supposed to be. The scanning line Ll for applying the read voltage to the TFT 15 is sequentially scanned, and the dark read value D (x, y) is read from all the radiation detection elements (x, y).

  As described above, in the radiation image detector 1, each radiation detection element (x, y) is calibrated in order to grasp the output value characteristic variation from each radiation detection element (x, y). In a normal case, in the calibration, the radiation image detector 1 is irradiated with radiation without a subject being present, and based on the actual image data F (x, y) output from each radiation detection element (x, y). A gain correction value is calculated for each radiation detection element (x, y).

  In the calibration, dark reading is also performed a plurality of times, and in the dark reading, the dark reading value D (x, y) is output from each radiation detection element (x, y) without irradiating the radiation image detector 1 with radiation. Then, an offset correction value is calculated for each radiation detection element (x, y) based on the output dark read value D (x, y).

  Furthermore, in order to obtain an offset correction value under the same imaging conditions as the imaging conditions such as the temperature characteristics of the radiation detection element (x, y) at the time when the actual image data F (x, y) is obtained by radiographic imaging, For each radiographic image capture, the dark read value D (x, y) output from each radiation detection element (x, y) is detected a plurality of times in a state where no radiation is irradiated immediately before or immediately after the radiographing, and the average value thereof is detected. As described above, may be calculated as an offset correction value in radiographic imaging.

[Principle of offset correction value acquisition of each radiation detection element of the radiation image detector]
Prior to the description of the defective pixel determination system according to the present invention, the principle of acquiring the offset correction value of each radiation detection element (each pixel) of the radiation image detector in image generation using the radiation image detector will be described.

  In order to correct the photographed image data F (x, y) output from each radiation detection element (x, y) according to the above equation (1), an offset correction value (hereinafter referred to as O (x, y)). The true value of is required. However, as described above, the data actually obtained by performing the dark reading is the dark reading value D (x, y) that fluctuates (fluctuates or has an error) due to the influence of the electric noise or the like described above. The true value of the offset correction value O (x, y) cannot be obtained directly.

  Therefore, conventionally, as described above, a plurality of darks output from one radiation detection element (x, y) by performing dark reading multiple times in succession immediately before and immediately after radiographic imaging. A method of calculating an average value of the read values D (x, y) and mitigating or canceling fluctuations and setting the average value as an offset correction value O (x, y) was adopted.

Specifically, assuming that dark reading is performed K times, as shown in FIG. 4, the dark reading value output from the radiation detection element (x, y) in the k-th (k = 1 to K) dark reading is obtained. Assuming D _k (x, y), for example, as shown in FIG. 5, each dark read value D _k (x, y) output from the radiation detection element (x, y) in each dark reading is an average value D. It is known to fluctuate so as to form a normal distribution of _kave (x, y) and standard deviation σ _Dk (x, y). This is because when the dark reading is repeatedly performed on the one radiation detection element (x, y), the obtained dark reading value fluctuates (fluctuates) at every reading even though the pixel position is the same. Therefore, this fluctuation is defined as “temporal fluctuation”. The standard deviation of the fluctuation distribution (standard deviation σ _Dk (x, y) in the case of the dark reading value D _k (x, y)) is usually used as a statistical index representing the magnitude of fluctuation.

Further, since the average value D _kave (x, y) is an average value on the time axis of each dark reading value D _k (x, y), this average value is defined as a “temporal average value”. If the average value D _kave (x, y) is used as the offset correction value O (x, y) of the radiation detection element (x, y), the value of each dark read value D _k (x, y) Since temporal fluctuations can be reduced or offset, in general, the temporal average value of each dark reading value D _k (x, y) may be used as the offset correction value O (x, y). Many. The offset correction value O (x, y) is shown in the following equation (2).

The dark reading value D _k (x, y) fluctuates with the standard deviation σ _Dk (x, y). When this is averaged K times, the fluctuation magnitude is generally 1 / √K. That is, the offset correction value O (x, y) has a distribution having a fluctuation of standard deviation {1 / √K · σ _Dk (x, y)}. Therefore, the above equation (2) indicates that the value of fluctuation 1 / √K is used as the offset correction value rather than using the dark read value D _k (x, y) itself as the offset correction value. Show.

  However, when the offset correction value O (x, y) is calculated in this way, dark reading and transmission of the dark reading value must be performed a plurality of times, and thus there is a problem that power is consumed. That's right.

  In order to avoid this problem, a plurality of dark readings are performed at the time of calibration, and the offset correction value O (x, y) calculated in advance using the above equation (2) is used. When the temperature of each element in the radiographic image detector at the time of performing and the temperature of each element in the radiographic image detector at the time of performing radiographic imaging are different, the offset correction value O (x, y) is appropriate. As described above, there is a problem of deviating from a certain value.

  Therefore, a description will be given of an offset correction method together with a calibration method and a temperature compensation method.

  Since the calibration method and the temperature compensation method taken up in this description are also used for defective pixel determination according to the present invention, the basic concept thereof will be discussed in the following description. The offset correction method is described by performing dark reading at least once just before or immediately after radiographic imaging, even if the dark reading is performed only once. , Y) is used. However, the defective pixel determination according to the present invention is not limited to the offset correction method described below.

  This method (a method of calculating an offset correction value O (x, y) with high accuracy even if dark reading is performed at least once before or immediately after radiographic imaging, and even if dark reading is performed only once. ) Is as follows.

First, at the time of calibration, for each pixel position (x, y) corresponding to each radiation detection element (x, y), a plurality of dark images d _k (x, y) under a predetermined temperature condition without temperature change. (Hereinafter, the dark reading value d _k (x, y) is distinguished from the dark reading value D (x, y) obtained by one dark reading performed immediately before or immediately after radiographic imaging as described above.) And an offset correction value δ (x, y) (hereinafter, finally obtained offset correction value O (x, y)) using a plurality of acquired dark images d _k (x, y). In order to distinguish from y), an offset correction value δ (x, y) is obtained). It is assumed that the temperature characteristics of each pixel position (x, y) are the same while the dark image d _k (x, y) is acquired.

  At this time, a temperature compensation variable is defined to represent the amount of change accompanying the temperature change of the signal value at each pixel position (x, y) when the offset correction value δ (x, y) is obtained, and this is defined for each pixel. Calculate and store for each position (x, y).

  Here, the positioning of the temperature compensation variable will be described. The temperature compensation variable is not intended to measure the temperature itself of the radiation detection element (x, y) at each pixel position (x, y). What we want to know is not the temperature itself of each radiation detection element (x, y), but the signal at each pixel position (x, y) due to the temperature change of each element such as the radiation detection element (x, y) or an electric circuit. Because the value has changed (for example, even if the temperature of each radiation detection element (x, y) itself changes), the signal value of the pixel corresponding to that radiation detection element (x, y) changes. If it is not, there is no problem even if it is considered that there was no temperature change of the radiation detection element (x, y).), By the temperature change of each element such as the radiation detection element (x, y) or an electric circuit, It is desirable that the variable be a variable that quantitatively suggests how the signal value at each pixel position (x, y) has changed.

  Next, just before or immediately after radiographic imaging, dark reading is performed only once to obtain a dark reading value D (x, y). At this time, the temperature compensation variable similar to the above is calculated and stored for each pixel position (x, y) for the dark read value D (x, y).

Next, the temperature compensation variable at each pixel position (x, y) when the offset correction value δ (x, y) is obtained and the temperature at each pixel position (x, y) of the dark read value D (x, y). Comparing the compensation variables, based on the result, the offset correction value δ (x, y) obtained in advance is converted into the offset correction value O (x, y) in the temperature environment when the radiation imaging is actually performed. The final image data F _O (x, y) is used for calculation.

  Normally, when the offset correction value O (x, y) acquired at the past calibration is used, the dark reading value D (x, y) acquired immediately before or after the radiographic image capturing is affected by temperature fluctuation. This is often used as the offset correction value O (x, y) (that is, O (x, y) = D (x, y)). On the other hand, in this method, the dark read value D (x, y) acquired every time radiographic imaging is not directly used as the offset correction value itself, but the temperature at each pixel position in the temperature environment when radiography is performed. The point used for calculation of the compensation variable is different from the conventional method. Then, instead of using the offset correction value δ (x, y) calculated in advance using a plurality of dark images as it is, using the above-described temperature compensation variable, the time point when the radiographic image is actually taken is reached. The point that it is used as the offset correction value O (x, y) after performing the temperature correction is a point that the conventional method does not have.

  The advantages of this method are as follows.

First of all, an offset correction value δ (x, y) serving as a reference for calculating an offset correction value O (x, y) used to obtain final image data F _O (x, y). However, since it is calculated from a plurality of dark images d _k (x, y) under a predetermined temperature condition that can be regarded as having no temperature change, error components (temporal fluctuations in signal values) such as electrical noise are canceled out. In addition, the value is close to the true offset correction value. Therefore, the final image data F with a small correction error and a good S / N ratio as compared with the case where the single dark read value D (x, y) itself acquired at the time of radiographic imaging is used as the offset correction value. _O (x, y) can be obtained.

Secondly, by calculating a temperature compensation variable for each pixel position (x, y), pixel value variations due to temperature changes are individually corrected for each pixel position (x, y). No matter how the signal value of the pixel position (x, y) is affected by the temperature change, this is corrected, and final image data F _O (x, y) having a good SN ratio is obtained. is there. This temperature correction for each pixel can be performed by acquiring a dark image only once.

  What is important here is what is used for the temperature compensation variable for each pixel position (x, y). Although it is sufficient if a temperature sensor can be built in each pixel position (x, y) in advance, this method is not practical. Although it is conceivable to incorporate a realistic number of temperature sensors in the radiation image detector 1, it is not possible to achieve accurate temperature correction for each pixel position (x, y). What is desired is not the temperature itself of the radiation detection element (x, y) but the temperature change of each element such as the radiation detection element (x, y) or an electric circuit at each pixel position (x, y). It is how the signal value has changed.

From the above, the requirements for temperature compensation variables are considered to be the following four items.
Requirement 1) Variable required for each pixel unit (or for each small area unit close to a pixel).
Requirement 2) A variable that can grasp changes in signal value due to temperature changes.
Requirement 3) A variable obtained from the dark image D (x, y).
Requirement 4) Variables with little temporal fluctuation.

  As a temperature compensation variable that satisfies the requirements of these four items, the present inventor has calculated the average value (or median value, etc.) of the pixel values of the peripheral pixels with respect to the pixel of interest (x, y) of the dark read value D (x, y). I paid attention to.

  The result of each element such as the radiation detection element (x, y) and the electric circuit being affected by the temperature change is reflected in the dark read value D (x, y). What we want to know is how the signal value of each pixel has changed due to the temperature change of each element such as the radiation detection element (x, y) and the electric circuit. It makes sense to use (x, y) and satisfies requirements 1-3.

  Next, whether or not the selected temperature compensation variable satisfies requirement 4 will be described with reference to FIG.

  As shown in FIG. 6, for example, as a plurality of radiation detection elements (x, y), the one radiation detection element (x, y), that is, in a 5 × 5 square region centered on the pixel position (x, y). The signal values at a plurality of pixel positions (x−2, y−2) to (x + 2, y + 2) existing in are used for temperature compensation variable calculation. A method of selecting a radiation detection element (x, y) other than the one radiation detection element (x, y) will be discussed later.

However, the dark read values D (x−2, y + 2) output from the 25 radiation detection elements (x−2, y−2) to (x + 2, y + 2) including the one radiation detection element (x, y). The distribution of temporal fluctuations of the signal values y-2) to D (x + 2, y + 2) is not always the distribution as shown in FIG. 5, but actually, for example, as shown in FIG. to, the average value mu _D and the standard deviation sigma _D is different distributions are common.

In this method, as shown in the following equation (3), all dark read values D output from the radiation detection elements (x−2, y−2) to (x + 2, y + 2) in the one dark read. An average value W (x, y) of (x−2, y−2) to D (x + 2, y + 2) is calculated and adopted as a temperature compensation variable. In this case, the average value (temperature compensation variable) W (x, y) is a spatial average value for the pixel position (x, y) of the dark read value D (x, y). N in the formula (3) is 25 (= 5 × 5) in this case.

  Here, it is considered whether or not the spatial average value W (x, y) as the temperature compensation variable satisfies the requirement 4).

となる（図８参照）。 Since each of the dark reading values D (x−2, y−2) to D (x + 2, y + 2) is an independent variable, and its fluctuation is random in terms of time and space, the dark reading value D The temporal average value of temporal fluctuations of (x, y) is expressed as μ _D (x, y), the standard deviation of the temporal fluctuation distribution is expressed as σ _D (x, y), and the temperature compensation variable W (x, y) ) Is represented by μ _W (x, y), and the standard deviation of the temporal fluctuation distribution is represented by σ _W (x, y).

(See FIG. 8).

Here, if σ _D (x, y) does not include a defective pixel having a very large value (abnormal value) or many abnormal pixels,
σ _W (x, y) <σ _D (x, y)
The following relational expression holds.

となり、σ_Ｗ（ｘ，ｙ）は、σ_Ｄ（ｘ，ｙ）の値の１／√Ｎとなる。Ｎ＝２５の場合は、１／√Ｎ＝１／５であるから、温度補償変数Ｗ（ｘ，ｙ）は、時間的ゆらぎの分布の広がり（標準偏差）が、ダーク読取値Ｄ（ｘ，ｙ）にくらべて１／５の分布となる。すなわち、図８に示すように、広がりが狭い（標準偏差が小さい）分布となり、要件４）を満足する。 For example, assuming that each value of σ _D (x−2, y−2) to σ _D (x + 2, y + 2) is different, but taking an average, it is almost equal to σ _D (x, y). )

Σ _W (x, y) is 1 / √N of the value of σ _D (x, y). In the case of N = 25, 1 / √N = 1/5. Therefore, the temperature compensation variable W (x, y) has a dark reading D (x, Compared to y), the distribution is 1/5. That is, as shown in FIG. 8, the distribution is narrow (standard deviation is small) and satisfies requirement 4).

となり、
σ_Ｗ（ｘ，ｙ）＜σ_Ｄ（ｘ，ｙ）
が成立する。 As an extreme example, among the pixel positions (x−2, y−2) to (x + 2, y + 2), the dark reading value temporally at the pixel position of (N−1) other than the target pixel (x, y). It is assumed that the standard deviation value of fluctuation has a magnitude N times the standard deviation value σ _D (x, y) of temporal fluctuation of the dark read value D (x, y) at the target pixel position. If N> 2,

And
σ _W (x, y) <σ _D (x, y)
Is established.

If N = 25,
SQRT [{1+ (N−1) · N} / N ² ]
= SQRT {(1 + 24 × 25) / (25 × 25)}
≒ 0.981
So,
σ _W (x, y) ≈0.981 · σ _D (x, y) <σ _D (x, y)
Normally, each value of σ _D (x−2, y−2) to σ _D (x + 2, y + 2) is different, but taking an average takes a value substantially equal to σ _D (x, y). It can be said that the method satisfies requirement 4).

The distribution of the spatial average value W (x, y) as the temperature compensation variable is a plurality of radiation detection elements 14 (hereinafter referred to as predetermined regions) in a square region centered on the one radiation detection element (x, y). The greater the number of radiation detection elements 14 _{R in} the sense of the radiation detection elements 14 belonging to R, that is, the larger the value of N, the spatial average value W (x, y) as a temperature compensation variable. ) Of the time fluctuation standard deviation σ _W (x, y) becomes small. This means that when σ _W (x, y) takes a sufficiently small value,
W (x, y) ≈μ _W (x, y) (6)
It can be approximated.

That is, dark read value D (x-2, y- 2) output from the immediately preceding or one dark read the radiation detecting element 14 _R, which is performed immediately after the radiation image capturing ~D (x + 2, y + 2) ( hereinafter, in the sense that dark read value output from the radiation detecting element 14 _R belonging to a predetermined region R is denoted as dark read value D _R.), the time distribution of the temporal fluctuations as illustrated in FIG. 7 The average value of the target fluctuations μ _D (x−2, y−2) to μ _D (x + 2, y + 2) fluctuates around, but the spatial average value (temperature compensation variable) W ( As x, y), a value approximately equal to the temporal average value μ _W (x, y) of the distribution of the spatial average value (temperature compensation variable) W (x, y) is calculated.

Thus, the one of the radiation detecting element (x, y) be the average value mu _D and the standard deviation sigma _D of the distribution of the temporal fluctuations of the dark read value D _R to be output from the radiation detection elements 14 _R are different, including the (see FIG. 7), by calculating their spatial mean value (temperature compensation variable) W (x, y), the temporal fluctuation of the dark read value D _R output from the radiation detection elements 14 _R is valid The spatial variation (temperature compensation variable) W (x, y) is almost equal to the average value μ _W (x, y) of the distribution of temporal fluctuations. Can do.

  In other words, the average value of temporal fluctuations that cannot be obtained unless a plurality of dark readings are performed is the spatial average value W (x, y) of the dark reading values D (x, y) obtained by one dark reading. y) can be substituted with a small error.

  Here, the radiation detection elements (x ′, y ′) other than the one radiation detection element (x, y) when calculating the temperature compensation variable W (x, y) for the one radiation detection element (x, y). ) Will be described.

In the radiation detection element (x, y), the radiation detection element (x, y) is usually output when the temperature of the radiation detection element (x, y) or the reading circuit for reading out the signal value is low. The dark read value D (x, y) is a small value, and as the temperature T of the radiation detection element (x, y) increases, the output dark read value D (x, y) also increases. It is known. Therefore, for example, the average value μ _D (x, y) of the temporal fluctuation distribution of the dark read value D (x, y) output from each radiation detection element (x, y) as shown in FIG. As the temperature rises, it shifts to the right on the graph in the figure.

  Further, for example, when the power of the radiation image detector 1 is turned on, the temperature of the sensor panel unit 4 gradually increases. However, each radiation detection element (x, y) on the sensor panel unit 4 and elements of the readout circuit The temperature variation is not uniform. For example, the radiation detection element (x, y) at the center of the sensor panel unit 4 in which a large number of radiation detection elements (x ′, y ′) are present around the sensor panel unit 4 The radiation detection element (x, y) in the peripheral portion has a different temperature variation method, and the temperature variation method differs if the readout IC constituting the readout circuit is different. Even within the same readout IC, if the amplifier circuit 17 is different, the manner of temperature variation is different.

  In such a situation, other than the one radiation detection element (x, y) when calculating the spatial average value (temperature compensation variable) W (x, y) for one radiation detection element (x, y). As the radiation detection element (x ′, y ′), for example, when a radiation detection element (x ′, y ′) far from the one radiation detection element (x, y) on the sensor panel unit 4 is selected, its temperature It is predicted that the fluctuation is different from the temperature fluctuation of the one radiation detection element (x, y).

  Therefore, for example, as shown in FIG. 7, the dark reading value D (x ′, y ′) output from each radiation detection element (x ′, y ′) in the left-right direction depends on the temperature of the distribution of temporal fluctuations. The degree of shift is different from the degree of shift of the temporal fluctuation distribution of the dark reading value D (x, y) of the one radiation detection element (x, y), and the relative positions on the graph are respectively It fluctuates depending on the temperature of the pixel position.

In this way, the distribution of temporal fluctuations of the dark reading value D (x, y) of the radiation detection element (x, y) and the distribution of temporal fluctuations of the other radiation detection elements (x ′, y ′) depending on the temperature. If the position on the graph is relatively shifted depending on the temperature of each pixel position, as shown in FIG. 9, the spatial average value (temperature compensation variable) W (x, y) The distribution becomes broad and a distribution with a small standard deviation σ _W (x, y) as shown in FIG. 8 cannot be obtained. Therefore, the spatial average value (temperature compensation variable) W (x, y) calculated according to the above equation (3) is the temporal average value μ _{W of the} spatial average value (temperature compensation variable) W (x, y). The beneficial effect of being approximately equal to (x, y) cannot be used.

  Therefore, in this method, the spatial average value (temperature compensation variable) W (x, y) for one radiation detection element (x, y) from the result of one dark reading performed immediately before or after radiographic imaging. As a radiation detection element (x ′, y ′) other than the one radiation detection element (x, y) for calculating the radiation detection, a radiation detection that changes in temperature in the same manner as the one radiation detection element (x, y). The element (x ′, y ′) is selected in advance and is associated with the one radiation detection element (x, y) in advance.

At this time, it is not necessary to select all the radiation detection elements (x, y) on the sensor panel unit 4 in the same manner, and the selection method may be different for each radiation detection element. As a selection method, for example, as described above, a plurality of radiation detection elements (x, y) _R existing in an n × n square region centered on the one radiation detection element (x, y) are selected. Although it can be used, as shown in FIGS. 16 (A) to 16 (C) and FIGS. 19 (C) and 19 (D) which will be described later, the square region does not necessarily have the one radiation detection element (x, y). It does not have to be set to exist in the center.

  In addition, in this embodiment, the size of the square area is 5 × 5 or 7 × 7 (refer to the description in FIGS. 24 to 26 to be described later for an example of 7 × 7), but is limited to this size. Not what you want. As long as the size is larger than 1 × 1, it may be 2 × 2, 3 × 3, or 9 × 9 or larger. In addition, a rectangular area can be set instead of the square area, or an indefinite area can be used.

Further, the radiation detection elements (x ′, y ′) other than the one radiation detection element (x, y) are radiation detection elements that change in temperature in the same manner as the one radiation detection element (x, y). Well, it is scattered on the sensor panel unit 4 instead of the radiation detection elements (x, y) _R existing in the spatially continuous region including the one radiation detection element (x, y) as described above. It may be a radiation detection element (x, y). The detection element group exhibiting the same temperature characteristics can be grasped in advance during factory inspection or the like. It can also be reset after the establishment of the facility.

  Next, the relationship between the spatial average value W (x, y) used as the temperature compensation variable and the true value of the offset correction value O (x, y) of the one radiation detection element (x, y). explain.

  In the dark reading performed once just before or immediately after radiographic imaging, as shown in FIG. 10, one dark reading value including temporal fluctuation is received from one radiation detection element (x, y). Only D (x, y) is obtained, and the true value of the offset correction value O (x, y) is not known. Further, each dark read value D () output from each of the radiation detection elements (x−2, y−2) to (x + 2, y + 2) associated with the one radiation detection element (x, y). x−2, y−2) to D (x + 2, y + 2) also include temporal fluctuations, and based on them, the offset correction value O (1) of the one radiation detection element (x, y) is directly selected. The true value of x, y) cannot be obtained.

  Therefore, in this method, as will be described in detail below, the dark detection is performed a plurality of times at the time of past calibration (performed before radiography of the current subject), and the one radiation detection element (x, y) A dark reading value d (x, y) output from itself is obtained, and first, a plurality of dark reading values d (x, y) output from one radiation detection element (x, y) are obtained. A temporal average value δ (x, y) is calculated.

  Further, as a temperature compensation variable, from each of the plurality of radiation detection elements (x, y) in the square region centered on the one radiation detection element (x, y) in each dark reading at the time of past calibration. Spatial average value of dark reading value d to be output (hereinafter, spatial average value W (x, y) as a temperature compensation variable obtained by one dark reading performed immediately before or after the above radiographic imaging) In order to distinguish them from each other, a spatial average value as a temperature compensation variable obtained at the time of calibration is expressed as w (x, y)), and a spatial average value w (x, The time average value ω (x, y) for a plurality of times of y) is calculated.

  The temporal average value δ (x, y) of the dark read value d (x, y) of the one radiation detection element (x, y) itself and the spatial average value w ( (x, y) temporal average value ω (x, y) and spatial average value W (as a temperature compensation variable) obtained by one dark reading performed immediately before or after the current radiographic imaging. x, y) is used to calculate the offset correction value O (x, y) of the one radiation detection element (x, y).

  In addition, in the case of the calibration in the present embodiment, not only the calibration for updating both the gain correction value and the offset correction value performed periodically (for example, every month) as described above, but also compared with the gain correction value, Only offset correction values with relatively large temperature fluctuations are targeted, and offset calibration executed as necessary is also included.

  In other words, in the present embodiment, calibration refers to the dark reading value D (x) as necessary when the radiation image detector 1 is shipped from the factory, introduced into the facility, or during standby when radiation is not exposed. , Y) is an operation for acquiring and creating all or part of the data for correcting the actual image data F (x, y) including the above, and is based on the calibration performed periodically as described above. Is also a broad concept.

  Therefore, the calibration after introduction into the facility may also be performed with a shorter cycle (for example, every day) than the periodic calibration. In one calibration and offset calibration, dark reading is usually performed a plurality of times.

  It should be noted that if the temporal average value δ (x, y), the spatial average value w (x, y) as the temperature compensation variable, and the temporal average value ω (x, y) are calculated using too old data, Therefore, the dark reading value d (x, y) from which these values are calculated is the value obtained by the dark reading from the present to the predetermined number of times or the latest (that is, It is preferable to use a value obtained by a plurality of dark readings performed in the previous calibration or offset calibration.

  Further, the process of calculating the temporal average value δ (x, y) or the like from the dark read value d (x, y) may be performed by the radiation image detector 1 or from the radiation image detector 1 to the console 31 ( It is also possible to transmit the dark reading value d (x, y) to an external device such as FIG.

Now, the temporal average value δ (x, y) of each dark read value d (x, y) of the one radiation detection element (x, y) itself obtained by the above-mentioned multiple times of dark reading is as follows. Is calculated according to the following equation (7). In the following equation (7), M represents the number of dark read values d _m (x, y) used for calculating the temporal average value δ (x, y) or the like, that is, the predetermined number of times.

As shown in FIG. 11, the dark read value d _m (x, y) is output from the one radiation detection element (x, y) in the m-th (m = 1 to M) dark read, and these values are output. When collectively expressed in a histogram, as shown in FIG. 12, the distribution is in a normal distribution with the temporal average value δ (x, y) as the center and the standard deviation σ _d (x, y). This distribution represents the temporal fluctuation of the dark reading value d _m (x, y).

Further, in the m-th (m = 1 to M) dark reading at the time of past calibration, a plurality of radiation detecting elements (x, x, y) in the square region with the one radiation detecting element (x, y) as the center. y) The spatial average value w _m (x, y) of each dark read value d _m (x, y) output from _R is calculated according to the following equation (8) similar to the above equation (3). In this case, N in the equation (8) is also 25 (= 5 × 5).

Then, the temporal average value ω (x, y) for M times of the spatial average value w _m (x, y) of each time is calculated according to the following equation (9).

As shown in FIG. 11, spatial average values w _m (x, y) are calculated in the m-th (m = 1 to M) dark reading, and these values are collectively shown in a histogram as shown in FIG. 12. As shown, the distribution is in a normal distribution with a standard deviation σ _w (x, y) centered on the temporal average value ω (x, y).

となることから分かるように、過去のキャリブレーション時のＭ回のダーク読取での各放射線検出素子（ｘ，ｙ）についての時間的平均値δ（ｘ，ｙ）が算出されれば、各回ごとにそれぞれ空間的平均値ｗ_ｍ（ｘ，ｙ）を算出しなくても、各放射線検出素子（ｘ，ｙ）ごとの時間的平均値δ（ｘ，ｙ）を用いてダーク読取値ｄ_ｍ（ｘ，ｙ）の空間的平均値ｗ_ｍ（ｘ，ｙ）のＭ回分の時間的平均値ω（ｘ，ｙ）を算出することができる。 In addition, by substituting the equations (8) and (7) into the above equation (9),

As can be seen from the above, if the temporal average value δ (x, y) for each radiation detection element (x, y) in M dark readings in the past calibration is calculated, each time Even if the spatial average value w _m (x, y) is not calculated for each, the dark reading value d _m (using the temporal average value δ (x, y) for each radiation detection element (x, y) is used. A temporal average value ω (x, y) of M times of the spatial average value w _m (x, y) of x, y) can be calculated.

  Here, the temporal average value δ (x, y) of the dark read value d (x, y) output from the one radiation detection element (x, y) in a plurality of dark readings at the time of past calibration. And the temporal average value ω (x, y) of the spatial average value w (x, y) as the temperature compensation variable and one dark reading performed immediately before or after the current radiographic image capturing. The relationship between the spatial average value W (x, y) as a temperature compensation variable and the true value of the offset correction value O (x, y) of the one radiation detection element (x, y) will be considered.

In general, calibration is often performed during start-up preparation or patient preparation without performing patient imaging, and dark reading at the time of calibration is usually performed under conditions of the same preset temperature T ₀ . It can be assumed that it is done below. Therefore, the temporal average value δ (x, y) and the spatial average value w (x, y) as the temperature compensation variable regarding the one radiation detection element (x, y) obtained by dark reading at the time of past calibration. ) Is a value at the temperature T ₀ .

On the other hand, the temperature T in the vicinity of the radiation detection element (x, y) or the radiation detection element (x, y) on the sensor panel unit 4 when radiographic imaging is performed is not necessarily calibrated as shown in FIG. It is not the temperature T ₀ during the operation.

  In particular, in the radiographic image detector 1, even when the power is on, when the radiographic image detector 1 is not used for radiographic imaging, the power consumption state is set to the sleep mode automatically or manually in order to reduce power consumption. As shown in FIG. 13, the temperature T of the radiation detection element (x, y) and the like rises in the imageable mode, and the sensor panel unit 4 in the sleep mode (see the portion S in the figure). As a result, the temperature T of the radiation detection element (x, y) or the like decreases.

  Further, as described above, in the radiation detection element (x, y), the dark read value D (x, y) that is normally output when the temperature T of the radiation detection element (x, y) or the like is low is small. As the temperature T of the radiation detection element (x, y), etc. increases, the dark read value D (x, y) output also increases.

Therefore, as shown in FIG. 14, a dark read value D (x, y) output from one radiation detection element (x, y) in one dark read performed immediately before or after the current radiographic imaging. ) Estimated from the center value (average value) of the temporal fluctuation distribution and the dark reading output from the radiation detection element (x, y) _R associated with the one radiation detection element (x, y). distribution of temporal fluctuation estimated for spatial average value W as a temperature compensation variable is the mean value of the values D _R (x, y) also increases or decreases the temperature T.

  That is, the distribution of the dark reading value D (x, y) in FIG. 14 and the distribution of the spatial average value W (x, y) as the temperature compensation variable are varied on the horizontal axis in FIG. Fluctuates left and right. In general, when the temperature T rises, the distribution tends to shift to the right on the horizontal axis, and when the temperature T falls, the distribution tends to shift to the left on the horizontal axis. The distribution of fluctuations such as the dark read value D (x, y) output from one radiation detection element (x, y) in one dark read performed immediately before or immediately after the current radiographic image capture is as follows. Since it is only estimated, it is displayed with a broken line in FIG.

However, as described above, even in that case, the temperature of the radiation detection element (x ′, y ′) _R including the one radiation detection element (x, y) similarly varies. Therefore, the position on the graph of FIG. 14 of the fluctuation distribution estimated for the dark read value D (x, y) output from the one radiation detection element (x, y) and the one radiation detection element ( x, it is estimated for radiation detection element associated with y) (x, y) spatial average value W as a temperature compensation variable is the mean value of the dark read value D _R to be output from the _R (x, y) The relative positional relationship of the temporal fluctuation distribution with respect to the position on the graph (that is, the distance between the average values of the respective distributions) should not change even if the temperature fluctuates.

By the way, if the current radiographic imaging is performed in an environment of temperature T ₀ , the average value μ _D (of the temporal fluctuation distribution estimated for the dark read value D (x, y) shown in FIG. x, y) and the average value μ _W (x, y) of the temporal fluctuation distribution estimated for the spatial average value W (x, y) as the temperature compensation variable are the past calibrations shown in FIG. Average value δ (x, y) of temporal fluctuation distribution of a plurality of dark reading values d _m (x, y) obtained by a plurality of dark readings at the time of the measurement, or the one radiation detection element (x, The average value ω (x, y) of the distribution of temporal fluctuations of the spatial average value w _m (x, y) of each dark read value d _m (x, y) output from the radiation detection element 14 _R including y) ).

Further, the above dark read values D (x, y) of the distribution of the temporal fluctuation of the average value μ _{D (x,} y) and the spatial average value W (x, y) as a temperature compensation variable temporal fluctuations Since the relative positional relationship with the average value μ _W (x, y) of the distribution does not change even when the temperature T changes as described above, the difference between the average values of the distributions, that is, the dark read value D (x , the average value of the distribution of the temporal fluctuation of _{y) μ D (x, y} ) and the spatial average value W as a temperature compensation variable (x, mean value of the distribution of the temporal fluctuation of y) mu _{W (x,} y The difference from) does not change even if the temperature T changes.

Therefore, the temporal average of the distribution of the dark reading values D (x, y) of the one radiation detection element (x, y) obtained by one dark reading performed immediately before or immediately after the current radiographic image capturing. The difference between the value μ _D (x, y) and the temporal average value μ _W (x, y) of the distribution of the spatial average value W (x, y) as the temperature compensation variable is expressed as ε (x, y) When placed (see Figure 14),
ε (x, y) = μ _D (x, y) −μ _W (x, y) (11)
Is the temporal average value δ (x) of the distribution of a plurality of dark reading values d _m (x, y) obtained by a plurality of dark readings at the time of past calibration shown in FIG. , Y) and the spatial average value w _m (x, y) of each dark read value d _m (x, y) output from the radiation detection element 14 _R including the one radiation detection element (x, y). The difference from the temporal average value ω (x, y) of the distribution of
δ (x, y) −ω (x, y)
Is equal to

That is,
ε (x, y) = μ _D (x, y) −μ _W (x, y)
= Δ (x, y) −ω (x, y) (12)
Is established.

Equation (12) is
μ _D (x, y) = ε (x, y) + μ _W (x, y) (13)
If the variable that can be calculated in advance at the time of calibration, that is, the value of ε (x, y) = δ (x, y) −ω (x, y) is known, then μ _W (x, y ) Indicates that μ _D (x, y) can be estimated.

Here, as described above, from the equation (6), W (x, y) ≈μ _W (x, y) (6)
Substituting this into equation (13) gives
μ _D (x, y) ≈ε (x, y) + W (x, y) (14)
The relationship is obtained.

If a variable that can be calculated in advance during calibration, that is, a value of ε (x, y) = δ (x, y) −ω (x, y) is known, the value is acquired later when radiographic images are taken. This means that μ _D (x, y) can be estimated if only the temperature compensation variable W (x, y) calculated from one dark image can be calculated.

Similarly to the above-described conventional case, the average value μ _D (x, y) of the temporal fluctuation distribution estimated for the dark read value _D (x, y) is exactly the radiation detection element (x, y). ) Offset correction value O (x, y) can be regarded as a true value.
O (x, y) ≈ε (x, y) + W (x, y) (15)
Can be rewritten.

  If a variable that can be calculated in advance during calibration, that is, a value of ε (x, y) = δ (x, y) −ω (x, y) is known, the value is acquired later when radiographic images are taken. This means that if only the temperature compensation variable W (x, y) calculated from one dark image can be calculated, O (x, y), that is, the true value of the offset correction value can be estimated. In other words, if a variable that can be calculated in advance at the time of calibration, that is, a value of ε (x, y) = δ (x, y) −ω (x, y) is known, If the temperature compensation variable W (x, y) for each pixel position in a single dark image is calculated no matter how the temperature of each element such as an electric circuit thereafter changes, radiation at the time of radiographic imaging This indicates that an ideal offset correction value O (x, y) in the temperature state of each element such as a detection element or an electric circuit is obtained.

Therefore, in this method,
O (x, y) = ε (x, y) + W (x, y) (16)
(Where ε (x, y) = δ (x, y) −ω (x, y))
Accordingly, the offset correction value O (x, y) of the radiation detection element (x, y) is calculated.

  As a result, a technique that becomes a framework for solving the above-described problems has been proposed.

  In the description so far, the temperature of each element in the radiation image detector 1 at the time of calibration has been assumed to be constant. However, when calibration is performed, the number of dark images acquired (the above-described M When the value of () increases, the temperature of each element in the radiation image detector 1 also varies during calibration.

As is clear from the above formulas (7) to (10), the larger the value of M, the more the temporal fluctuations of d _m (x, y) and w _m (x, y) are offset, and (16 The accuracy of ω (x, y) and δ (x, y), which influences the accuracy of the expression), is improved. The larger the value of M, the greater the value of ω (x, y) and δ (x, y). Means that it is more susceptible to temperature fluctuations.

  Therefore, the idea of temperature correction described so far is also applied to this problem.

Substituting (7) and (9) into (12),

here,
_{λ m (x, y) =} d m (x, y) -w m (x, y) ... (18)
After all,

Here, λ _m (x, y) is a temperature-corrected dark in which the temperature change of the dark read value d _m (x, y) at the time of calibration is corrected by the temperature compensation variable w _m (x, y). The reading value λ _m (x, y) can be considered.

If the temperature of each element in the radiation image detector 1 fluctuates during calibration, the distribution of d _m (x, y) becomes a very broad distribution as shown in FIG. 15. Therefore, the above (7) The temporal average value δ (x, y) of d _m (x, y) obtained by the equation cannot indicate a correct value.

Therefore, in general, when dark reading is performed a plurality of times during calibration and an average value thereof is obtained (in the past, the offset correction value is directly obtained by this method as described above), direct d _m (x, Instead of calculating the average value of y), as described above, ε (x, y) that is the average value (temporal average value) of the temperature-corrected dark read value λ _m (x, y) is obtained. In this case, it is possible to obtain an accurate dark reading temporal average value in which the temperature fluctuation is corrected. That is, ε (x, y) can be called a temporal average value of temperature-corrected dark reading.

The temperature-corrected dark reading value λ _m (x, y) and the temperature-corrected dark reading temporal average value ε (x, y) are used for dark reading at the time of calibration and also at the time of radiographic imaging. It can be applied to all operations that are performed multiple times and take the average (temporal average). Moreover, even if the value of M, which is the number of time averages, is large, it can be made less susceptible to temperature fluctuations.

  Similar temperature compensation can also be applied when obtaining the gain correction value G (x, y) during calibration. When obtaining the gain correction value G (x, y), radiation image data in a state where the subject is not placed is acquired with a predetermined radiation dose determined in advance, and offset correction is performed on this to obtain the gain correction value. In order to calculate G (x, y), if the radiographic image data acquired at this time is regarded as the actual image data F (x, y) of the present method, the gain correction subjected to the temperature correction similar to the present method is performed. Data G (x, y) can be obtained. In addition, it is preferable to similarly perform the above-described temperature correction on the offset correction value used when obtaining the gain correction value G (x, y). Thereby, a highly accurate gain correction value G (x, y) can be obtained, and as a result, good image data with a high SN ratio can be obtained.

In this method, in the distribution (see FIG. 12) of each dark read value d _m (x, y) output from the radiation detection element (x, y) in M dark readings during calibration, a temporal average is obtained. The value δ (x, y) is calculated, and further, the standard deviation σd (x, y) of the distribution is calculated,
d _m (x, y) −δ (x, y) (20)
d _m (x, y) / σd (x, y) (21)
Or
(D _m (x, y) −δ (x, y)) / σd (x, y) (22)
It is also possible to normalize each dark read value d _m (x, y) by performing the above-described calculation and configure the above processing based on each normalized dark read value.

[Example of method for selecting other radiation detection elements to be associated with radiation detection elements in advance]
As described above, in order to calculate the spatial average value W (x, y) as a temperature compensation variable for one radiation detection element (x, y), the radiation detection element (x, y) is associated in advance. The other radiation detection elements (x ′, y ′) may be any element that varies in temperature in the same manner as the one radiation detection element (x, y), and the method for selecting them is as described above. There can be various methods. Here, an example of a selection method according to the practical configuration of the radiation image detector 1 will be described.

  For example, in FIG. 6, a sufficient number of other radiation detection elements (x ′, x) in the row and column directions around one radiation detection element (x, y) on the sensor panel unit 4 of the radiation image detector 1. If there is y ′), the other radiation detection elements (x−2, y−2) to (x + 2) existing in, for example, a 5 × 5 square region centering on the one radiation detection element (x, y). , Y + 2) is shown. However, when the one radiation detection element (x, y) is in the peripheral portion of the sensor panel unit 4, the one radiation detection element (x, y) is centered in the square area of 5 × 5, for example. ) May not be located.

  Therefore, in such a case, for example, as shown in FIGS. 16A to 16C, the position of the one radiation detection element (x, y) represented by being darkly colored in the drawing is selected. Just do it.

In this case, in the calculation of the spatial average value w _m (x, y) in the dark reading at the time of calibration and the calculation of the temporal average value ω (x, y), the radiation detection element (x, The region R for y) is set in the same manner. Further, in FIGS. 16A to 16C, the numbers described above and to the left of each radiation detection element represented in a lattice pattern are the row direction of each radiation detection element (x, y) and It is a number representing the position in the column direction, and the left coordinate (x, y) represents the coordinate of the radiation detection element (x, y).

  On the other hand, as shown in FIG. 17, the readout circuit 9 (see FIG. 2) for reading out and amplifying an electrical signal from each radiation detection element (x, y) normally has 128 pixels or 256 pixels of radiation in the row direction. One readout IC 91, 92,... Is connected to each detection element (x, y), and the necessary number of them are arranged in parallel. In addition, since the temperature characteristics and noise characteristics of the readout ICs 91, 92,... On the readout circuit 9 side are not necessarily the same, the offset correction value O (x, y) of each radiation detection element (x, y) is the readout IC 91. , 92,...

  Therefore, as shown in FIG. 18, a read IC that reads an electrical signal from each radiation detection element belonging to the region R set to calculate the offset correction value O (x, y) of the radiation detection element (x, y). Are read out separately by the adjacent readout ICs 91 and 92, the values of the temperature compensation variables W (x, y) and ω (x, y) are set for each radiation detection element (x, y) belonging to the region R. There may be an error, and the offset correction value O (x, y) may not be calculated satisfactorily.

  In such a case, for example, as shown in FIGS. 19C and 19D, the radiation detection element (x, y) connected to the readout IC 91 and the radiation detection element connected to the readout IC 92, for example. Consider (x, y) separately. Then, the radiation detection element (x, y) existing in the vicinity of the boundary L between the readout ICs 91 and 92 is handled in the same manner as in the case of the peripheral portion of the sensor panel unit 4 so that one radiation detection element (x , Y), a plurality of radiation detection elements associated in advance can be selected from the radiation detection elements connected to the same readout IC 91.

  In FIGS. 19A and 19B, a plurality of radiation detection elements associated in advance with one radiation detection element (x, y) may be selected in the normal manner shown in FIG. It is shown.

[Features of defective pixel determination method and abnormal pixel determination method]
Hereinafter, features of the defective pixel determination method and the abnormal pixel determination method according to the present embodiment will be described.

  First, defective pixels are detected by carrying out post-manufacturing shipping inspections and inspections involving radiation irradiation during periodic calibration, and the pixel positions of the detected defective pixels are registered in the defective pixel map, As described above, it is operated and managed in association with the radiation image detector.

  As an example of a defective pixel, a pixel that constantly outputs a large pixel value (for example, a pixel whose signal value always reaches a saturation level), or a pixel that constantly outputs only a small pixel value (for example, a signal value that always remains) Can be easily determined for pixels that output only zero level values, and pixels that always output only constant signal values (pixels whose output pixel values do not change even when irradiated with radiation). Therefore, it is assumed that such a defective pixel has already been determined and registered in the defective pixel map.

  What is going to be solved here is an abnormal pixel that outputs an abnormal signal value at a certain probability, and whether or not these abnormal pixels are detected accurately without irradiation and whether they are defective pixels or not. Or a method of registering abnormal pixels in the defective pixel map, a method of appropriately processing abnormal pixels as abnormal pixels without registering them in the defective pixel map, or in image data irradiated with radiation This is a method for dealing with the abnormal pixels.

  For this reason, the main method uses dark reading values (reading signals that do not require radiation irradiation) acquired during calibration or before and after radiographic imaging. Also, a method for determining defective pixels in consideration of temperature fluctuations of each element in the radiation image detector when determining defective pixels and abnormal pixels will be described. In addition, a method for detecting abnormal pixels from image data obtained by irradiating radiation and appropriately handling them will be described.

First, defective pixels and abnormal pixels that are desired to be solved by the present invention will be described. As shown in the equation (1), in order to obtain the final image data F _O (x, y), the offset correction value O (x, y) is subtracted from the photographed image data F (x, y), The difference is multiplied by a gain correction value G (x, y).
F _O (x, y) = (F (x, y) −O (x, y)) × G (x, y) (1)

Therefore, even if the actual image data F (x, y) has a normal pixel value, if the offset correction value O (x, y) or the gain correction value G (x, y) is abnormal, the calculation is performed. Since the final image data F _O (x, y) as a result shows an abnormal value, this is an abnormal pixel (a pixel that should be corrected originally).

Further, even if the offset correction value O (x, y) and the gain correction value G (x, y) are normal, if the value of the actual image data F (x, y) is abnormal, the calculation result is obtained. Since the final image data F _O (x, y) shows an abnormal value, this is also an abnormal pixel (a pixel that should be corrected originally).

  The pixel position (x, y) that outputs an abnormal pixel value with an offset correction value O (x, y), gain correction value G (x, y), and actual image data F (x, y) with a certain probability Abnormal pixels that are classified according to the intensity of the value (guideline indicating how abnormal) and the frequency of output of abnormal values are determined to be unacceptable or should not be permitted are registered in the defective pixel map as defective pixels. Do.

  Also, even if a pixel suddenly outputs an abnormal value, if it is determined that the intensity of the abnormal value (a guideline indicating how abnormal) or the frequency of outputting the abnormal value is an acceptable level, the abnormal pixel The same processing (correction processing, etc.) as that of the defective pixel is performed, but it is allowed not to register the defective pixel in the defective pixel map. Of course, all abnormal pixels may be registered as defective pixels in the defective pixel map.

  Here, the gain correction value G (x, y) is calculated by irradiating with radiation at the time of calibration, but the offset correction value O (x, y) can be calculated without irradiating with radiation. It is a parameter.

The inventors pay attention to the dark read value D (x, y) or dark read value d _m (x, y) used for calculation of the offset correction value, and detect abnormal pixels generated at a certain probability as dark read values. A method of finding and determining from D (x, y) or dark read value d _m (x, y) and processing as a defective pixel or an abnormal pixel (in some cases, registering an abnormal pixel as a defective pixel in a defective pixel map) I found it. The present inventors also found a method for detecting abnormal pixels in radiographic image data and appropriately processing them.

[Defective Pixel Determination Method and Defective Pixel Determination Program]
Hereinafter, an embodiment of a defective pixel determination method and a defective pixel determination program according to the present embodiment will be described with reference to the drawings. However, the present invention is not limited to the following illustrated examples.

  As shown in FIG. 5 and the like described above, even when the environment is at the same temperature T, the dark read value D (x, y) output from one radiation detection element (x, y) of the radiation image detector 1. A fluctuation (variation) occurs in y), and the fluctuation distribution is a normal distribution having a standard deviation σ (x, y) around the mean value μ (x, y).

  However, depending on the radiation detection element (x, y), the average value μ (x, y) of the distribution is a dark read value D (x ′, y ′) output from another radiation detection element (x ′, y ′). ) Distribution is far from the average value μ (x ′, y ′) or the fluctuation is very large, that is, the standard deviation σ (x, y) of the distribution is different from that of the other radiation detection elements (x ′, y). In some cases, it is much larger than the standard deviation σ (x ′, y ′) of the distribution of the dark reading value D (x ′, y ′) output from “′).

Therefore, in such a radiation detection element (x, y), the stability of the finally obtained image data F _O (x, y) is also deteriorated and becomes an abnormal value that can be regarded as a defective pixel. May end up. Hereinafter, the radiation detection element (x, y) that can be regarded as a defective pixel in this way is represented as a defective pixel (xs, ys).

  Further, although the standard deviation σ (x, y) of the dark read value D (x, y) falls within the normal range, there are pixels that occasionally output an abnormal value. Such pixels cannot be detected even if attention is paid only to the standard deviation σ (x, y) of the dark read value D (x, y). As described above, although the standard deviation σ (x, y) of the dark read value D (x, y) falls within the normal range, the pixel that occasionally outputs an abnormal value is also regarded as an abnormal pixel if it exceeds the allowable range. .

  Therefore, in the defective pixel determination method and the defective pixel determination program according to the present embodiment, when the radiation detection element (x, y) becomes an abnormal pixel with a certain probability, the defective pixel (xs, ys) is determined from the occurrence probability and the degree of abnormality. ) To determine whether or not to register. In particular, in the defective pixel determination program according to the present invention, the control means 6 that is a computer of the radiation image detector 1 described above and a defective pixel determination device that is a computer of a defective pixel determination system 30 described later (that is, a console 31 described later) The server means 39) realizes this defective pixel determination function, that is, the defective pixel determination function described in detail below. Needless to say, it may be performed in the server 31 (or the server means 39).

As described above, in the present embodiment, a plurality of times during the past calibration (M times) the radiation detecting element in made a dark read (x, y) dark read values output from its d _{m (x,} y) A plurality of times (M times) of data are accumulated for each radiation detection element (x, y). And their dark read values _d m (x, y) expressed together in a histogram, as shown in FIG. 12, each radiation detection element (x, y) dark read value for each _d m (x, y ) Temporal statistical values such as the temporal average value δ (x, y) and standard deviation σd (x, y) of the temporal fluctuation distribution.

Therefore, in the present embodiment, the dark read value d _m (x, x, y) output from one radiation detection element (x, y) obtained in the dark reading performed a plurality of times at the time of past calibration in this way. y) itself or a temporal statistical value such as a temporal average value δ (x, y) and a standard deviation σd (x, y) in the distribution of temporal fluctuations of the dark reading value d _m (x, y). Based on the determination methods described below, it is determined whether or not the one radiation detection element (x, y) is a defective pixel.

[Determination method 1]
First, attention is paid to the standard deviation σd (x, y) in the temporal fluctuation distribution of the dark read value d _m (x, y) output from one radiation detection element (x, y) as a temporal statistical value. When the standard deviation σd (x, y) is larger than a preset threshold σdth, the one radiation detection element (x, y) is determined as a defective pixel (xs, ys) and registered. Is possible.

Dark read value d _{m (x,} y) of the temporal fluctuation standard deviation .sigma.d (x, y) in the distribution of that large, dark read value d _{m (x,} y) output an abnormal value with some probability Therefore, if the pixel is larger than a preset threshold value σdth, it is determined as a defective pixel (xs, ys) and is registered in the defective pixel map. Can solve the problem.

That is, the standard deviation in the fluctuation distribution of the dark read value d _m (x, y) output from one radiation detection element (x, y) obtained in the dark reading performed a plurality of times during the past calibration. σd (x, y) is equal to or less than the threshold σdth, that is,
σd (x, y) ≦ σdth (23)
If so, the fluctuation (variation) in the dark read value d _m (x, y) of the one radiation detection element (x, y) is within an allowable range, and is determined to be a normal pixel. In this case, the one radiation detection element (x, y) is not registered as a defective pixel.

However, as shown in the normal distribution on the right side of FIG. 20, the standard deviation σd (x) in the temporal fluctuation distribution of the dark read value d _m (x, y) output from one radiation detection element (x, y). , Y) is an average standard deviation in the temporal fluctuation distribution of the dark read value d _m (x, y) output from the normal radiation detection element (x, y) shown in the normal distribution on the left side of FIG. When larger than σd (x, y) and exceeding the threshold σdth, that is,
σd (x, y)> σdth (24)
Is satisfied, the temporal fluctuation (variation) of the dark read value d _m (x, y) of the one radiation detection element (x, y) exceeds the allowable range and is determined to be a defective pixel. Is done. In this case, the one radiation detection element (x, y) is registered in the defective pixel map as a defective pixel (xs, ys).

The threshold value σdth is, for example, 3 of the average standard deviation σd (x, y) in the distribution of temporal fluctuations of the dark reading value d _m (x, y) output from the radiation detection element (x, y). The value is preferably set to a value of about 10 to 10 times, but is not limited to this value.

Note that the variance σd ² (x, y) is used instead of the standard deviation σd (x, y), a threshold is set for the variance σd ² (x, y), and processing is performed in the same manner as described above. Is also possible.

[Determination method 2]
In addition, as a temporal statistical value, a temporal average value δ (x, y) in a temporal fluctuation distribution of the dark read value d _m (x, y) output from one radiation detection element (x, y) is used. Paying attention, another threshold value in which the temporal average value δ (x, y) is larger than a preset threshold value δth1 (hereinafter referred to as the first threshold value δth1) or smaller than the threshold value δth1. If it is smaller than δth2 (hereinafter referred to as the second threshold δth2), it can be considered that the one radiation detection element (x, y) has a defect, so this is regarded as a defective pixel (xs , Ys) can be determined and registered.

If dark read value d _{m (x,} y) the temporal average value in the distribution of the temporal fluctuations of [delta] (x, y) is too large or too small, the dark read value d _{m (x,} y) in the future This also indicates that there is a high possibility that an abnormal value is output with a certain probability. Therefore, if the value is larger than the preset first threshold value δth1 or smaller than the second threshold value δth, the defective pixel (xs, ys ) And registering it in the defective pixel map can solve the problem recognized by the present invention as a problem.

That is, the time of the dark read value d _m (x, y) output from one radiation detection element (x, y) obtained in dark reading performed a plurality of times (M times) at the time of past calibration. Similar to the other radiation detection elements (x ^* , y ^* ), the temporal average value δ (x, y) in the fluctuation distribution is equal to or higher than the second threshold δth2 and lower than the first threshold δth1, that is,
δth2 ≦ δ (x, y) ≦ δth1 (25)
If so, the magnitude of the dark read value d _m (x, y) output from the one radiation detection element (x, y) is output from the other radiation detection element (x ^* , y ^* ). The magnitude of the dark read value d _m (x ^* , y ^* ) does not change so much and is within the allowable range and is determined to be a normal pixel. In this case, the one radiation detection element (x, y) is not registered as a defective pixel.

However, as shown in FIG. 21, the temporal average value δ (x, y) in the temporal fluctuation distribution of the dark read value d _m (x, y) output from one radiation detection element (x, y) is obtained. Is greater than the threshold δth1, that is,
δ (x, y)> δth1 (26)
Or the temporal average value δ (x, y) in the temporal fluctuation distribution of the dark read value d _m (x, y) output from one radiation detection element (x, y) is small, If it is smaller than the threshold δth2, that is,
δ (x, y) <δth2 (27)
Is satisfied, the magnitude of the dark read value d _m (x, y) output from the one radiation detection element (x, y) is output from the other radiation detection element (x ^* , y ^* ). The difference between the dark read value d _m (x ^* , y ^* ) and the read value is large and exceeds the allowable range, and is determined to be a defective pixel. In this case, the one radiation detection element (x, y) is registered in the defective pixel map as a defective pixel (xs, ys).

[Determination method 3]
Further, as the temporal statistics, one radiation detecting element (x, y) consideration of the magnitude itself of the value of the dark read values output from the d _{m (x,} y), dark read value d _{m (x,} when the value of y) is greater than the threshold value d _m th set in advance, since regarded as the the one of the radiation detecting element (x, y) has a defect from what defective pixels it It can be configured to determine and register as (xs, ys).

The value of the dark reading value d _m (x, y) itself being too large indicates that there is a high possibility that the dark reading value d _m (x, y) will output an abnormal value with a certain probability in the future. because there, defective pixel (xs, ys) is larger than the threshold value d _m th set in advance is determined as the, is possible to solve the problem of the present invention by registering the defective pixel map is recognized as problems it can.

That is, the value of the dark read value d _m (x, y) output from one radiation detection element (x, y) obtained in dark reading performed a plurality of times (M times) at the time of past calibration is obtained. , other radiation detecting elements ^(x *, ^{y *)} as well as the threshold _{d m} th or less, i.e.,
d _m (x, y) ≦ d _m th (28)
If so, the magnitude of the dark read value d _m (x, y) output from the one radiation detection element (x, y) is output from the other radiation detection element (x ^* , y ^* ). The magnitude of the dark read value d _m (x ^* , y ^* ) does not change so much and is within the allowable range and is determined to be a normal pixel. In this case, the one radiation detection element (x, y) is not registered as a defective pixel.

However, when the value of the dark read value d _m (x, y) output from one radiation detection element (x, y) is larger than the threshold value d _m th, that is,
d _m (x, y)> d _m th (29)
Is satisfied, the magnitude of the dark read value d _m (x, y) output from the one radiation detection element (x, y) is output from the other radiation detection element (x ^* , y ^* ). The difference between the dark read value d _m (x ^* , y ^* ) and the read value is large and exceeds the allowable range, and is determined to be a defective pixel. In this case, the one radiation detection element (x, y) is registered in the defective pixel map as a defective pixel (xs, ys).

In the above [determination method 1] to [determination method 3], the dark read value d output from the radiation detection element (x, y) itself in the dark reading performed a plurality of times (M times) at the time of past calibration. A case has been described in which only _m (x, y) is used to determine whether or not the one radiation detection element (x, y) is a defective pixel from the temporal statistical value. That is, during the dark reading period that is performed a plurality of times (M times), the signal value output from the radiation detection element (x, y) is not affected by the temperature change, or the signal value is not affected by the temperature change. It was considered small enough to be ignored.

  However, if the signal value output from the radiation detection element (x, y) is affected by a temperature change during a plurality of (M times) dark reading periods, it is more accurate to compensate for this temperature change. A good judgment can be made.

Therefore, by using the concept of temperature compensation at the time of calibration in the offset correction value method for each radiation detection element of the radiation image detector described above, the dark read value d output from one radiation detection element (x, y) itself. _m (x, y) and a plurality of radiation image detection elements that are associated with the one radiation detection element (x, y) in advance and change in temperature similarly to the one radiation detection element (x, y). Using the spatial average value w _m (x, y) of each dark read value d _m (x ′, y ′) output from (x ′, y ′), the one radiation detection element (x, A method of determining whether or not y) is a defective pixel and registering it will be described in [Determination Method 4] to [Determination Method 6].

  By adopting this temperature compensation concept, errors due to temperature can be removed when determining defective pixels, so that defective pixels can be determined accurately without being affected by temperature. It becomes.

[Determination Method 4]
The determination method described here is a method in which the concept of temperature compensation is incorporated in the above [determination method 1]. First, in a past calibration during several times (M times) carried out a dark read, in each time of the dark read, one radiation detecting element (x, y) dark output from its own readings d _{m (x,} y) and a plurality of radiation detection elements (x ′, y ′) that are associated with the one radiation detection element (x, y) in advance in the same manner as the one radiation detection element (x, y). ) And the difference e _m (x, y) represented by the following equation (30) with the spatial average value w _m (x, y) of each dark read value d _m (x ′, y ′) output from Is newly defined. That is, the spatial average value w _m (x, y) is used as a temperature compensation variable for the dark read value d _m (x, y).
_{e m (x, y) =} d m (x, y) -w m (x, y) ... (30)

The space of each dark read value d _m (x ′, y ′) output from a plurality of radiation detection elements (x ′, y ′) previously associated with the one radiation detection element (x, y). The average value w _m (x, y) is output from the one radiation detection element (x, y) itself in order to determine whether or not the one radiation detection element (x, y) is a defective pixel. Since it is used in contrast with the dark read value d _m (x, y), a plurality of radiation detection elements (x ′, y ′) that are targets for calculating the spatial average value w _m (x, y). It is preferable not to include the dark read value d _m (x, y) output from the one radiation detection element (x, y) in each dark read value d _m (x ′, y ′) from And a plurality of radiation detection elements (x ′, y ′) that are targets for calculating the spatial average value w _m (x, y) Each dark read values d _{m (x',} y') of et al in, the one of the radiation detecting element (x, y) dark read values output from the d _{m (x,} y) that may be included There is a way of thinking.

  In the present invention, both of these are allowed and are not limited to either one. The same applies to the offset correction value acquisition method described above.

  In addition, a plurality of other radiation detection elements (x ′, y ′) that are associated with the one radiation detection element (x, y) in advance and change in temperature in the same manner as the one radiation detection element (x, y). ) Is selected as described above.

Now, in this way, past calibration at several times (M times) is calculated for each performed a dark read each time the dark reading difference e _{m (x,} y) expressed together the value of the histogram , the one of the radiation detecting element (x, y) if the normal pixel, for example, as shown in FIG. 22, the distribution of the difference e _{m (x,} y) of the temporal fluctuation (variation), the average standard The distribution is a normal distribution having a deviation σe (x, y).

However, the one of the radiation detecting element (x, y) is the defective pixel (xs, ys) If, dark read value output from the one radiation detection device _{(xs, ys) d m (} xs, ys) Since the fluctuation itself is much larger than that in the case of a normal pixel, the distribution of temporal fluctuation (variation) of the difference e _m (xs, ys) has a large standard deviation σe (xs, ys) as shown in FIG. It becomes a distribution of normal distribution having

Therefore, focusing on the standard deviation σe (x, y) in the temporal fluctuation distribution of the difference e _m (x, y), the standard deviation σe (x, y) is larger than a preset threshold σeth. In addition, the one radiation detection element (x, y) can be determined as a defective pixel (xs, ys) and registered.

That is, in the past upon calibration several times (M times) carried out a dark read, in each time of the dark read, one radiation detecting element (x, y) dark output from its own readings d _{m (x,} y) and each dark read value d _m (x ′, y ′) output from a plurality of radiation detection elements (x ′, y ′) previously associated with the one radiation detection element (x, y). spatial average _w m of (x, y) the difference _e m with (x, y) to calculate the standard deviation in the distribution of the temporal fluctuations of the difference _e m (x, y) Sigma] e (x, y) Is calculated as a temporal statistical value, and the standard deviation σe (x, y) is larger than a preset threshold σeth, that is,
σe (x, y)> σeth (31)
Is satisfied, it is determined that the one radiation detection element (x, y) is a defective pixel. In this case, the one radiation detection element (x, y) is registered in the defective pixel map as a defective pixel (xs, ys).

The above threshold σeth, as in the case of the threshold σdth described above, for example, 3 times to 10 times the average standard deviation σe in the distribution of the temporal fluctuations of the difference _{e m (x, y) (} x, y) Although it is preferable to set to a value of about, it is not limited to this value. Further, the variance σe ² (x, y) is used instead of the standard deviation σe (x, y), a threshold is set for the variance σe ² (x, y), and processing is performed in the same manner as described above. Is also possible.

The difference e _m (x, y) is a variable obtained by performing temperature compensation with w _m (x, y) on the dark read value d _m (x, y) as shown in the equation (30). Therefore, the error due to the temperature change of the dark reading value d _m (x, y) is canceled out. Accordingly, the standard deviation σe (x, y) of the temporal fluctuation of the difference e _m (x, y) is equal to the standard deviation σd (x, y) of the temporal fluctuation of the dark read value d _m (x, y). And
σe (x, y) ≦ σd (x, y) (32)
The relationship is established. That is, the spread of the histogram distribution of the difference e _m (x, y) is equal to or smaller than the spread of the histogram distribution of the dark read value d _m (x, y). Thus, it can be said that [Determination Method 4] has a smaller error with respect to temperature change than [Determination Method 1].

[Determination method 5]
The determination method described here is a method in which the concept of temperature compensation is incorporated into the above-mentioned [determination method 2].

As a temporal statistical value, attention is paid to a temporal average value in the distribution of temporal fluctuations of the difference ε _m (x, y). The time mean value of the distribution of the temporal fluctuations of even a difference cases shown above in FIGS. 22 and 23 _e m (x, y) is the aforementioned (12) epsilon (x, y) = δ (x, y) −ω (x, y) (12)
The calculation is equal to the temporal average value ε (x, y) of the temperature-corrected dark reading calculated in (1).

  Therefore, in this case, the temporal average value ε (x, y) of the temperature-corrected dark reading is larger than a preset threshold value εth1 (hereinafter, referred to as a first threshold value εth1), or is larger than the threshold value εth1. When the threshold value is smaller than another threshold value εth2 (hereinafter referred to as the second threshold value εth2) set to a small value, the one radiation detection element (x, y) is determined as a defective pixel (xs, ys), and the defect is detected. It is configured to be registered in the pixel map.

That is, the difference ε (x, y) obtained in the dark reading performed a plurality of times (M times) at the time of the past calibration is not less than the second threshold εth2 and not more than the first threshold εth1, that is,
εth2 ≦ ε (x, y) ≦ εth1 (33)
If so, the magnitude of the dark read value d _m (x, y) output from the one radiation detection element (x, y) is output from the other radiation detection element (x ^* , y ^* ). The dark read value d _m (x ^* , y ^* ) does not vary greatly and is determined to be within the allowable range and to be a normal pixel. In this case, the one radiation detection element (x, y) is not registered as a defective pixel.

However, if the temporal average value ε (x, y) of the temperature-corrected dark reading is larger than the threshold value εth1, that is,
ε (x, y)> εth1 (34)
Or, when the temporal average value ε (x, y) of the temperature-corrected dark reading is smaller than the threshold value εth2, that is,
ε (x, y) <εth2 (35)
Is satisfied, the magnitude of the dark read value d _m (x, y) output from the one radiation detection element (x, y) is output from the other radiation detection element (x ^* , y ^* ). The difference between the dark read value d _m (x ^* , y ^* ) and the read value is large and exceeds the allowable range, and is determined to be a defective pixel. In this case, the one radiation detection element (x, y) is registered in the defective pixel map as a defective pixel (xs, ys).

Note that the temporal average value ε (x, y) of the dark corrected temperature reading is the temporal average value δ (x, y) of the dark reading value d _m (x, y) as shown in the equation (12). Therefore, the error due to the temperature change of the dark read value d _m (x, y) is canceled out. Thus, it can be said that [Determination Method 5] is a method with less error with respect to temperature change than [Determination Method 2].

[Determination method 6]
The determination method described here is a method in which the concept of temperature compensation is incorporated into the above-mentioned [determination method 3]. That is, as a temporal statistical value used for determination, instead of each dark reading value d _m (x, y), it is the absolute value of the difference e _m (x, y) defined by the equation (30) | e _{Use m} (x, y) |.

What is the one radiation detection element (x, y) when the absolute value | e _m (x, y) | of the difference e _m (x, y) is larger than a preset threshold value e _m th Therefore, it is possible to determine that this is a defective pixel (xs, ys) and register it as a defective pixel (xs, ys).

That is, the dark read value d _m (x, y) output from one radiation detection element (x, y) obtained in the dark reading performed a plurality of times (M times) at the time of past calibration and the spatial Similar to the other radiation detection elements (x ^* , y ^* ), the absolute value of the difference e _m (x, y) of the average value w _m (x, y) is equal to or less than the threshold value e _m th, that is,
| E _m (x, y) | ≦ e _m th (36)
If so, the magnitude of the dark read value d _m (x, y) output from the one radiation detection element (x, y) is output from the other radiation detection element (x ^* , y ^* ). The magnitude of the dark read value d _m (x ^* , y ^* ) does not change so much and is within the allowable range and is determined to be a normal pixel. In this case, the one radiation detection element (x, y) is not registered as a defective pixel.

However, if the absolute value of one radiation detecting element (x, y) the difference _e m calculated for (x, y) is larger than the threshold value _{e m} th, i.e.,
| E _m (x, y) |> e _m th (37)
Is satisfied, the magnitude of the dark read value d _m (x, y) output from the one radiation detection element (x, y) is output from the other radiation detection element (x ^* , y ^* ). The difference between the dark read value d _m (x ^* , y ^* ) and the read value is large and exceeds the allowable range, and is determined to be a defective pixel. In this case, the one radiation detection element (x, y) is registered in the defective pixel map as a defective pixel (xs, ys).

The difference e _m (x, y) is a variable obtained by performing temperature compensation with w _m (x, y) on the dark read value d _m (x, y) as shown in the equation (30). Therefore, the error due to the temperature change of the dark reading value d _m (x, y) is canceled out. Thus, it can be said that [Determination Method 6] is a method with less error with respect to temperature change than [Determination Method 1].

  Further, in the above [determination method 1] to [determination method 6], a pixel exceeding a predetermined threshold is not registered as a defective pixel immediately, but is registered as a defective pixel, for example, based on the number of times (frequency). You may make it determine whether to do. Further, a plurality of threshold values are set at a predetermined interval, a histogram (the number of abnormal pixels entering a predetermined interval) for each interval is calculated, and the interval position and the occurrence frequency of abnormal pixels in the interval are calculated. Accordingly, it may be determined whether or not to register a defective pixel.

As described above, according to the defective pixel determining method and defective pixel determining program according to the present embodiment, dark read value d _{m (x,} y) in the distribution of fluctuation of, for example, the standard deviation .sigma.d (x, y) is A radiation detection element that is significantly large, the temporal average value δ (x, y) in the fluctuation distribution is an abnormal value, or the dark read value d _m (x, y) itself is an abnormal value ( It is possible to accurately find x, y), accurately determine that the radiation detection element (x, y) is a defective pixel, and register it in the defective pixel map.

Therefore, by registering such defective pixels in advance, the offset correction value O (x, y) is calculated, or the actual image data F (x, y) is corrected to obtain the final image data F _O. When generating (x, y), it is possible to accurately eliminate the adverse effect of such defective pixels, and to improve the SN ratio of the final image data F _O (x, y). Is possible.

  In addition, by registering defective pixels in advance, it is possible to register and manage them collectively in the form of, for example, a defective pixel map in the same way as a defective pixel group registered as a defective pixel due to other factors. It becomes.

Furthermore, as described above, the dark read value d _m (x, y) output from the radiation detection element (x, y) increases or decreases due to temperature fluctuation. However, according to the above [Determination Method 4], [Determination Method 5], and [Determination Method 6], from each radiation detection element (x, y) itself for each dark reading at the time of past calibration. The dark read value d _m (x, y) that is output and each dark that is output from a plurality of radiation detection elements (x ′, y ′) associated in advance with the one radiation detection element (x, y). A difference ε _m (x, y) from the spatial average value w _m (x, y) of the read value d _m (x ′, y ′) is calculated, and based on the difference, the one radiation detection element (x, y) is calculated. ) Is a defective pixel.

Therefore, by paying attention to the difference ε _m (x, y), the radiation detection element (x, y) is removed in the state where the influence of the value variation due to the temperature variation of the dark read value d _m (x, y) is removed. It is possible to determine whether or not the pixel is a defective pixel, and the defective pixel can be found more stably without being affected by the temperature.

  [Decision method 1] to [determination method 3] or [determination method 4] to [determination method 6] can be used independently to register defective pixels, but [determination method 1] to [Determination Method 3] or a combination of [Determination Method 4] to [Determination Method 6], for example, [Determination Method 1] and [Determination Method 3] or [Determination Method 4] and [Determination Method 6] , [Determination Method 1], [Determination Method 2], [Determination Method 3] or [Determination Method 4], [Determination Method 5], and [Determination Method 6] are used to register defective pixels with higher accuracy. It can be performed.

Also, for each radiographic imaging, immediately before or immediately after the dark read value D of shooting (x, y) (dark read more dark read value if value is a plurality of D _{m (x,} the mean value of y)) the offset When used as the correction value O (x, y), the dark read value D (x, y), the dark read value D _m (x, y), or the dark read value D immediately before and after radiographic imaging. _Using the [determination method 3] or [determination method 6] of the present embodiment for the average value of _m (x, y), the dark reading value D (x, y) or the dark reading value D _m ( It is possible to determine and register defective pixels by determining whether or not the average value of x, y) or the dark read value D _m (x, y) is an abnormal value.

[Abnormal pixel determination method and abnormal pixel determination program]
In the above [determination method 1] to [determination method 6], dark reading values of dark reading performed a plurality of times (M times) at the time of calibration and dark reading performed immediately before or immediately after radiographic imaging are used. Therefore, statistical determination including the occurrence frequency can be performed by sufficiently increasing the value of M. For this reason, it was possible to determine whether or not to determine as a defective pixel or whether or not to register.

  However, when using dark read values D (x, y) immediately before and after radiographic imaging, a sufficient number of dark read values D (x, y) cannot always be used as in calibration. In some cases (especially when dealing with one or two dark scans), the recognized abnormal pixel value is an unusual phenomenon that occurs very rarely, or an abnormal value that can be registered as a defective pixel. In some cases, it is difficult to determine whether or not is frequently occurring.

  Therefore, in this description, a case where such an abnormal pixel is temporarily registered as an abnormal pixel without immediately registering it as a defective pixel will be described. Whether or not to register these abnormal pixels as defective pixels will be described later.

  Hereinafter, an abnormal pixel determination method in the case where [Determination Method 6] is applied to the dark read value D (x, y) immediately before and after radiographic imaging will be described.

[Judgment method 7]
First, the difference E (x, y) between the dark read value D (x, y) and the spatial average value W (x, y) is defined as follows.
E (x, y) = D (x, y) -W (x, y) (38)

  When the absolute value of the difference E (x, y) is expressed as | E (x, y) |, the absolute value | E (x, y) | of the difference E (x, y) is greater than a preset threshold value Eth. Is larger, the one radiation detection element (x, y) can be regarded as having any abnormality, so that it is determined as an abnormal pixel and temporarily stored.

That is, the difference E (x, y) between the dark read value D (x, y) output from one radiation detection element (x, y) immediately before and after radiographic imaging and the spatial average value W (x, y). ) Is equal to or smaller than the threshold Eth, similarly to the other radiation detection elements (x ^* , y ^* ), that is,
| E (x, y) | ≦ Eth (39)
If so, the magnitude of the dark read value D (x, y) output from the one radiation detection element (x, y) is the dark output from the other radiation detection element (x ^* , y ^* ). The read value D (x ^* , y ^* ) is not much different from the size of the read value D (x ^* , y ^* ) and is within the allowable range, and is determined to be a normal pixel.

However, when the absolute value of the difference E (x, y) calculated for one radiation detection element (x, y) is larger than the threshold value Eth, that is,
| E (x, y) |> Eth (40)
Is satisfied, the magnitude of the dark read value D (x, y) output from the one radiation detection element (x, y) is output from the other radiation detection element (x ^* , y ^* ). The dark read value D (x ^* , y ^* ) is greatly different from the allowable range and is determined to be an abnormal pixel.

  The difference E (x, y) is a variable obtained by performing temperature compensation with W (x, y) on the dark read value D (x, y) as shown in the equation (38). The error due to the temperature change of the dark reading value D (x, y) is cancelled.

If it is determined as an abnormal pixel, even if the pixel position is not registered in the defective pixel map, it is temporarily regarded as equivalent to the defective pixel, and, as with the defective pixel, abnormalities are detected by interpolation using surrounding pixel values. By performing a correction process for correcting an abnormal value of the pixel itself, it is possible to prevent an abnormal pixel from being generated in the final image data F _O (x, y).

  The method shown in [Determination method 7] can be further developed into [Determination method 8] shown below.

[Judgment method 8]
First, as in [Determination method 7], the difference E (x, y) between the dark read value D (x, y) and the spatial average value W (x, y) is calculated according to the equation (38).

Next, in the dark reading performed a plurality of times (M times) at the time of past calibration, the temporal average value δ (x) of the distribution of the dark reading value d _m (x, y) calculated according to the equation (12) , Y) and the difference ε (x, y) between the temporal average value ω (x, y) of the distribution of the spatial average value w _m (x, y) of the dark reading value d _m (x, y), The difference Λ (x, y) with the difference E (x, y) calculated by the equation (38) is obtained according to the equation (41).
Λ (x, y) = ε (x, y) −E (x, y) (41)

  Here, when the absolute value of the difference Λ (x, y) is expressed as | Λ (x, y) |, when the absolute value is larger than a preset threshold Λth, the one radiation detection element (x , Y) can be regarded as having an abnormality, so that it is determined as an abnormal pixel and temporarily stored.

That is, the absolute value of the difference Λ (x, y) is equal to or less than the threshold value Λth, like the other radiation detection elements (x ^* , y ^* ), that is,
| Λ (x, y) | ≦ Λth (42)
If so, the magnitude of the dark read value D (x, y) output from the one radiation detection element (x, y) is the dark output from the other radiation detection element (x ^* , y ^* ). The read value D (x ^* , y ^* ) is not much different from the size of the read value D (x ^* , y ^* ) and is within the allowable range, and is determined to be a normal pixel.

However, when the absolute value of the difference Λ (x, y) calculated for one radiation detection element (x, y) is larger than the threshold Λth, that is,
| Λ (x, y) |> Λth (43)
Is satisfied, the magnitude of the dark read value D (x, y) output from the one radiation detection element (x, y) is output from the other radiation detection element (x ^* , y ^* ). The dark read value D (x ^* , y ^* ) is greatly different from the allowable range and is determined to be an abnormal pixel.

  Note that the difference Λ (x, y) is an arithmetic operation between the difference ε (x, y) and the difference E (x, y), both of which are temperature compensated, as shown in the equation (41). Therefore, the error due to the temperature change of the dark read value D (x, y) is cancelled.

If it is determined as an abnormal pixel, even if the pixel position is not registered in the defective pixel map, it is temporarily regarded as equivalent to the defective pixel, and, as with the defective pixel, abnormalities are detected by interpolation using surrounding pixel values. By performing a correction process for correcting an abnormal value of the pixel itself, it is possible to prevent an abnormal pixel from being generated in the final image data F _O (x, y).

  For handling abnormal pixels shown in [Determination method 7] and [Determination method 8], any of the following options 1) to 5) can be employed.

Option 1) It is regarded as an abnormal pixel that occurs accidentally and is not registered as a defective pixel.
Option 2) Immediately register in the defective pixel map as defective pixel (xs, ys).
Option 3) Decide whether or not to register as a defective pixel according to the value of the abnormal pixel value. That is, when the value of an abnormal pixel is very large (when a threshold value for determining registration to a defective pixel determined separately is exceeded), it is registered in the defective pixel map as a defective pixel (xs, ys).
Option 4) As an abnormal pixel, a pixel position is stored separately from the defective pixel. After that, if the signal value at the same pixel position becomes an abnormal pixel with a certain frequency, the occurrence probability is determined at that time and determined in advance. When the generated occurrence probability threshold value is exceeded, it is registered in the defective pixel map as a defective pixel (xs, ys).
Option 5) The above options 3) and 4) are combined.

Regarding the above [Determination Method 4] to [Determination Method 8], the spatial average value w _m (x, y) and the spatial average value W (x, y), which are temperature compensation variables, and these temperature compensation variables are determined. Differences ε (x, y), differences e _m (x, y), differences E (x, y), absolute values | Λ (x, y) | ing. Here, if there is a sudden abnormal pixel that is not registered as a defective pixel in the spatial average value w _m (x, y) or the spatial average value W (x, y), the temperature compensation coefficient There is a concern that the accuracy of the value of the value decreases.

However, in the case of temperature compensation coefficient, the dark reading values (D (x, y) and _D m (x, y) and _d m (x, y)) surrounding the target pixel position (x, y) of Is a spatial average value of N pixel values, and even if there is a sudden abnormal pixel that is not registered as a defective pixel among the N peripheral pixels, the effect is reduced to 1 / N. However, it does not cause a big problem, but a separate means may be provided to detect sudden abnormal pixels and exclude them from the pixels used for calculating the temperature compensation variable.

Up to this point, the method for discriminating defective pixels and abnormal pixels using dark read values has been described. However, as described above, in order to obtain good final image data F _O (x, y), not only the offset correction value O (x, y) and the gain correction value G (x, y) but also the actual image. The image data F (x, y) is also required to be good.

That is, even if the defective pixel and the abnormal pixel are appropriately dealt with with respect to the offset correction value O (x, y) and the gain correction value G (x, y), the pixel of the actual image data F (x, y) If a sudden abnormal pixel is included in the value, the abnormal pixel is also recognized in the final image data F _O (x, y), which is not preferable. Therefore, it is preferable to take appropriate measures not only for defective pixel correction but also for abnormal pixels that occur suddenly for the real image data F (x, y).

  However, since the photographed image data F (x, y) has the following characteristics, it is not as easy to handle as the dark read value described so far.

1) In most cases, the photographed image data F (x, y) needs to be handled as the photographed image data F (x, y) that is taken only once. In some cases, the same subject is photographed multiple times. In this case, since the subject is moved or the direction of the subject is changed during multiple radiographic image capturing, each of the independent photographed image data F (x , Y).

2) Radiation image capturing for acquiring actual image data F (x, y) is solid image capturing for acquiring gain correction data (solid image capturing is a uniform radiation without placing a subject. Except for shooting that is irradiated), the subject is always placed and the subject image is reflected in the real image data F (x, y).

  In addition, as described above, since the actual image data F (x, y) is independent data that is only radiographed once, the value of the actual image data F (x, y) accurately reflects subject information. It is difficult to judge objectively whether or not there is. For example, if radiation image capturing under the same conditions is performed a plurality of times on the same subject (assuming that the subject does not move during the plurality of capturing operations), a plurality of times are performed for a predetermined pixel position (x, y). By comparing the variations in image data, it can be determined whether or not subject information is reflected almost accurately from a statistical point of view.

  For example, when the average value of the signal values of the pixel position (x, y) in multiple radiographic imaging is calculated and the variation of each signal value from the average value is measured, if there is little variation, those signal values Can be determined to reflect the subject information almost accurately. On the other hand, if the value of a certain signal value deviates greatly from the average value with respect to the average value of the signal values at the pixel position (x, y) in multiple radiographic imaging, the signal value is The subject information is not accurately reflected, and it can be determined that the pixel is an abnormal pixel affected by noise or the like.

  In this way, if radiographic image capturing under the same conditions is performed for the same subject a plurality of times, it can be determined whether or not subject information is reflected almost accurately from a statistical viewpoint, but the actual image data F ( Since x, y) is independent data that is radiographed only once, such a statistical judgment cannot be made.

  Therefore, it is not known what value should be originally for each pixel value (value that each pixel value should be, that is, the true value of each pixel value). For example, if it is a dark reading value, it can be determined by statistical processing as described above whether the current data is normal or abnormal with respect to a plurality of data acquired at the time of calibration. There is no way to make the same determination for the real image data F (x, y) that is taken only once.

  A technique for discriminating suddenly abnormal pixels generated in the actual image data F (x, y) and taking appropriate measures in such a problem will be described below.

[Determination method 9]
First, each pixel value of the live-action image data F (x, y) varies depending on the radiation absorption rate of the subject, but this is a signal value that changes very gently from a microscopic viewpoint of the pixel size level. There are many cases (the normal pixel size is often selected in the range of 100 to 200 microns, and the pixel size exceeding 200 microns is often not selected even though the selection is 100 microns or less). In particular, it can be said that there is no pixel in which only a specific pixel changes greatly. In addition, as a lesion having a size closest to the pixel level, there is microcalcification in mammography, which acts in the direction of absorbing radiation, and thus the signal value changes in the direction in which the signal value attenuates.

  Most of the abnormal pixels to be affected here act so that the signal value becomes large (abnormalities that act so that the signal value becomes small are already registered as defective pixels), so they act in the direction of absorbing radiation. A lesion that does not get confused with an abnormal pixel.

  On the other hand, since there are no lesions that act to increase the signal value and have a pixel-level size, an independent pixel that has a large value relative to surrounding pixels is regarded as a sudden abnormal pixel. No problem. By utilizing this characteristic, sudden abnormal pixels are discriminated from the real image data F (x, y).

  Furthermore, a pixel that occurs suddenly is always an isolated pixel. In the first place, abnormal pixels that occur suddenly have a low frequency of occurrence (those with a high frequency of occurrence are already registered as defective pixels), and one flat panel detector usually has a pixel count of around 5 megapixels. Therefore, the probability that the sudden abnormal pixels in question are generated side by side or as a clustered cluster is astronomically small.

  Therefore, as described above, it may be considered that the sudden abnormal pixel taken up as a problem this time is always generated as an isolated pixel. Of course, it goes without saying that the case where they occur next to each other or the case where they occur as a cluster-like lump may be taken into consideration. However, in the following description, it is assumed that they occur as isolated pixels.

  In summary, the suddenly abnormal pixels in the actual photographed image data F (x, y) that are the subject of this time are considered to have the following properties.

1) An abnormal pixel always has a large signal value (behaves like a pixel irradiated with a lot of radiation)
2) An abnormal pixel is an isolated point. The pixels around the abnormal pixel are always normal pixels other than those registered as defective pixels.
3) A normal pixel changes smoothly in the direction in which the signal value increases, and does not show a large pixel change in units of one pixel.

  One embodiment of the method for discriminating the suddenly abnormal pixels will be described below with reference to FIG.

For example, the one radiation detection element (x, y), that is, the pixel position (x, y) as the plurality of radiation detection elements (x, y) is set as the target pixel position of the actual image data. Then, 25 pixel positions (x−2, y−2) to (x + 2, y + 2) existing in a 5 × 5 square area centered on the pixel position (x, y) are defined as comparison areas, and this When the average value of 24 signal values excluding the signal value at the pixel position (x, y) that is the target pixel from the signal value in the comparison area is expressed as A (x, y),

  Here, the value of L is 24 (= 5 × 5-1) in this case. In this example, L = 24 and a 5 × 5 square area centered at the pixel position (x, y) is described as the comparison area. However, the comparison area selection method is not limited to this. As L = 8 (= 3 × 3-1), a 3 × 3 square region centered at the pixel position (x, y) may be used as the comparison region, and for example, a square region having a size of 7 × 7 or larger may be used. Even if it adopts as a comparison field, the effect of the present invention is not spoiled. Further, if the region surrounding the pixel position (x, y) is selected as the comparison region, the comparison region is not necessarily a square region. In order to reduce the number of calculations, the comparison area may be four pixels above and below the pixel position (x, y), or the comparison area may be two pixels above and below the pixel position (x, y) or left and right.

Next, the difference V (x, y) is calculated according to the equation (45).
V (x, y) = F (x, y) −A (x, y) (45)

If this difference V (x, y) is less than or equal to a preset threshold value Vth, that is,
V (x, y) ≦ Vth (46)
If so, the value of the actual image data F (x, y) output from the one radiation detection element is used as it is without being changed.

On the other hand, when the difference V (x, y) is larger than a preset threshold value Vth, that is,
V (x, y)> Vth (47)
If so, the photographed image data F (x, y) output from the one radiation detection element is determined as an abnormal pixel, and the photographed image is obtained with the average value B (x, y) defined in the equation (49). The value of data F (x, y) is replaced (see equation (48)).
F (x, y) ← B (x, y) (48)

Here, nine pixel positions (x−1, y−1) to (x + 1, y + 1) existing in a 3 × 3 square area centered on the pixel position (x, y) are defined as calculation areas, The average value of the eight signal values excluding the signal value at the pixel position (x, y), which is the target pixel, from the signal value in this calculation area is defined as an average value B (x, y).

  In the equation (49), the value of P is 8 (= 3 × 3-1) in this case. In this example, P = 8 and a 3 × 3 square area centered at the pixel position (x, y) is described as the calculation area. However, the calculation comparison area selection method is not limited to this. As P = 24 (= 5 × 5-1), a 5 × 5 square area centered at the pixel position (x, y) may be used as the comparison area, and for example, a square area having a size of 7 × 7 or larger may be used. You may employ | adopt as a comparison area | region. Further, if a region surrounding the pixel position (x, y) is selected as the calculation region, the calculation region is not necessarily a square region. In order to reduce the number of calculations, the calculation area may be four pixels above and below the pixel position (x, y), or the calculation area may be two pixels above and below the pixel position (x, y).

  Here, the comparison area A (x, y) and the calculation area B (x, y) are individually defined (the comparison area and the calculation area do not necessarily match), but the comparison area A (x , Y) and the calculation area B (x, y) may be treated as the same. If a defective pixel exists in the comparison area or calculation area, the average value A (x, y) or average value B (x, y) can be calculated using normal pixels adjacent to the defective pixel. good.

As described above, if an abnormal pixel is suddenly dealt with in the real image data F (x, y), no abnormal pixel is generated in the final image data F _O (x, y). It is possible to stably supply a good diagnostic image. In addition, regarding the handling of the abnormal pixel detected here (whether or not to register as a defective pixel), any of the methods 1) to 5) described after [Determination Method 8] described above should be adopted. Is possible.

[Determination Method 10]
In the above description, the abnormal pixel in the actual image data F (x, y) is described as having a large signal value (behaves like a pixel irradiated with a lot of radiation). Also, the case where the signal value changes in a decreasing direction may be considered. That is,

1) An abnormal pixel always has an abnormal value in a direction of increasing or decreasing with respect to surrounding pixel values.
2) An abnormal pixel is an isolated point. The pixels around the abnormal pixel are always normal pixels other than those registered as defective pixels.
3) Normal pixels change smoothly and do not show large pixel changes in units of one pixel.
Judgment may be made.

That is, an absolute value | V (x, y) | for V (x, y) in the equation (45) is defined, and this absolute value | V (x, y) | is equal to or less than a preset threshold value V′th. If so, ie
| V (x, y) | ≦ V′th (50)
If so, the value of the actual image data F (x, y) output from the one radiation detection element is used as it is without being changed.

On the other hand, when the absolute value | V (x, y) | of the difference V (x, y) is larger than a preset threshold value V′th, that is,
| V (x, y) |> V′th (51)
If so, the photographed image data F (x, y) output from the one radiation detection element is determined as an abnormal pixel, and the photographed image is obtained with the average value B (x, y) defined in the equation (49). The value of the data F (x, y) may be replaced (see formula (48)). Hereinafter, it is the same as that described in [Determination Method 9].

[Determination method 11]
Further, as described above, since the lesion size of the microcalcification lesion in mammography is close to the pixel size level, the photographed image data F (x) that is significantly different from the signal value output from the surrounding pixels after mammography is performed. , Y), even if there is a radiation detection element that outputs, it is difficult to determine whether it is an image of actual microcalcifications present in the lesion or just an abnormal value is output. It is difficult to determine whether or not the radiation detection element can be regarded as an abnormal pixel based on the actual image data F (x, y) obtained by performing mammography.

  However, on the other hand, especially for the radiation image detector 1 used for mammography, it is necessary to know in advance which pixel (radiation detection element) outputs an abnormal signal value when radiation is irradiated. Otherwise, there is a problem that microcalcification is present but erroneously determined as an abnormal value, or the abnormal value is misdiagnosed as microcalcification.

  Therefore, the radiation image detector 1 is uniformly irradiated with radiation in a state where no subject is present, and the actual image data F (x, y) is transmitted from each radiation detection element (x, y). y) is output, and the abnormal pixel is determined in the same manner as in the above [determination method 10] based on each photographed image data F (x, y) acquired in a state where the subject does not exist. be able to.

  That is, in the state where no subject exists, the radiation incident surface X (see FIG. 1) of the radiation image detector 1 is irradiated with radiation so as to have a substantially constant dose over the entire surface, and each radiation detection element (x, It is assumed that the actual image data F (x, y) having substantially the same signal value is to be output from y).

In this state, V (x, y) is calculated according to the above equations (44) and (45) for the actual image data F (x, y) actually output from each radiation detection element (x, y). Then, an absolute value | V (x, y) | for V (x, y) is calculated, and if this absolute value | V (x, y) | is equal to or smaller than a preset threshold value V′th, that is, ,
| V (x, y) | ≦ V′th (50)
If so, it is determined that the one radiation detection element is not an abnormal pixel. In actual mammography, the value of the actual image data F (x, y) output from the one radiation detection element is used as it is without being changed. Further, it is more preferable to determine whether or not the pixel is an abnormal pixel based on the dark reading value, and to use the actual image data as it is only when it is determined that both are not abnormal pixels.

However, the absolute value of the difference V (x, y) is satisfied in spite of the situation where the actual image data F (x, y) having substantially the same signal value is to be output from each radiation detection element (x, y). When V (x, y) | is larger than a preset threshold value V′th, that is,
| V (x, y) |> V′th (51)
If so, the one radiation detection element is determined as an abnormal pixel. In such a case, since the peripheral pixel is an output obtained by detecting normal tissue, it is not possible to use an interpolated value using the value of the peripheral pixel for the abnormal pixel. Announce the presence of abnormal pixels and encourage other radiological image detectors to be used. (2) Display the address of the abnormal pixel and set the area where the abnormal pixel exists as an area outside the radiation field (outside the breast area) An operator such as an engineer is notified via the console that he / she is urged to change the position of the radiation image detector with respect to the subject to perform imaging.

  In the case of the above-described mammography, a pixel whose abnormality determination count is at least once from the past to the present, that is, a pixel whose threshold value for defective pixel determination is one abnormality determination count is registered as a defective pixel. Since it is displayed on the console, the engineer can make the above judgment. Since these defective pixels are not subjected to any processing that is performed on normal pixels, the processing time can be shortened.

  In addition, when the abnormal pixel determination based on the dark read value is used in combination, the threshold value is set in the same manner, and a pixel having the number of times determined to be an abnormal pixel is treated as a defective pixel in the same manner. In regions other than the breast, it is possible to create an alternative value by performing an interpolation process using output values of pixels around the defective pixel, and a threshold value can be set a plurality of times.

So far, the description has been given of performing the abnormal pixel determination method and correction method individually for the offset correction value O (x, y) and the actual image data F (x, y). The pixel determination method and the correction method are not individually applied to the offset correction value O (x, y) and the actual image data F (x, y), but the final image data F _O which is the calculation result thereof. There is also a method implemented for (x, y).

The final image data F _O (x, y) is expressed by the following formula (1): F _O (x, y) = (F (x, y) −O (x, y)) × G (x, y) (1) )
Therefore, if there is an abnormal pixel in the offset correction value O (x, y) or the photographed image data F (x, y), the final image data F _O (x, y) also becomes an abnormal pixel. .

In most cases, the abnormal pixel changes in the direction in which the pixel value of the offset correction value O (x, y) or the photographed image data F (x, y) increases, so that the offset correction is performed in the calculation result of the above formula (1). If there is an abnormal pixel in the value O (x, y), the signal value of the final image data F _O (x, y) changes in a decreasing direction, and the actual image data F (x, y) has an abnormal pixel. For example, the signal value changes in the direction in which the signal value of the final image data F _O (x, y) increases.

For this reason, as a determination method, if the real image data F (x, y) of the above [determination method 10] is replaced with the final image data F _O (x, y), it will be described in [determination method 10]. It is possible to determine abnormal pixels by the same method as described above.

[Calculation method of spatial average when there are defective pixels]
Next, in the case where a radiation detection element having a defect (hereinafter referred to as a defective pixel) is included in another radiation detection element associated in advance with one radiation detection element (x, y), the one radiation detection element. A method of calculating the spatial average value W (x, y) as the temperature compensation variable of (x, y) will be described.

  In order to make the following explanation easy to understand, here, as shown in FIG. 24, the one radiation detection element (target pixel) for calculating the spatial average value W (x, y) as a temperature compensation variable is radiation. 7 × 7 radiation detection elements (1, 1) to (7, 7) centered on the radiation detection element (4, 4) are associated with the detection element (4, 4), A case where the radiation detection element (6, 6) is a defective pixel will be described.

により放射線検出素子（４，４）の空間的平均値Ｗ（４，４）を算出することが考えられる。しかし、この算出手法では、Ｗ（４，４）が欠陥画素（６，６）のダーク読取値Ｄ（６，６）の影響を受けてしまう可能性が残る。 [Calculation method 1]
The simplest calculation method includes a plurality of radiation detection elements (1, 1) to (7, 7) associated with the one radiation detection element (4, 4) in advance including the defective pixel (6, 6). 7) Using the dark reading values D (1, 1) to D (7, 7) output from 7),

It is conceivable to calculate the spatial average value W (4, 4) of the radiation detection element (4, 4) by the above. However, with this calculation method, there is a possibility that W (4, 4) is affected by the dark read value D (6, 6) of the defective pixel (6, 6).

により放射線検出素子（４，４）の空間的平均値Ｗ（４，４）を算出することが可能である。 [Calculation method 2]
Output from the radiation detection elements excluding the defective pixel (6, 6) among the plurality of radiation detection elements (1, 1) to (7, 7) previously associated with the one radiation detection element (4, 4). Using only the dark read values D (1,1) to D (7,7) (excluding D (6,6)),

Thus, the spatial average value W (4, 4) of the radiation detection element (4, 4) can be calculated.

[Calculation method 3]
The value of the dark read value D (6, 6) output from the defective pixel (6, 6) is replaced with the dark read value output from the radiation detection element in the vicinity of the defective pixel (6, 6) to obtain a spatial average. It is also possible to configure to calculate the value W (4, 4).

For example, the dark reading value D (6, 6) output from the defective pixel (6, 6) is used as the dark reading value output from the radiation detection element (5, 6) in the vicinity of the defective pixel (6, 6). Using the value D (5,6)
D (6,6) ← D (5,6) (54)
Replace with Then, the spatial average value W (4, 4) of the radiation detection element (4, 4) is calculated according to the above equation (52).

  In this case, the dark reading value D (6, 6) output from the defective pixel (6, 6) is converted into the radiation detection element (5, 6) adjacent to the defective pixel (6, 6) inside the region R. For example, a radiation detection element that exists at a position outside the region R, such as the radiation detection element (8, 6) shown in FIG. , 6) can be configured to replace the dark read value D (6, 6) output from the defective pixel (6, 6) with the dark read value D output from the radiation detection element adjacent to the pixel. .

[Calculation method 4]
The dark read value D (6, 6) output from the defective pixel (6, 6) is interpolated with each dark read value D output from a plurality of radiation detection elements in the vicinity of the defective pixel (6, 6). The spatial average value W (4, 4) may be calculated.

  For example, the dark read value D (6, 6) output from the defective pixel (6, 6) is used as the radiation detection element (5, 5), (6, 5) in the vicinity of the defective pixel (6, 6). , (7,5), (5,6), (7,6), (5,7), (6,7), (7,7) , D (6,5), D (7,5), D (5,6), D (7,6), D (5,7), D (6,7), D (7,7) For example, the average value of these values is calculated, and the value of D (6,6) is interpolated with the average value. Then, the spatial average value W (4, 4) of the radiation detection element (4, 4) is calculated according to the above equation (52).

  In this case, for example, when the defective pixel is a radiation detection element (7, 6) present at the end of the region R as shown in FIG. 26, the defective pixel (7, 6) is output. The dark read value D (7,6) is calculated from the radiation detection elements (5,5), (6,5), (7,5), (5,6) in the vicinity of the defective pixel (7,6). , (6,6), (5,7), (6,7), (7,7), the dark read values D (5,5), D (6,5), D (7,7) 5), D (5,6), D (6,6), D (5,7), D (6,7), D (7,7) can be used for interpolation. is there.

  In addition, as shown in FIG. 26, radiation detection elements (8, 5), (8, 6), (8) that are radiation detection elements that are located outside the region R and are close to the defective pixel (7, 6). , 7) using the dark read values D (8, 5), D (8, 6), D (8, 7) output from the defective pixels (7, 6). The values of (7, 6) are set to the radiation detection elements (6, 5), (7, 5), (8, 5), (6, 6), (8, 6), (6, 7), (7 , 7), (8, 7), the dark read values D (6, 5), D (7, 5), D (8, 5), D (6, 6), D (8, 6) ), D (6, 7), D (7, 7), and D (8, 7) may be used for interpolation.

  In each of the calculation methods described above, the case where there is one defective pixel, that is, one point defect in the region R has been described. However, the same calculation method can be used when there are a plurality of point defects. . A similar calculation method can also be used when there is a line defect in which defective pixels exist in a line.

[Defect pixel determination in radiation image detector]
[Determination method 1] to [Determination method 8] described above all use dark reading values. That is, since it is possible to determine and register a defective pixel without irradiation with radiation, it can be automatically performed without human intervention. For example, if this method is used at the time of shipping inspection, the inspection cost can be reduced. If this method is used for calibration after installation in a user facility, newly generated defective pixels can be automatically registered in the defective pixel map without the user irradiating radiation.

  Further, for example, as shown in FIG. 1, the radiographic image detector 1 is provided with an input button 23 for enabling an instruction from the user to be input. By operating and registering newly generated defective pixels in the defective pixel map automatically, the defective pixel map can be updated very easily. Furthermore, if the program of this method operates regularly and automatically registers newly generated defective pixels in the defective pixel map, even if the user does not give instructions every time, automatically, The defective pixel map can be updated.

[Image generation in radiation image detector]
Here, a case will be described in which final image data is generated by the radiation image detector using the principle of obtaining the offset correction value of each radiation detection element of the radiation image detector. In this embodiment, as described above, the dark reading for obtaining the offset correction value used for the offset correction of the radiographed image data is performed at least once before or after the radiographic imaging. The offset correction value O (x, y) of each radiation detection element (x, y) is calculated based on the dark reading value acquired by the dark reading.

  In the following description, the case where the dark reading is performed only once after the radiographic image capturing will be described. However, the dark reading may be performed before the radiographic image capturing, and the dark reading is performed about twice. It is also possible to do. It goes without saying that even if the dark reading is performed twice or more, the power consumption increases, but the other effects of the present embodiment can be obtained.

  The image generation using the radiation image detector 1 is performed according to the flowchart shown in FIG. 27, and includes a defective pixel determination step (step S1), a real image data acquisition step (step S2), and a dark reading step. (Step S3), temperature compensation variable calculation step (step S4), temporal average value calculation step (step S5), offset correction value calculation step (step S6), and image correction step (step S7). After that, final image data can be obtained. In the present embodiment, the statistical value is the above-described spatial average value, temporal average value, or the like. However, it is possible to adopt other methods of taking statistical values, which will be described later.

  In the defective pixel determination step (step S1), based on the above-described defective pixel determination method, the result of dark reading performed a plurality of times at the time of past calibration, or dark performed one or more times immediately before or after radiographic imaging From the result of reading, it is determined whether or not all of the plurality of radiation detection elements (x, y) arranged two-dimensionally in the sensor panel unit 4 of the radiation image detector 1 are defective pixels. Information such as the number (x, y) of the radiation detection element determined to be present is registered in the defective pixel map stored in the storage means 7 of the radiation image detector 1.

  In addition, even when the pixel is determined to be abnormal based on the above-described abnormal pixel determination method and the pixel position is not registered in the defective pixel map, it is temporarily regarded as equivalent to the defective pixel and correction processing is performed in the same manner as the defective pixel. The information such as the number (x, y) is temporarily stored in the storage means 7 of the radiation image detector 1 and is treated in the following processing as equivalent to a defective pixel.

When the defective pixel determination step (step S1) is performed by an external device such as the console 32, it is output from each radiation detection element (x, y) by dark reading at the time of calibration by a wired method or a wireless method. The dark reading value d _m (x, y) is transmitted to the external device. Information such as the number (x, y) of the radiation detection element determined to be a defective pixel is stored and registered in a storage means (not shown) of the console 32 or a storage means 38 (see FIG. 28) of the server means 39. The

  Further, the defective pixel determination step (step S1) may be configured to be performed after the actual image data acquisition step (step S2), the dark reading step (step S3), or the like.

  In the actual image data acquisition step (step S2), the radiographic image detector 1 is illustrated with a subject part such as a patient's hand or chest (not shown) placed or approached on the radiation incident surface X (see FIG. 1) side. Ordinary radiation irradiation is performed by irradiating radiation from a radiation generator that does not.

  As described above, the radiation image detector 1 uses a photoconductive material such as a-Se as a detection means to directly convert radiation energy into charges, and the charges are converted into two-dimensionally arranged TFTs or the like. A direct method of reading out electrical signals in pixel units by a switch element for signal readout, or converting radiation energy into light with a scintillator or the like, and converting this light into electric charge with a photoelectric conversion element such as a photodiode arranged two-dimensionally An indirect method for reading out an electrical signal by a TFT or the like is well known.

  At this time, if the radiation image detector 1 is an indirect method, the radiation transmitted through the subject is converted into light by the scintillator layer 3 and is incident on a plurality of radiation detection elements (x, y) arranged in pixel units. Then, after being photoelectrically converted by the radiation detection element (x, y), it is read out as an electric signal in units of pixels, and actual image data F (x, y) is generated. If the radiation image detector 1 is a direct system, the radiation that has passed through the subject is directly converted into electric charges by a photoconductive substance such as a-Se, and is read out as an electrical signal in pixel units by the plurality of radiation detection elements 14. Actual captured image data F (x, y) is generated.

  Then, by the signal reading operation, the actual image data F (x, y) is sequentially read as an output value from each radiation detection element (x, y) (each pixel), and each radiation is detected by the control means 6 of the radiation image detector 1. The detection element number (x, y) and each photographed image data F (x, y) are associated with each other and stored in the storage unit 7.

  In the dark reading step (step S3), subsequently, dark reading is performed at least once after radiographic image capturing. That is, the read voltage is sequentially applied to each scanning line Ll of the radiation image detector 1 in the state where no radiation is irradiated, and the charge accumulated in each radiation detection element (x, y) is converted into the dark read value D (x, y). In the same manner as described above, each dark read value D (x, y) is associated with the number (x, y) of each radiation detection element by the control means 6 of the radiation image detector 1 and stored in the storage means 7. Saved in.

  In addition, the abnormal pixel determination program stored in the storage unit 7 is read, and the abnormal pixel determination method described above (see [Determination Method 7] to [Determination Method 11]) is executed in accordance with the program. Based on the image data F (x, y) and the dark read value D (x, y), it is determined whether or not each radiation detection element (x, y) is an abnormal pixel.

  If there is a radiation detection element (x, y) that can be determined as an abnormal pixel, the number (x, y) is temporarily stored in the storage means 7 and determined as an abnormal pixel according to the above equation (48) or the like. The photographed image data F (x, y) output from the radiation detection element (x, y) thus obtained is replaced with a normal value and stored in the storage means.

  A temperature compensation variable calculation step (step S4), a temporal average value calculation step (step S5), an offset correction value calculation step (step S6), and an image correction step (step S7) described later are performed by an external device such as the console 31. In this case, the actual image data F (x, y) and the dark read values D (x, y) associated with the number (x, y) of each radiation detection element are respectively obtained by a wired method or a wireless method. It is transmitted from the radiation image detector 1 to an external device.

  In the temperature compensation variable calculation step (step S4), the dark read value D (x, y) output from one radiation detection element (x, y) on the sensor panel unit 4 and the one radiation detection element (x , Y), and the dark read values D (x) output from the plurality of radiation detection elements (x ′, y ′) associated with the one radiation detection element (x, y) in advance. ′, Y ′) and the temperature compensation variable that is a spatial average value for the one radiation detection element (x, y) in the dark reading acquired at the time of radiographic imaging this time according to the above equation (3). W (x, y) is calculated. The temperature compensation variable W (x, y) is calculated for all the radiation detection elements (x, y) arranged two-dimensionally on the sensor panel unit 4.

  At this time, in calculating the temperature compensation variable W (x, y) for one radiation detection element (x, y), among the other radiation detection elements previously associated with the one radiation detection element (x, y). When a defective pixel is included, the dark read value D (xs, ys) output from the defective pixel using any one of [Calculation method 1] to [Calculation method 4] described above is radiation in the vicinity thereof. The temperature compensation variable W (x, y) of the one radiation detection element (x, y) is calculated by replacing or interpolating with the dark reading value output from the detection element.

In the temporal average value calculation step (step S5), the dark reading output from one radiation detection element (x, y) itself obtained in the dark reading performed a plurality of times (M times) at the time of past calibration. A time average value δ (x, y) of M times of the value d _m (x, y) is calculated according to the above equation (7). Also, in the temporal average value calculation step (step S5), each dark outputted from a plurality of radiation detection elements (x ′, y ′) associated in advance with the one radiation detection element (x, y). The temporal average value ω (x, y) of M times of the spatial average value w _m (x, y) of the read value d _m (x ′, y ′) is determined according to the above equation (9) or (10). Calculated.

  This temporal average value calculation step (step S5) is not necessarily performed after the temperature compensation variable calculation step (step S4), or after the actual image data acquisition step (step S2) and the dark reading step (step S3). It is not necessary to be arranged, and it is also possible to configure so as to be performed in advance at an appropriate timing.

  In the offset correction value calculation step (step S6), from the temporal average value δ (x, y) and temporal average value ω (x, y) calculated in the temporal average value calculation step (step S5), A difference between them, that is, a temporal average value ε (x, y) (= δ (x, y) −ω (x, y)) of dark correction after temperature correction is calculated. Then, the calculated ε (x, y) and the temperature compensation variable W (x, y) relating to the one radiation detection element (x, y) calculated in the temperature compensation variable calculation step (step S4) are added. Then, an offset correction value O (x, y) for the one radiation detection element (x, y) is calculated. Further, the offset correction value O (x, y) is calculated for all the radiation detection elements (x, y) arranged two-dimensionally on the sensor panel unit 4.

In the image correction step (step S7), the offset correction value described above is obtained from the actual image data F (x, y) read from each radiation detection element (x, y) in the above-described actual image data acquisition step (step S2). The actual image data F (x, y) is corrected by subtracting each offset correction value O (x, y) calculated for each radiation detection element (x, y) in the calculation step (step S6). Then, the real image data F (x, y), the offset correction value O (x, y), and the gain correction value G (x, y) calculated in advance for each radiation detection element (x, y) are obtained. Substituting into the above equation (1), final image data F _O (x, y) is generated for each radiation detection element (x, y).

In this way, the characteristic variation for each pixel (radiation detection element (x, y)) is corrected with respect to the actual captured image data F (x, y) generated in the actual captured image data acquisition step (step S2). Image data F _O (x, y) is generated.

Note that the dark read value d _m (x) output from one radiation detection element (x, y) itself obtained by dark reading at the time of past calibration processed in the offset correction value calculation step (step S6). , Y), the actual image data F (x, y) processed in the image correction step (step S7), and the like, are data output from the radiation detection elements that are determined to be defective pixels and registered. In some cases, the data output from the defective pixel may be subjected to processing such as replacement or interpolation in the same manner as the method for calculating the spatial average value when there is a defective pixel. It is performed appropriately.

  If comprised as mentioned above, according to the principle of offset correction value O (x, y) acquisition of each radiation detection element (x, y) of the radiation image detector 1 mentioned above, dark reading will be carried out before or after radiographic imaging. It is possible to acquire an effective offset correction value O (x, y) that approximates the true value for each radiation detection element (x, y) of the radiation image detector 1 by performing at least once.

In addition, when photographing a subject, the number of times of dark reading can be suppressed to at least one. Further, when data is transmitted from the radiation image detector (FPD) 1 to an external device, pixels (radiation detection elements (x, Since only the final image data F _O (x, y) whose characteristic variation for each y)) is corrected needs to be transmitted only once, there is no need to transmit the dark read value D (x, y) or the like. Thus, it is possible to reduce power consumption required for data transmission.

In particular, the radiation image detector 1 is the built-in battery, even when the image data F _{O (x,} y) and such that the wireless communication, since the transmission of the image data F _{O (x,} y) is only once, It is possible to reduce power consumption and prevent the internal battery 21 from being consumed. Further, an effective offset correction value O (x, y) equivalent to that of the conventional method is acquired for each radiation detection element (x, y), and good image data F _O (x, y) can be obtained. Become.

Further, in the defective pixel determination step (step S1), by using the defective pixel determination method according to the above-described embodiment, it is accurately determined whether or not the radiation detection element is a defective pixel, and is determined to be a defective pixel. Since the offset correction value O (x, y) is calculated by performing appropriate processing such as replacement and interpolation on the registered radiation detection element, the effective offset correction value O (x, y) is calculated. Can be obtained accurately. Therefore, it is possible to improve the SN ratio of the finally obtained image data F _O (x, y).

In addition, in the image generation using the radiation image detector of the present embodiment and the above-mentioned principle that is the premise thereof, the radiation detection element (x, y) is associated with the dark reading before or after the radiation image capturing. A spatial statistical value is calculated as a temperature compensation variable W (x, y) that is a spatial average value of the dark read values D (x ′, y ′) output from the plurality of radiation detection elements (x ′, y ′). Explained when to do. Further, the temporal average value δ (x, y) of the dark read value d _m (x, y) output from the radiation detection element (x, y) in a plurality of dark readings in the past calibration, and Spatial average value w _m (x) of dark read values d _m (x ′, y ′) output from a plurality of radiation detection elements (x ′, y ′) associated with the radiation detection elements (x, y). , Y) The case where the temporal statistical value is calculated as the temporal average value ω (x, y) has been described.

  However, it is possible to adopt various methods other than the calculation method of taking a simple average for taking the spatial statistical value and the temporal statistical value.

  That is, as a method of taking the spatial statistical value, for example, among the plurality of radiation detection elements (x ′, y ′) previously associated with one radiation detection element (x, y), the one radiation detection element The weighted average value is weighted so that the weight for each dark read value (x ′, y ′) output to the radiation detection element (x ′, y ′) closer to (x, y) is larger. Can be calculated as a spatial statistical value.

  Further, for example, the median value of each dark read value (x ′, y ′) output from a plurality of radiation detection elements (x ′, y ′) associated in advance with one radiation detection element (x, y). It is also possible to configure so that (median value) is calculated as a spatial statistical value.

In addition, as a method of taking a temporal statistical value, for example, in dark reading performed a plurality of times during past calibration, a plurality of dark reading values d _m output from one radiation detection element (x, y) itself. (X, y) and each dark read value d _m (x ′, y ′) output from a plurality of radiation detection elements (x ′, y ′) previously associated with the one radiation detection element (x, y). Among the temperature compensation variables w _m (x ′, y ′), which are a plurality of spatial statistics values of y ′), the darker readings d _m (x, y) and spatial statistics are obtained for data closer to the present in time. It is possible to weight the value w _m (x ′, y ′) so as to increase the weight and calculate the weighted average value as a temporal statistical value.

Further, for example, a plurality of dark reading values d _m (x, y) output from one radiation detection element (x, y) itself or the one radiation detection element (x, y) are associated in advance. The median value of a plurality of spatial statistical values w _m (x ′, y ′) of each dark read value d _m (x ′, y ′) output from the plurality of radiation detection elements (x ′, y ′). May be calculated as a temporal statistical value.

Note that the weighted average value and the median value are used instead of the spatial average value w _m (x, y) and the spatial average value W (x, y) used in the above defective pixel determination method. Is possible.

[Defective pixel determination system and radiation image generation system]
As described above, all of the above-mentioned [determination methods 1] to [determination method 8] use dark reading values and can determine and register defective pixels without irradiation with radiation. It can be implemented automatically without intervention. In addition, although the above-mentioned [determination method 9] to [determination method 11] involve irradiation of radiation, if the user presses a switch for starting determination, for example, the irradiation image is automatically irradiated and real image data is obtained. It is possible to determine whether or not the radiation detection element is an abnormal pixel based on the acquired information and register it as a defective pixel.

  In the above [Defect pixel determination in the radiation image detector], the case where the determination and registration of the defective pixel is performed by the radiation image detector 1 has been described. However, the determination and registration of the defective pixel can be performed as a system including the radiation image detector 1 and a console. Below, an example of embodiment of the defective pixel determination system which concerns on this invention is shown.

  In the present embodiment, the defective pixel determination system 30 is constructed in a radiation image generation system including a radiation image detector 1 (see FIG. 1 and the like), a console 31, and server means 39, as shown in FIG. ing. In the following, a case where the console 31 functions as a defective pixel determination device that determines a defective pixel will be described. However, the server unit 39 or the like can be configured to have the function of a defective pixel determination device. is there.

  The radiation image detector 1 is used by being loaded into a holding portion 33a of a bucky device 33 provided in the photographing room R1. The bucky device 33 is provided with a small operation unit 33b similar to a portable information terminal.

  In the present embodiment, one radiation source 34a of the radiation generator 34 is shared and used by the respective bucky devices 33, and the radiation image detector 1 is loaded into the bucky device 33 and used. In this case, when radiation is emitted from the corresponding radiation source 34a, the radiation generation timing control of the radiation generator 34 is linked with the control means 6 of the radiation image detector 1, and the radiation generator 34 Various controls of the radiation image detector 1 are performed based on the timing of radiation generation. In addition, it is also possible to constitute so that each radiation source 34a of the radiation generator 34 is provided in association with each Buckie device 33.

  The radiation image detector 1 can be used alone in a free state without being loaded into the bucky device 33. In this case, for example, the radiographic image detector 1 can be used on the bed-type bucky device 33, or the patient can use the radiographic image detector 1 with his / her hand.

  The radiation image detector 1 is provided with an antenna device 3 (see FIG. 1) which is a communication means for transmitting the dark read value D (x, y) or the like in a wireless manner. Further, the radiation image detector 1 is connected to an electrode (not shown) provided on the holding unit 33a when the radiation image detector 1 is loaded on the holding unit 33a of the bucky device 33, and the dark read value D (x, The terminal 13 (see FIG. 2) which is a communication means for transmitting y) etc. in a wired manner and a power supply means to the radiation image detector 1 is present in the antenna device 3 and the power switch 11 of the radiation image detector 1. It is provided in the side part opposite (facing) to the side part of the housing 2 to be operated. Further, a terminal 22 which is a power supply means for charging the battery 21 is provided on the same side surface portion as the terminal 13.

  As described above, when the radiation image detector 1 is loaded in the bucky device 33, the terminal 13 and the electrode of the holding unit 33 a are connected, and the dark read value D (x, y) and the like are transmitted via the cable 32. It is sent to the relay terminal 35 in a wired manner and is sent to the console 31 via the relay terminal 35. The radiographic image detector 1 is configured to supply power from the relay terminal 35 to the radiographic image detector 1 via the cable 32 and the terminal 13 when the radiographic image detector 1 is loaded in the bucky device 33.

  Further, when the radiation image detector 1 is used alone without being loaded into the bucky device 33, the dark read value D (x, y) or the like is transmitted via the antenna device 3 in a wireless manner. Yes. The radiographing room R1 is provided with a relay terminal 35 including a radio antenna 36 that relays when the radiological image detector 1 communicates the dark read value D (x, y) or the like to the console 31 in a wireless manner. Information such as the dark reading value D (x, y) transmitted from the antenna device 3 of the radiation image detector 1 by the wireless method is received by the wireless antenna 36 and transmitted to the console 31 via the relay terminal 35. It has become so.

  The radiographic image detector 1 is supplied with power from the relay terminal 35 via the cable 32 and the terminal 13 when loaded in the bucky device 33, but when used alone, the built-in battery 21 is used. It is designed to operate with the power of Note that the relay terminal 35 is provided with a terminal which is a power supply means (not shown) for charging the battery. When the terminal 22 of the radiation image detector 1 is brought into contact with the terminal on the relay terminal 35, the relay terminal 35 is not shown. Electric power is supplied to the battery 21 of the radiation image detector 1 via the terminal 22 and the battery 21 is charged. However, it is not limited to this form.

  The radiographic image detector 1 includes a tag (not shown). In this case, a so-called RFID (Radio Frequency IDentification) tag is used as the tag, and the tag includes a control circuit that controls each part of the tag and a storage unit that stores unique information such as the ID of the radiation image detector 1. Built in compact.

  A tag reader 37 that reads the RFID tag of the radiation image detector 1 is installed in the vicinity of the entrance of the front chamber R2. The tag reader 37 transmits predetermined instruction information on radio waves or the like via a built-in antenna (not shown), detects the radiation image detector 1 entering or leaving the front chamber R2, and detects the radiation image detector 1 An ID or the like is transmitted to the console 31.

  In addition, although the example in the case of incorporating an RFID tag in the radiation image detector 1 is shown, instead of incorporating the RFID tag in the radiation image detector 1, a radiation image is formed on the outer surface of the housing of the radiation image detector 1. A bar code in which the ID of the detector 1 and the like are written may be attached and read with a bar code reader. In this case, the tag reader 37 becomes a barcode reader, reads the barcode of the radiation image detector 1 entering or leaving the front chamber R2, and transmits the ID of the radiation image detector 1 to the console 31.

  The tag reader 37 and the barcode reader are installed in the vicinity of the entrance of the imaging room R1 instead of in the vicinity of the entrance of the front room R2, and the entrance / exit of the radiation image detector 1 to the imaging room R1 is managed. good.

  In this way, by using the RFID or the barcode, the radiation image detector 1 is informed of which radiation generating device (by which the radiation image detector 1 enters or leaves the anterior chamber R2 or the radiographing room R1 is notified to the console 31. In the case of this example, it is possible to automatically notify the console 31 and the radiation generator 34 of whether or not it is necessary to link with the radiation generator 34). Such management of entering or leaving the radiographic image detector 1 using the RFID or barcode functions effectively in a facility where there are a plurality of radiographing devices and radiation generators. When RFID or barcode is not used, the user may directly input information on entering or leaving the radiographic image detector 1 into the imaging room or the like to the console 31.

  The front chamber R2 is provided with a console 31 that controls the defective pixel determination system 30 and the entire radiation image generation system. The console 31 includes main units of the relay terminal 35, the tag reader 37, and the radiation generator 34 described above. 34 b and the like are connected, and a bucky device 33 and the like are connected via the relay terminal 35.

  The console 31 includes a computer (not shown), a storage unit such as a ROM (Central Processing Unit), ROM, RAM, and hard disk, an input / output interface, and the like connected to a bus.

  The console 31 is connected to server means 39 comprising a computer via a network NW. The server means 39 is connected to a storage means 38 composed of a hard disk or the like.

  In the storage means 38 and the storage means (not shown) of the console 31 itself, in order to calculate the spatial average value described above for each radiation image detector 1 usable in the radiation image generation system, each radiation detection element (x, Information on which radiation detection element (x ′, y ′) is associated with y) is stored in advance in association with the ID of the radiation image detector 1.

Further, the console 31 from the radiographic image detector 1, the radiation detection elements in the read dark made during calibration (x, y) dark read values output from the d _{m (x,} y) and radiographic imaging When the dark reading values D (x, y) and D _m (x, y) obtained in the dark reading performed one or more times immediately before or after are transmitted by the wired method or the wireless method, In association with the ID of the radiation image detector 1, it is stored in its own storage means or the storage means 38 of the server means 39.

Further, when the dark read value d _m (x, y) or the like of the radiation image detector 1 is transmitted from the console 31, the server means 39 stores them in the storage means 38 and also stores the radiation detector 1 of the radiation detector 1. The temporal average value δ (x, y) of the dark read value d _m (x, y) output from each radiation detection element (x, y) and the radiation detection element (x, y) are associated with each other. Time of spatial average value w _m (x ′, y ′) as a temperature compensation variable of each dark read value d _m (x ′, y ′) output from a plurality of radiation detection elements (x ′, y ′) The statistical value (temporal average value) ω (x, y) is calculated and stored in the storage means 38.

  The dark read value d (x, y) can be stored and the temporal average value δ (x, y) and the like can be calculated and stored by the console 31 or the storage means of the console 31. is there.

  The console 31 reads a program necessary for executing various processes stored in a storage means such as a ROM or a hard disk, expands it in a work area of the RAM, and executes the process according to the program.

  In addition, the above-described defective pixel determination program is stored in a storage unit such as a ROM or a hard disk. In this embodiment, the console 31 as a defective pixel determination device performs the above defective pixel determination according to the defective pixel determination program. The method is supposed to be executed.

  In addition, it is possible to perform the defective pixel determination by any of the determination methods [Determination Method 1] to [Determination Method 8] of the defective pixel determination method described above or a determination method in which those determination methods are combined. is there. Further, the radiation detection element (x, y) is abnormally caused by irradiating the radiation image detector 1 with radiation from the radiation source 34a of the radiation generator 34 in the state where the subject exists in accordance with [Determination method 9] or [Determination method 10]. It is determined whether or not the pixel is an abnormal pixel, and when it is determined that the pixel is an abnormal pixel, the radiation detection element (x, y) when the actual image data F (x, y) is an abnormal value exceeding a preset threshold value. It is also possible to configure so as to determine a defective pixel.

  In the present invention, the defective pixel determination program is configured to perform the abnormal pixel determination described in [Determination Method 11] among the above-described abnormal pixel determination methods. When the photographed image data F (x, y) output from the radiation detection element (x, y) of the device 1 is an abnormal value exceeding a preset threshold value, and is output more than a preset number of times. The radiation detection element (x, y) is determined as a defective pixel and registered in the defective pixel map.

  Specifically, first, the user sets the radiation image detector 1 so that radiation is uniformly emitted from the radiation source 34a of the radiation generator 34 to the radiation image detector 1 in a state where no subject exists. The radiation image detector 1 is set at a predetermined position, for example, by being loaded into the holding unit 33a of the bucky device 33.

  Then, when a signal instructing the start of determination is input by the user's operation, the console 31 irradiates the radiation image detector 1 with radiation from the radiation source 34a of the radiation generator 34 in a state where no subject exists. Each real image data F (x, y) output from each radiation detection element (x, y) is transmitted from the radiation image detector 1. When the actual captured image data F (x, y) is transmitted from the radiation image detector 1 via the relay terminal 35, the console 31 transmits the actual captured image data F (x, y) to the corresponding radiation image detector 1. The ID is stored in its own storage means or in the storage means 38 of the server means 39 (hereinafter simply referred to as storage means).

  Subsequently, the console 31 calculates the difference V (x, y) according to the above equations (44) and (45) with respect to each photographed image data F (x, y) according to the above-described [determination method 11]. An absolute value | V (x, y) | with respect to the difference V (x, y) is calculated, and the absolute value | V (x, y) | is equal to or less than a preset threshold value V′th according to the above equation (50). If there is, it is determined that the radiation detection element (x, y) that has output the photographed image data F (x, y) is not an abnormal pixel, but the difference V (x, y) as shown in equation (51). If the absolute value | V (x, y) | is larger than the threshold value V′th, it is determined that the radiation detection element (x, y) that has output the actual image data F (x, y) is an abnormal pixel. To do.

  And the console 31 memorize | stores the information of the radiation detection element (x, y) determined to be this abnormal pixel temporarily in a memory | storage means.

  Note that, as described above, when the [determination method 11] is applied particularly to abnormal pixel determination and defective pixel determination in the radiographic image detector 1 used for mammography or the like, there is a microcalcification lesion in actual mammography. However, it is effective because it is possible to accurately prevent the abnormal value from being misjudged as an abnormal value or misdiagnosing the abnormal value as a microcalcification lesion although it is an output value in which the lesion is detected. In addition to the normal defective pixel map, the storage means is provided with a defective pixel map for a state where no subject exists, that is, a defective pixel map for mammography, and the defective pixel map is used for the mammography defective pixel map. It is preferable to be configured to register.

Furthermore, using at least one method described in the determination methods 1 to 8, abnormal / defective pixel determination based on the dark reading value is performed, and combined with the abnormal / defective pixel determination result by the determination method 11,
The engineer can recognize the presence of abnormal / defective pixels in the image that may produce an output value that is almost equivalent to the output value corresponding to the microcalcification lesions by normal pixels. It is possible to maintain the diagnostic accuracy by devising the arrangement of the region of interest with respect to the detector (to prevent abnormal / defective pixels from coming into the region of interest) or using a new detector.

  As described above, in the live-action image data in which a microcalcification lesion having a size approximately equal to the pixel size obtained by mammography or the like is captured, when there is a pixel whose data value is significantly different from the surrounding pixels, Although it is difficult to distinguish whether an actual microcalcification lesion existing in the lesion is taken or whether an abnormal value is output, according to the defective pixel determination system 30 of the present invention, the radiation image detection The radiation image detector 1 is uniformly irradiated with radiation to the device 1 in the state where no subject exists, and the actual image data F (x, y) is output from each radiation detection element (x, y). .

  In such a situation, if there is no abnormal pixel or defective pixel, the actual image data F (x, y) having substantially the same signal value should be output from each radiation detection element (x, y). In such a situation, if there is a pixel whose data value is significantly different from the surrounding pixels, it can be accurately determined that it is an abnormal pixel.

  Therefore, according to the defective pixel determination system 30 of the present invention, it is possible to accurately detect an abnormal pixel that outputs an abnormal value in a situation where radiation is irradiated. In addition, during actual mammography, it is possible to accurately prevent an erroneous determination of an abnormal value or a misdiagnosis of an abnormal value as a microcalcification lesion even though a microcalcification lesion is actually captured. Is possible.

  As described above, according to the defective pixel determination system 30 of the present invention, it is possible to accurately determine whether or not the radiation detection element (x, y) is a defective pixel. It is possible to accurately detect a microcalcification lesion in imaging, and mammography can be accurately performed using the radiation image detector 1.

  On the other hand, in the storage means such as the ROM or hard disk of the console 31, the above-described actual image data acquisition step (step S2 in FIG. 27), dark reading step (step S3), temperature compensation variable calculation step (step S4), time An image generation program for realizing a step of calculating a mean value or the like (step S5), an offset correction value calculation step (step S6), and an image correction step (step S7) is stored, and the console 31 stores the image generation program. Are expanded in the work area of the RAM, necessary processing is executed according to the image generation program, and each part of the apparatus is controlled.

  Hereinafter, a case where the system functions as a radiation image generation system will be described. When the console 31 receives a radiographic image capturing instruction from a radiographer, doctor, or the like, the actual image data acquisition step is performed by driving the radiation generating device 34 that controls the imaging conditions (tube current, tube voltage, etc.) and the start of irradiation. (Step S2 in FIG. 27) is executed. Further, before or after radiographic imaging, an instruction signal is transmitted to the radiographic image detector 1 so as to perform at least one dark reading, and a dark reading step (step S3) is executed.

  The radiation image detector 1 has its own ID, the actual image data F (x, y) of the radiographic image acquired in the actual image data acquisition step (step S2) and the dark reading step for each radiation detection element (x, y). The dark read value D (x, y) acquired in (Step S3) is transmitted to the console 31 by a wired method or a wireless method. When the real image data F (x, y) and the dark read value D (x, y) are transmitted from the radiation image detector 1, the console 31 stores them in its storage means.

  Then, thinned data compressed to 1/15 or the like is first generated and displayed from the acquired real image F (x, y), so that the necessity of re-photographing can be quickly determined. When the radiation image detector generates and transmits thinned data, the thinned data is displayed.

  At this stage, if a positioning defect or the like is recognized, re-photographing is executed, and the process of waiting for re-photographing data acquisition is not performed without performing the offset correction value generation process or the determination process for the presence or absence of defects or abnormal pixels.

In addition, when displaying thinned data, whether or not each pixel output is within a predetermined range together with positioning,
In other words, in order to know whether or not the output value is proportional to the X-ray dose that has passed through the subject rather than being in a saturated state, an image that has undergone offset / gain correction processing may be displayed for the thinned data.

  Further, the console 31 executes the above-described abnormal pixel determination method (see [Determination Method 7] to [Determination Method 11]) according to the abnormal pixel determination program stored in the storage unit such as a ROM, and the radiation image detector. Whether each radiation detection element (x, y) is an abnormal pixel is determined based on the actual image data F (x, y) and the dark read value D (x, y) transmitted from 1. In the case of executing [Determination Method 11], the radiation image detector 1 is irradiated with radiation from the radiation source 34a of the radiation generator 34 in a state where no subject exists.

  If there is a radiation detection element (x, y) that can be determined as an abnormal pixel, the console 31 temporarily stores the number (x, y) in the storage unit.

  Subsequently, the console 31 executes a temperature compensation variable calculation step (step S4). The console 31 refers to the ID of the radiation image detector 1, reads out information of a plurality of radiation detection elements (x ′, y ′) associated with each radiation detection element (x, y ′) from the storage means, and outputs from the information Using each of the dark read values D (x ′, y ′) thus obtained, a temperature compensation variable W (x, y) is calculated for each radiation detection element (x, y) according to the above equation (3).

  At that time, when calculating the temperature compensation variable W (x, y) for one radiation detection element (x, y), the console 31 associates with another radiation detection element (x, y) in advance. When a defective pixel is included in the element, the dark read value D (xs, ys) output from the defective pixel using any one of [Calculation method 1] to [Calculation method 4] described above is used. The temperature compensation variable W (x, y) of the one radiation detection element (x, y) is calculated by replacing or interpolating with the dark reading value output from the radiation detection element in the vicinity thereof.

In addition, the console 31 executes a temporal average value calculation step (step S5) based on data obtained at the time of past calibration or the like in advance, but in this embodiment, the radiographic image detector 1 is in the present embodiment. The temporal average value δ (x, y) and the temporal average value ω (x, y) of the spatial average value w _m (x, y) as the temperature compensation variable are previously stored in the server means 39 as described above. It is calculated in advance and stored in the storage means 38. The console 31 may be configured to calculate these values by itself at this stage.

Subsequently, the console 31 executes an offset correction value calculation step (step S6). At that time, the console 31 transmits the ID of the radiation image detector 1 transmitted via the tag reader 37 to the server means 39 via the network NW. Based on the ID of the radiation image detector 1, the server unit 39 determines the temporal average value δ (x, y) of the radiation image detector 1 or the spatial average value w _m (x, y) as a temperature compensation variable. Is read from the storage means 38 and transmitted to the console 31.

The console 31 calculates the temporal average value δ (x, y) of the dark read value d _m (x, y) obtained from the server means 39 for each radiation detection element (x, y) of the radiation image detector 1. From the temporal average value ω (x, y) of the spatial average value w _m (x ′, y ′) as the temperature compensation variable, the difference ε (x, y) (= δ (x, y)) -Ω (x, y)) is calculated. The calculated ε (x, y) and each temperature compensation variable W (x, y) for each radiation detection element (x, y) of the radiation image detector 1 calculated in the temperature compensation variable calculation step (step S4). y) is added to calculate an offset correction value O (x, y) for each radiation detection element (x, y).

Subsequently, the console 31 executes an image correction step (step S7), and each captured image data F (x, y) acquired from each radiation detection element (x, y) in the captured image data acquisition step (step S2). From each offset correction value O (x, y) calculated for each radiation detection element (x, y) in the offset correction value calculation step (step S6), each radiation detection element (x, y) is subtracted from the difference. Is multiplied by a gain correction value G (x, y) that is calculated in advance to generate final image data F _O (x, y) for each radiation detection element (x, y).

By configuring the radiation image generation system in this way, the same effect as that of the image generation method using the above-described radiation image detector is effectively exhibited, and each radiation detection element (x, y) is effective. An offset correction value O (x, y) is acquired, and there is a characteristic variation for each pixel (radiation detection element 14) with respect to the captured image data F (x, y) generated in the captured image data acquisition step (step S2). The final image data F _O (x, y) is generated after correction.

  At the same time, it is only necessary to perform the dark reading at least once before or after the radiographic imaging, and the actual image data F (x, y) of the radiographic image and the dark reading value D (x, y) for at least one time are obtained. Since only the transmission from the radiation image detector 1 is required, it is possible to reduce power consumption required for radiographic image capturing.

  In particular, when the radiation image detector 1 is a battery built-in type, the dark reading needs to be performed at least once and the dark reading value D (x, y) needs to be transmitted at least once. The power consumption required for reading and transmitting the dark read value D (x, y) is reduced, and it is possible to prevent the internal battery 21 from being consumed. Further, even when the radiation image detector 1 is used in a wired manner, the number of times of dark reading and the number of transmissions of the dark reading value D (x, y) can be reduced, so that power consumption can be suppressed.

  Also, by performing defective pixel determination using the above-described defective pixel determination method according to the present embodiment, it is accurately determined whether or not the radiation detection element is a defective pixel, and it is determined and registered as a defective pixel. It is possible to perform imaging using only normal elements without using the radiation detection elements that are used, and it is possible to maintain diagnostic accuracy when the lesion size such as microcalcification is approximately equal to the pixel size .

  In this embodiment, the method has been described as a particularly effective method when the dark reading is performed once before or after the radiographic image capturing. However, the same effect can be obtained even when the dark reading is performed twice or more. be able to. That is, when the number of dark readings performed at the time of calibration is M times and the number of dark readings performed before or after radiographic imaging is K times, if the relationship of M> K is established, the present invention has been described. Needless to say, similar effects can be obtained.

を本発明の実施例のＤ（ｘ，ｙ）にあてはめて考えれば良い。すなわち、
Ｄ（ｘ，ｙ）=Ｄｋ_ave（ｘ，ｙ） …（５６）
として考えれば良い。 That is, when K> 2, the dark reading value is set as Dk (x, y) (k = 1 to K, K> 2),

May be applied to D (x, y) in the embodiment of the present invention. That is,
D (x, y) = Dk _ave (x, y) (56)
Think of it as

Further, even when there is a defective pixel in the radiation detection element (x, y) of the radiation image detector 1, output from the defective pixel by using the above-described spatial statistical value (spatial average value) calculation method. The obtained dark reading value D is appropriately replaced or interpolated to obtain temperature compensation variables W (x, y) and w _m (x ′, y ′) which are effective spatial statistics (spatial average values). It is possible to calculate. Thereby, the offset correction value O (x, y) can be accurately calculated.

  Needless to say, the present invention is not limited to the above-described embodiment, and can be changed as appropriate. The concept of the present invention is to select in advance a radiation detection element (x ′, y ′) that varies in temperature in the same manner as one radiation detection element and associate it with the one radiation detection element (x, y) in advance. . Therefore, the radiation detection element itself does not need to be a radiation detection element type as in the above-described embodiment, and is also applied to a radiation image detector including a sensor panel unit in which elements having other structures are two-dimensionally arranged. It goes without saying that it is possible.

  Further, in the present invention, the temperature variation in the pixel unit of the radiation detection element has been described, but even if it is a factor other than the temperature variation, a certain pixel value causes a similar change to a neighboring pixel value. Needless to say, the same effect can be obtained by applying the same method if the fluctuation is small.

Further, in the present embodiment, the case where the data processing is performed by the console 31 has been described. However, the offset image O (x, y) is subtracted for each pixel from the actual image data F (x, y), and the actual image is obtained. The radiation image detector 1 may be configured to perform all data processing from the correction of the data F (x, y) to the generation of the final image data F _O (x, y).

  Further, for example, in the above embodiment, the dark reading is performed at the time of calibration performed separately from the radiographic image capturing. However, the dark reading value to be performed before or after the radiographic image capturing is stored and stored. You may comprise so that it may be used. In this case, for example, the dark reading values for a plurality of consecutive radiographic image captures are stored by normalization and the like, and the plurality of dark reading values are used as in the above embodiment. Alternatively, a temporal statistical value (temporal average value) δ (x, y) or ω (x, y) may be calculated.

  Further, in this embodiment, the offset correction is taken as an example, and the description has been made. However, even when the gain correction value G (x, y) is obtained, the radiation image detector is subjected to uniform radiation under predetermined conditions. The offset correction is also applied to this read value (corresponding to the actual image data F (x, y) in this embodiment). For this reason, in the present embodiment, the case where offset correction is performed on the real image data F (x, y) has been described as an example. However, when the gain correction value O (x, y) is obtained for the above-described reason. Needless to say, the same treatment can be performed and the same effect can be obtained.

It is a figure which shows the external appearance structure of the radiographic image detector which concerns on this embodiment. It is an equivalent circuit diagram shown in the structure of the sensor panel part and reading part of a radiographic image detector. It is a figure explaining the number allocated to the radiation detection element. It is a figure explaining the dark reading value output from a radiation detection element for every multiple times of dark reading. It is a graph explaining distribution of temporal fluctuation of a plurality of dark reading values outputted from a radiation detection element by a plurality of dark readings. It is a figure which shows an example of how to take the some radiation detection element matched with a radiation detection element. It is a graph explaining that the average value and the standard deviation differ in the distribution of temporal fluctuation of the dark reading value when the radiation detection elements are different. It is a graph explaining that it becomes distribution with a small standard deviation of distribution of spatial average value W (x, y). It is a graph which shows the distribution of the temperature compensation variable which becomes broad when the shift degree by the temperature change of the temporal fluctuation distribution of D (x, y) and the temporal fluctuation distribution of D (x ', y') is different. It is a graph explaining that it is difficult to estimate the true value of the offset correction value from the radiation detection element in one dark reading. It is a figure explaining the dark reading value and spatial average value which are output from a radiation detection element for every multiple times of dark reading. It is a graph explaining that the dark reading value and spatial average value output for every multiple times of dark reading are distributed in a normal distribution. It is a graph explaining that the temperature of a radiation detection element fluctuates temporally. It is a graph explaining the distribution of the temporal fluctuation estimated about the dark reading value output from a radiation detection element and the spatial average value by one dark reading performed before or after radiographic imaging. If temperature fluctuation occurs during a plurality of times of dark reading, the distribution of dark reading values and the like becomes broad, but the distribution of temperature-corrected dark reading values does not become broad. (A)-(C) are figures which show the example of the area | region set so that one radiation detection element may be located in the position which is not the center of a square area | region. It is a figure explaining the read-out IC connected one for every predetermined number of radiation detection elements. It is a figure explaining the case where the electrical signal from each radiation detection element which belongs to a field is read by two adjacent read-out ICs separately. (A) and (B), when a plurality of radiation detection elements are selected in a normal manner, (C) and (D) treat the vicinity of the boundary of adjacent readout ICs in the same manner as the peripheral part of the sensor panel unit. It is a figure explaining a case. It is a graph showing the average distribution and the distribution which is much larger than the standard distribution of the fluctuation distribution of the dark reading value from the radiation detection element. It is a graph showing the case where the average value of the fluctuation distribution of the dark reading value from a radiation detection element is larger than a threshold value. It is a graph in which the standard deviation of the difference fluctuation distribution represents an average distribution. 23 is a graph showing a distribution in which the standard deviation of the fluctuation distribution of the difference is larger than that shown in FIG. It is a figure explaining an example in case a pixel of interest is a radiation detection element (4, 4) and a radiation detection element (6, 6) is a defective pixel among 7 × 7 radiation detection elements associated therewith. . It is a figure which shows the example which substitutes a defective pixel with the radiation detection element which exists in the position outside an area | region in the example of FIG. FIG. 25 is a diagram illustrating an example in which defective pixels are interpolated by radiation detecting elements existing at positions outside the region when the radiation detecting elements (7, 6) are defective pixels in the example of FIG. It is a flowchart which shows the process sequence of the image generation using a radiographic image detector. It is a figure which shows the structure of the radiographic image generation system containing the defect image determination system which concerns on this embodiment.

Explanation of symbols

1 Radiation image detector 3 Antenna device (communication means)
4, 5, 6 Sensor panel section, reading section 5, control means 6 (image data acquisition means)
13 terminals (communication means)
14, (x, y) radiation detection element 30 defective pixel determination system 31 console (defective pixel determination device)
34 Radiation generator 38 Storage means 39 Server means (defective pixel determination device)
A (x, y) Average value of actual image data (spatial average value of actual image data)
F (x, y) Real image data V (x, y) Difference V′th threshold value (xs, ys) Defective pixel

Claims (4)

A plurality of radiation detection elements arranged two-dimensionally;
Image data acquisition means for acquiring real image data based on radiation irradiation from the plurality of radiation detection elements;
A communication means for transmitting the photographed image data;
A radiation image detector comprising:
A radiation generator for irradiating the radiation image detector with radiation;
When a signal instructing the start of determination is input, radiation is emitted from the radiation generation device to the radiation image detector in a state where no subject is present, and is output from each radiation detection element of the radiation image detector. A defective pixel determination device that transmits each of the actual image data, and determines whether or not each of the radiation detection elements of the radiological image detector is a defective pixel based on each of the actual image data;
With
In the defective pixel determination device, the number of times that the actual image data output from the radiation detection element of the radiation image detector becomes an abnormal value exceeding a preset threshold is equal to or greater than the preset number. In this case, a defective pixel determination system, wherein the radiation detection element is determined as a defective pixel.   The defective pixel determination device outputs the actual image data output from the radiation detection element of one of the radiation image detectors and the radiation detection elements output from a plurality of radiation detection elements previously associated with the one radiation detection element. Comparing the absolute value of the difference with the spatial average value of each live-action image data and a predetermined threshold, and when the absolute value of the difference is greater than the threshold, the real-image data is the abnormal value The defective pixel determination system according to claim 1, wherein the defective pixel determination system according to claim 1 is determined.   3. The defective pixel according to claim 1, wherein the defective pixel determination device registers information on the radiation detection element determined to be a defective pixel in a defective pixel map for a state where the subject does not exist. Judgment system.   The said defective pixel determination apparatus is provided with the memory | storage means which memorize | stores the frequency | count that the said radiation detection element of the said radiographic image detector output the said real image data of the abnormal value for every said radiation detection element. The defective pixel determination system according to any one of claims 1 to 3.

Images