DE102010019735A1 - Automatically defect pixels detecting method, for flat panel detector in e.g. digital mammography system, involves recognizing defect pixels depending on appropriate pixel value of mean value and middle deviation of all pixel values - Google Patents

Abstract

The method involves capturing multiple images (1) e.g. dark images, by a detector and generating a noise image the images. An average value of all pixel values of the noise image is determined. An average deviation of all pixel values of the noise image is determined from the average value. The pixel of the detector is recognized as defect pixels depending on an appropriate pixel value within the noise image, of the mean value of all the values and middle deviation of the pixel values. The pixel of the detector is arranged in middle of a predetermined section. Independent claims are also included for the following: (1) an imaging fluoroscopy system comprising a control unit (2) a computer program having a set of instructions for automatically detecting defect pixels in a detector (3) an electronic readable data carrier comprising a set of instructions for automatically detecting defect pixels in the detector.

Description

The present invention relates to a plurality of methods and imaging radiographic systems, such as X-ray systems, with which automatically detected defect pixels of a detector of the imaging fluoroscopy system.

Detectors (eg, flat-panel detectors) used in medical imaging with X-ray systems (eg, stationary or mobile C-arms or digital mammography systems) are never free of so-called defect pixels due to their manufacturing process and for reasons of economy. A defect pixel is understood to mean that pixel within an image matrix which differs significantly in its behavior from a normal pixel and thus leads to artifacts in a result (image).

The defect pixels are subdivided into the following four defect pixel types:

1. dark current defect pixel

Dark-current defect pixels always provide too high or too low a dark current value without the radiation of a radiation generator compared to a correctly operating pixel.

2. Sensitivity defect pixel

Sensitivity defect pixels provide too high or too low a gray level compared to a correctly operating pixel when exposed to the radiation of a radiation generator. It should be noted that pixels with a conspicuous dark current behavior (defect pixel type 1), usually also have a noticeable sensitivity behavior (defect pixel type 2).

3rd turn signal

Turn signals vary widely in their characteristics, so that their regular or irregular behavior is purely statistical.

4. Saturation defect pixel

Saturation defect pixels "freeze" already at a radiation dose far below the actual saturation limit and thus have a lower dynamic range compared to a correctly working pixel.

In the prior art, each detector is subjected to defect pixel inspection prior to its delivery. In the process, the pixels are examined for abnormalities by means of light images at different radiation dosages. For each detector and for each detector mode specific criteria and parameters are set, which must be checked for the decision whether the corresponding pixel is defective or not defective. For the check, only the gray value or pixel value of the pixel to be checked is used as a criterion. The defect pixels detected during this inspection are stored on a data carrier as a so-called defect pixel map and serve as a mask for a later correction.

If further defects occur after delivery, which is not unusual in general, they are manually entered into the defect pixel map as soon as they have been detected. In addition, a complete and time-consuming search for defect pixels can be performed in a similar manner as before shipment.

The object of the present invention is to detect defect pixels automatically even after delivery in a simple manner.

According to the invention, this object is achieved by a method for automatic detection of defect pixels of a detector of an imaging fluoroscopy system according to claim 1 or claim 3, by an imaging fluoroscopy system according to claim 13 or 14, by a computer program product according to claim 17 or by an electronically readable data carrier according to claim 18 , The dependent claims define preferred and advantageous embodiments of the present invention.

• • Erfassen mehrerer Hellbilder oder mehrerer Dunkelbilder mit Hilfe des Detektors. D. h. gemäß diesem Verfahren werden entweder nur mehrere Hellbilder oder nur mehrere Dunkelbilder erfasst.
• • Erzeugen eines Rauschbildes, welches aus den vorab erfassten Bildern erzeugt wird. Dabei wird jedes Pixel (z. B. das Pixel in der oberen linken Ecke) des Rauschbildes abhängig von den entsprechenden Pixeln der mehreren vorab erfassten Bilder (den Pixeln in der oberen linken Ecke der mehreren Bilder) bestimmt, wobei jedes Pixel des Rauschbildes ein Maß für das Rauschen innerhalb der Pixelwerte der entsprechenden Pixel der mehreren vorab erfassten Bilder darstellt.
• • Bestimmen eines Mittelwerts aller Pixelwerte des Rauschbildes. Der Mittelwert kann dabei der arithmetische oder geometrische Mittelwert, aber auch der Modus, Median oder ein anderer Mittelwert sein.
• • Bestimmen einer Maßzahl, welche angibt, wie stark die Pixelwerte des Rauschbildes von dem vorbestimmten Mittelwert abweichen.
• • In Abhängigkeit von einem Pixelwert des Rauschbildes eines bestimmten Pixels, in Abhängigkeit von dem Mittelwert aller Pixelwerte des Rauschbildes und in Abhängigkeit von der Maßzahl wird der bestimmte Pixel als Defektpixel erkannt.

In the context of the present invention, a method for the automatic detection of defective pixels of a detector of an imaging fluoroscopy system is provided. The method according to the invention comprises the following steps:

• • Capture multiple light images or multiple dark images using the detector. Ie. According to this method, either only a plurality of light images or only a plurality of dark images are detected.
• • Generating a noise image, which is generated from the previously acquired images. At this time, each pixel (eg, the pixel in the upper left corner) of the noise image is determined depending on the corresponding pixels of the plurality of pre-acquired images (the pixels in the upper left corner of the plurality of images), each pixel of the noise image being a measure for the noise within the pixel values of the corresponding pixels of the plurality of prerecorded images.
• Determining an average of all pixel values of the noise image. The mean value may be the arithmetic or geometric mean value, but also the mode, median or another mean value.
• • Determining a measure indicating how much the pixel values of the noise image deviate from the predetermined average.
• Depending on a pixel value of the noise image of a particular pixel, as a function of the mean value of all pixel values of the noise image and depending on the dimension, the particular pixel is recognized as a defective pixel.

Since light images and dark images are generated anyway in the regular maintenance intervals, it is possible to search for newly added defect pixels without further effort in accordance with the method according to the invention, which is also known as "smart repair". Since the dark current in most detectors is constantly updated in the background, you can also search for new defect pixels. Thus, an automatic detection of defect pixels during operation by means of dark images is possible, which is also known as "on the fly detection".

According to a preferred embodiment of the invention, an i-th pixel of the detector is considered to be a defective pixel if one of the following two inequalities (1) and (2) in the noise image for the pixel value PVNI _i corresponding to the ith pixel ) is satisfied: PVNI _i > NM + F × NS (1), PVNI _i <NM - F × NS (2).

Here, NM ("Noise Mean") stands for the average of all pixel values of the noise image, F for a factor and NS ("Noise Sigma") for the standard deviation of all pixel values of the noise image.

Since the embodiment described with the inequalities (1) and (2) can be carried out both with dark images and with light images, this embodiment comprises quasi two variants of the method according to the invention, one variant working with light images and the other variant with dark images.

• • Erfassen mehrerer Hellbilder oder mehrerer Dunkelbilder mit Hilfe des Detektors. D. h. gemäß diesem weiteren Verfahren werden entweder nur mehrere Hellbilder oder nur mehrere Dunkelbilder erfasst.
• • Erzeugen eines Signalbildes aus den mehreren vorab erfassten Bildern. Dabei wird jedes Pixel dieses Signalbildes berechnet, indem für das jeweilige Pixel ein Pixelwert bestimmt wird, welcher einem Mittelwert der diesem Pixel entsprechenden Pixelwerte der vorab erfassten mehreren Bildern entspricht. Der Mittelwert kann dabei der arithmetische oder geometrische Mittelwert, aber auch der Modus, Median oder ein anderer Mittelwert sein.
• • Bestimmen eines Mittelwerts aller Pixelwerte des Signalbildes. Der Mittelwert kann dabei der arithmetische oder geometrische Mittelwert, aber auch der Modus, Median oder ein anderer Mittelwert sein.
• • Erzeugen eines Rauschbildes, welches aus den vorab erfassten Bildern erzeugt wird. Dabei wird jedes Pixel (z. B. das Pixel in der oberen linken Ecke) des Rauschbildes abhängig von den entsprechenden Pixeln der mehreren vorab erfassten Bilder (den Pixeln in der oberen linken Ecke der mehreren Bilder) bestimmt, wobei jedes Pixel des Rauschbildes ein Maß für das Rauschen innerhalb der Pixelwerte der entsprechenden Pixel der mehreren vorab erfassten Bilder darstellt.
• • Bestimmen einer Maßzahl, welche angibt, wie stark die Pixelwerte des Rauschbildes von einem Mittelwert aller Pixelwerte des Rauschbildes abweichen. Der Mittelwert kann dabei der arithmetische oder geometrische Mittelwert, aber auch der Modus, Median oder ein anderer Mittelwert sein.
• • In Abhängigkeit von einem Pixelwert des Signalbildes eines bestimmten Pixels, in Abhängigkeit von dem Mittelwert aller Pixelwerte des Signalbildes und in Abhängigkeit von der Maßzahl wird der bestimmte Pixel als Defektpixel erkannt.

In the context of the present invention, a further method for the automatic detection of defect pixels of a detector of an imaging fluoroscopy system is provided. The further method according to the invention comprises the following steps:

• • Capture multiple light images or multiple dark images using the detector. Ie. According to this further method, either only a plurality of light images or only a plurality of dark images are detected.
• • Generating a signal image from the several pre-acquired images. In this case, each pixel of this signal image is calculated by determining, for the respective pixel, a pixel value which corresponds to an average value of the pixel values of the previously acquired multiple images corresponding to this pixel. The mean value may be the arithmetic or geometric mean value, but also the mode, median or another mean value.
• • Determining an average of all pixel values of the signal image. The mean value may be the arithmetic or geometric mean value, but also the mode, median or another mean value.
• • Generating a noise image, which is generated from the previously acquired images. At this time, each pixel (eg, the pixel in the upper left corner) of the noise image is determined depending on the corresponding pixels of the plurality of pre-acquired images (the pixels in the upper left corner of the plurality of images), each pixel of the noise image being a measure for the noise within the pixel values of the corresponding pixels of the plurality of prerecorded images.
• • Determine a measure that indicates how much the pixel values of the noise image differ from an average of all pixel values of the noise image. The mean value may be the arithmetic or geometric mean value, but also the mode, median or another mean value.
• Depending on a pixel value of the signal image of a specific pixel, as a function of the mean value of all pixel values of the signal image and depending on the dimension, the specific pixel is recognized as a defective pixel.

The advantages of the further method according to the invention essentially correspond to the advantages of the method according to the invention, so that it is dispensed with a repetition.

According to a preferred embodiment of the invention, an i-th pixel of the detector is considered to be a defective pixel if one of the following inequalities (3) and (4) in the signal image corresponds to the pixel value PVSI _i ("pixel value signal image") corresponding to the ith pixel ) is satisfied: PVSI _i > SM + F × NS (3), PVSI _i <SM - F × NS (4).

Here, SM ("Signal Mean") stands for the mean value of all pixel values of the signal image, F for a factor and NS for the standard deviation of all pixel values of the noise image.

Since the embodiment described with the inequalities (3) and (4) can be performed both with dark images and with light images, this embodiment comprises quasi two variants of the method according to the invention, one variant working with light images and the other variant with dark images.

The noise image can be determined according to the present invention by calculating the standard deviation of the corresponding pixel values of the plurality of prerecorded images according to the following equation (5) for each pixel value PVNI _{i of} the noise image (the index i passes through all the pixels of the noise image).

entspricht dem Pixelwert des i-ten Pixels des j-ten Bildes und

wird gemäß der folgenden Gleichung (6) berechnet.Where N is the number of multiple pre-acquired images,

corresponds to the pixel value of the ith pixel of the jth image and

is calculated according to the following equation (6).

dem Mittelwert der Pixelwerte des i-ten Pixels aller N Bilder. Damit stellt Gleichung (6) auch die Möglichkeit dar, den Pixelwert des i-ten Pixels des Signalbildes zu berechnen. Anders ausgedrückt gilt PVSI_i

In other words, it corresponds

the average of the pixel values of the i-th pixel of all N pictures. Thus, equation (6) also provides the ability to compute the pixel value of the ith pixel of the signal image. In other words, PVSI _i

The measure indicating how much the pixel values of the noise image deviate from an average of all pixel values of the noise image may correspond to the standard deviation NS which is calculated according to the following equation (7):

Where M is the number of pixels in the noise image, PVNI _k is the pixel value of the k-th pixel of the noise image and PVNI is calculated according to the following equation (8).

In other words, it corresponds PVNI the average of the pixel values of all M pixels of the noise image.

The noise image does not have to comprise the same number of pixels as the pre-acquired light images or dark images, but the noise image can only correspond to a predetermined region within the light or dark images and thus have fewer pixels than either of the light images or each of the dark images. To generate the noise image, the same section (the same pixels) is used for each of the images.

Similarly, the signal image may also correspond only to a predetermined section within the images (light images or dark images). In this case, the same section (s) of the images are also evaluated for generating the signal image for each of the images.

According to a preferred embodiment of the invention, a noise image (hereinafter also referred to as overall noise image) and / or a signal image (also referred to below as the overall signal image) is generated which has the same number of pixels as the dark images or light images. Thus, this overall noise image or overall signal image also has the same resolution (number of pixels in the x-direction and y-direction) as the original image of the detector in which defect pixels are searched for.

In order to determine the statistical values (NM, NS, SM), it is then possible, in accordance with this embodiment, to work with a section of the overall noise image or overall signal image, so that the statistical values (NM, NS, SM) are determined only for the corresponding section. In this case, this section of the overall noise image or overall signal image can be regarded as a noise image or signal image in the sense of the present invention.

• 1. Möglichkeit: Es wird zuerst das Gesamtrauschbild bzw. Gesamtsignalbild erzeugt. Ausgehend davon wird dann das Rauschbild bzw. Signalbild erzeugt, indem der gewünschte Ausschnitt aus dem Gesamtrauschbild bzw. Gesamtsignalbild gewählt wird.
• 2. Möglichkeit: Aus den entsprechenden Ausschnitten der Hellbilder oder Dunkelbilder wird das Rauschbild bzw. Signalbild mit dem entsprechenden Ausschnitt erzeugt.

In other words, there are two possibilities for generating a noise image or signal image which represents the same section of the light images or dark images.

• 1st possibility: First, the overall noise pattern or overall signal image is generated. Based on this, the noise image or signal image is then generated by selecting the desired section from the overall noise image or overall signal image.
• 2nd possibility: From the corresponding sections of the light images or dark images, the noise pattern or signal image is generated with the corresponding section.

Of course, the noise image or signal image may also correspond to the overall noise image or overall signal image, if the corresponding section (which in this case is actually no true "section") contains all pixels of each of the light images or dark images.

While the pixel values within a dark image hardly differ in their amplitude, as far as they are correctly working pixels (and not defect pixels), the pixel values within the light images have large amplitude differences even with correctly operating pixels of the detector. In general, pixel values for pixels in the middle of a light image are larger than pixel values for edge pixels of a light image.

Therefore, a predetermined section within the light images, from which the statistic values of the signal image are generated, should not be too large. According to an embodiment of the invention, the predetermined cutout has an edge length of 8 to 15 pixels, so that the number of pixels within the cutout 64 is (8 x 8) to 225 (15 x 15), with a cutout of 10 x 10 pixels being preferred becomes.

Advantageously, the pixel of the detector, which is to be examined for its correct operation, lies in the middle of the predetermined section of the light images, from which the signal image or the section of the overall signal image is generated. In other words, in this embodiment, a signal image is generated for each pixel of the detector to be examined. If the pixel to be examined is located at the edge of the detector, the part of the predetermined section of the light images, which is arranged outside the light images, is ignored for generating the signal image (section of the total signal image) and for averaging. In other words, the averaging is performed only on the basis of the pixels of the predetermined section, which belong to the light images.

In particular, in embodiments according to the invention which work with light images, the section within the light images with which the signal image or the section of the overall signal image is generated need not correspond to the section within the light images with which the noise image is generated becomes. In a preferred embodiment according to the invention, only one noise image is generated for checking the pixels of the detector, but one signal image or section of the overall signal image is generated for each pixel to be checked. In this embodiment, the standard deviation NS resulting from the noise image is constant, while the mean value SM resulting from the respective signal image or section of the overall signal image changes depending on the pixel to be checked and thus dependent on the signal image or section of the overall signal image generated for this pixel.

According to the invention, several of the methods listed above can also be combined with one another in order to automatically detect defect pixels. In this case, a pixel of the detector is then considered to be a defective pixel if this pixel is recognized by at least one of the previously listed methods as a defective pixel. In such a combined method then also light images and dark images can be detected.

Advantageously, in several or all of the methods of the present invention, the factor may have the same value which ranges from 2 to 4, especially 3.

The use of the same factor in any or all of the methods of the present invention independently of the detector under test has the distinct advantage that no detector mode or detector specific thresholds or parameters need to be established, as is the case in the prior art.

In the context of the present invention, an imaging fluoroscopy system is also provided. The fluoroscopy system, in particular an X-ray system, comprises a detector, a radiation generator and control means. By means of the control means and the detector, the fluoroscopy system captures a plurality of dark images or a plurality of light images. Depending on the captured images, the control means generates a noise image and from this noise image determines an average of all pixel values of that noise image. Subsequently, the control means determines a degree for deviation of the pixels of the noise image from this average value. Depending on the pixel value of the noise image corresponding to a particular pixel, the average of all pixel values of the noise image, and the degree of deviation, the control means recognizes whether or not the particular pixel of the detector is a defect pixel.

In the context of the present invention, a further imaging fluoroscopy system is also provided. The further fluoroscopy system comprises a detector, a radiation generator and control means. By means of the control means and the detector, the further fluoroscopy system captures a plurality of images (light images or dark images). Depending on these images, the control means generates a signal image by the control means calculating each pixel of the signal image, the control means for each pixel determining a pixel value corresponding to an average of the pixel values of the plurality of images corresponding to that pixel. Subsequently, the control means determines an average of all pixel values of the signal image. In addition, the control means generates a noise image depending on the previously acquired images. From this noise image, the control means determines a degree of average deviation of the pixel values of the noise image from an average of the pixel values of the noise image. Depending on the pixel value of the signal image corresponding to a particular pixel, the average of all pixel values of the signal image, and the degree of deviation of the pixel values of the noise image, the control means detects whether or not the particular pixel is a defective pixel.

The advantages of the imaging radiographic systems according to the present invention essentially correspond to the advantages of the methods according to the invention, which have been carried out in detail in advance, so that a repetition is omitted here.

Furthermore, the present invention describes a computer program product, in particular a computer program or software, which can be loaded into a memory of a programmable controller or a computing unit of a fluoroscopy system. With this computer program product, all or various previously described embodiments of the inventive methods can be executed when the computer program product is running in the controller. The computer program product may require program resources, eg. As libraries and auxiliary functions to realize the corresponding embodiments of the method. In other words, with the claim directed to the computer program product, in particular a computer program or a software is to be protected, with which one of the above-described embodiments of the method according to the invention can be executed or which executes this embodiment. The software can be a source code (eg C ++), which can still be compiled and bound or which can only be interpreted must be, or an executable software code that is to be loaded for execution only in the appropriate processing unit.

Finally, the present invention discloses an electronically readable medium, for. As a DVD, a magnetic tape or a USB stick on which electronically readable control information, in particular software (see above), is stored. When this control information (software) is read from the data medium and stored in a control unit of a fluoroscopy system, all the embodiments according to the invention of the previously described methods can be carried out.

In particular, according to the present invention, whether or not a particular pixel is a defective pixel is determined via a reference point and the multiple of a standard deviation. The multiple (the factor F) of the standard deviation determines with which probability a pixel which works correctly is nevertheless classified as defective. At a factor of 3, this probability is 0.1%.

• 1. Bei einer Suche nach Defektpixeln in den Dunkelbildern wird die Standardabweichung als gemittelte zeitliche Standardabweichung über alle Pixel, d. h. als mittleres Elektronikrauschen, bestimmt. Der Bezugspunkt ist dabei der mittlere Grauwert im gesamten Dunkelbild, d. h. der Mittelwert aller Pixelwerte aller Dunkelbilder.
• 2. Gemäß dieser Möglichkeit wird das Elektronikrauschbild, welches über die zeitliche Standardabweichung aus einer Reihe von Dunkelbildern erzeugt wird, untersucht. Dabei ist die Standardabweichung die Variation des mittleren Elektronikrauschens, wobei der Bezugspunkt der Mittelwert im Elektronikrauschbild ist und damit dem mittleren Elektronikrauschen entspricht.
• 3. Bei einer Suche nach Defektpixeln in Hellbildern wird die Standardabweichung als gemittelte zeitliche Standardabweichung über alle Pixel, d. h. als mittleres Quantenrauschen, bestimmt. Der Bezugspunkt ist der mittlere Grauwert in einer lokalen Umgebung, d. h. in demselben Ausschnitt aller Hellbilder.
• 4. Ähnlich wie das Elektronikrauschbild kann auch das Quantenrauschbild, welches einer zeitlichen Standardabweichung einer Serie von Hellbildern entspricht, ausgewertet werden. Der Bezugspunkt ist dabei der Mittelwert im Quantenrauschbild. Bei dieser Möglichkeit der Suche nach Defektpixeln werden aber maximal diejenigen Defektpixel gefunden, welche ebenfalls über das Elektronikrauschbild gefunden werden.

The present invention discloses the following four ways to detect a defect pixel:

• 1. In a search for defect pixels in the dark images, the standard deviation is determined as the mean time standard deviation across all pixels, ie as mean electronic noise. The reference point is the average gray value in the entire dark image, ie the average value of all pixel values of all the dark images.
• 2. According to this possibility, the electronic noise image, which is generated by the temporal standard deviation from a series of dark images, is examined. The standard deviation is the variation of the average electronic noise, the reference point being the mean value in the electronics noise pattern and thus corresponding to the mean electronic noise.
• 3. In a search for defect pixels in light images, the standard deviation is determined as the mean time standard deviation over all pixels, ie as mean quantum noise. The reference point is the mean gray value in a local environment, ie in the same section of all the light images.
• 4. Like the electronic noise image, the quantum noise image, which corresponds to a time standard deviation of a series of light images, can also be evaluated. The reference point is the mean value in the quantum noise image. In this possibility of finding defect pixels, however, at most those defect pixels are found which are also found via the electronic noise image.

Modern detectors feature correlated double sampling as well as analog dark-current compensation (analog offset compensation) in the read-out process, ensuring a uniform dark current ("offset") across all readout amplifier units. This makes it possible to search for defect pixels also by means of dark images. Not only are the perturbations in the gray value of the pixels with respect to the global environment sought, but also the perturbations in the temporal behavior of the pixel (that is, the electronic noise) are searched for. With this, the defect pixel types 1, 2 and 3 (see above) can be detected.

According to the invention, in a series of light images it is also sought for abnormalities with respect to the gray value (pixel value) with regard to its local environment and abnormalities with respect to its temporal behavior (that is, with respect to its quantum noise). Defective pixel types 2 and 4 are detected.

Regardless of whether looking for defect pixels with dark images or with bright images, the selection criterion of whether a pixel is defective or not defective is in both cases formed by a single threshold (the multiple of the determined standard deviation) which is used for both the dark images for the light images have the same value, so that the present invention (the factor) is independent of the detector mode and the detector to be tested.

In other words, with only a single fixed parameter, i. H. the factor F, a defect pixel map is automatically created for all detectors and their modes, which contains all defects important for the image quality.

In particular, the present invention enables the detection of defect pixels in detectors of an X-ray system. Of course, the present invention is not limited to this preferred application, since the present invention also in any working with a detector fluoroscopy systems, for. B. sonography can be used.

In the following, the present invention will be described in detail with reference to inventive embodiments with reference to the figures.

1 to 3 represent several dark images, from which a signal image and a noise image are generated.

Similarly, the 4 to 6 a plurality of bright images, from which a plurality of signal images and a noise image are generated.

In 7 an inventive fluoroscopy system is shown.

8th to 10 each represent a program flowchart of a method according to the invention.

In 1 are several dark images 1 shown. Starting from these dark pictures 1 becomes a signal image SI _D ; SI ' _D ("Signal Image Dark") is generated as it is in 2 is shown. There are two possibilities. Either the signal image or overall signal image SI ' _{D has} exactly as many pixels as each of the dark images 1 or the signal image SI _D includes fewer pixels than each of the dark images 1 such that the signal image SI _D only has a predetermined section of each of the dark images 1 represents. In the second case, the signal image SI _{D can} also be regarded as a section of the overall signal image SI ' _D.

jedes Pixels des Signalbildes SI_D; SI'_D wird dabei durch die vorab ausgeführte Gleichung (6) aus den Pixelwerten der Dunkelbilder 1 berechnet. PVSI_i entspricht dem i-ten Pixel des Signalbildes SI_D; SI'_D, N entspricht der Anzahl der erfassten Dunkelbilder 1 und PVij entspricht dem Pixelwert des i-ten Pixels des j-ten Dunkelbildes 1 (siehe Gleichung (6)).The pixel value PVSI _i

each pixel of the signal image SI _D ; SI ' _D is calculated from the pixel values of the dark images by the previously executed equation (6) 1 calculated. PVSI _i corresponds to the i-th pixel of the signal image SI _D ; SI ' _D , N corresponds to the number of recorded dark images 1 and PVij corresponds to the pixel value of the i-th pixel of the jth dark image 1 (see equation (6)).

In addition, from the dark images 1 a noise picture NI _D ; NI ' _D ("Noise Image Dark") is generated as it is in 3 is shown. The calculation of the pixel values PVNI _{i of} each pixel of the noise image NI _D ; NI ' _D is performed according to equations (5) and (6) above. Similar to the generation of the signal image SI _D ; SI ' _D exists in the generation of the noise image NI _D ; NI _'D two ways. Either the noise image NI ' _D comprises the same number of pixels as each of the dark images 1 or a smaller number of pixels, so that the noise image NI _D only a certain section of the dark images 1 represents. In the second case, the noise image NI _{D can} also be regarded as a section of the overall noise image NI ' _D.

Equation (7) described above can be used to determine the standard deviation NS of the pixel values of the noise image and, via averaging, the mean value NM of all pixel values of the noise image (overall noise image NI ' _D or detail NI _{D of} the overall noise image). Now, for each pixel value PVNI _{i of} the noise image NI _D ; NI ' _{D be} checked whether one of the inequalities (1) and (2) is valid for this pixel value. If so, the associated pixel is a defect pixel.

In a similar manner, the mean value SM of the pixel values of the signal image SI _D ; Determine SI ' _D. Subsequently, for each pixel value PVSI _{i of} the signal image SI _D ; SI ' _{D be} checked whether one of the inequalities (3) or (4) is valid for this pixel value. If this is the case, the associated pixel is a defect pixel.

In the same way are in 4 several bright pictures 2 shown. Starting from these bright pictures 2 if several signal images SI _B ; SI ' _B ("Signal Image Bright") is generated as it is in 5 is shown. It can 'be produced _B, in which case, starting from this overall signal image SI', also the overall image signal SI _B, the plurality of cutouts (signal patterns) SI _B are generated. Because with the light images the difference between the pixel values of the same light image 2 is ready for correctly operating pixels of the detector is very large here, to complete inspection of the detector on defect pixels several sections or signal images SI _{B are} generated, each of which only a specific (small) excerpt of the light images 2 represent such. B. 10 × 10 pixels.

jedes Pixels des Signalbildes SI_B; SI'_B wird dabei durch die vorab ausgeführte Gleichung (6) aus den Pixelwerten der Hellbilder 2 berechnet. PVSI_i entspricht dem i-ten Pixel des Signalbildes SI_D; SI'_D, N entspricht der Anzahl der erfassten Hellbilder 2 und PVi_j entspricht dem Pixelwert des i-ten Pixels des j-ten Hellbildes 2 (siehe Gleichung (6)).The pixel value PVSI _i

each pixel of the signal image SI _B ; SI ' _B becomes from the pixel values of the bright images by the previously executed equation (6) 2 calculated. PVSI _i corresponds to the i-th pixel of the signal image SI _D ; SI ' _D , N corresponds to the number of recorded light images 2 and PVi _j corresponds to the pixel value of the ith pixel of the jth bright image 2 (see equation (6)).

In addition, out of the chiaroscuro 2 a noise image NI _B ; NI ' _B ("Noise Image Bright") is generated as it is in 6 is shown. The calculation of the pixel values PVNI _{i of} each pixel of the noise image NI _B ; NI ' _B is performed according to equations (5) and (6) above.

By the above-described equation (7), the standard deviation NS of the pixel values of the noise image NI _B ; NI ' _B and averaging the mean NM of all pixel values of the noise image NI _B ; Determine NI _'B. Now, for each pixel value PVNI _{i of} the noise image NI _B ; NI ' _{B are} checked whether one of the inequalities (1) and (2) is valid for this pixel value. If this is the case, the associated pixel is a defect pixel.

In a similar manner, the mean value SM of the pixel values of the signal image SI _B ; Determine SI ' _B. Subsequently, for each pixel value PVSI _{i of} the signal image SI _B ; SI ' _{B are} checked whether one of the inequalities (3) or (4) is valid for this pixel value. If this is the case, the associated pixel is a defect pixel.

In this case, for each pixel of the detector to be checked in such a way from the light images 2 A signal image SI _B generates that the pixel to be checked is arranged in each case in the middle of the signal image, as it 5 can be seen. In contrast, to determine the standard deviation NS required for inequalities (3) and (4), only one noise image NI _{B is} generated for all the pixels to be checked, the dimensions of which are substantially greater than the dimensions of the respective signal images SI _B. The cutout inside the light pictures 2 , with which the noise image NI _{B is} generated, lies advantageously in each case in the middle of the bright images 2 like it 6 can be seen.

In 7 schematically is an inventive X-ray system 20 shown. The X-ray system according to the invention 20 includes a C-arm 11 at which a radiation generator 6 and a detector 4 of the X-ray system 20 are attached. In addition, the X-ray system includes 20 control means 5 which interfaces both with the radiation generator 6 as well as with the detector 4 have, and a terminal 7 with a screen 8th , a keyboard 9 and a mouse 10 , Because the terminal 7 an interface to the tax means 5 an operator can use the terminal 7 and the control means 5 the x-ray system 20 Taxes. About the terminal 7 can also use the software for the tax 5 in the control means 5 getting charged. This software of the control means 5 may also include one of the methods of the invention or both methods of the invention. It is also possible that an inventive or both inventive method is / are included in a software which in the terminal 7 expires. Regardless of the software in which the method (s) according to the invention are / are contained, the software can be stored on a DVD 14 be stored so that this software then from the terminal 7 from the DVD 14 read and either in the control means 5 or in a computing unit of the terminal 7 itself can be copied.

In 8th a program flow chart of an embodiment of a method according to the invention is shown. In a first step S11, several dark images are formed 1 with the x-ray system 20 detected.

In the next step S12, these become dark images 1 generates a noise image NI _D by using Equations (5) and (6) each pixel of the noise image NI _D as a standard deviation of pixel values corresponding to that pixel within all the dark images acquired in the first step S11 1 is determined.

In steps S13 and S14, the average NM _{D of} all pixel values of the pixels of the noise image NI _D and the standard deviation NS _{D of} all pixel values of the pixels of the noise image NI _{D are} determined. The calculation of the mean value NM _D and the standard deviation NS _D is performed by the equations (7) and (8).

In step S15 it is checked for all pixels of the noise image NI _D whether the inequality (for the corresponding pixel or the corresponding pixel value PVNI _{Di of} the pixel in the noise image NI _D) 1 ) or the inequality ( 2 ), in particular with the factor F = 3 is used. If one of the two inequalities (1) or (2) is valid for one pixel in the noise image NI _D , the corresponding pixel of the detector is valid 6 malfunction.

After the step S15, the process ends.

In 9 a further embodiment of a method according to the invention is shown.

In the first step S21 again several dark images 1 detected.

With the equation (6), these become dark images 1 generates a signal image SI _D by making the pixel value of each pixel of the signal image SI _D equal to the average of the pixel values of the corresponding pixels in the dark images 1 is set.

aller Pixelwerte der Pixel des Signalbildes SI_D gemäß der folgenden Gleichung (9) bestimmt.In step S23, the average value

of all pixel values of the pixels of the signal image SI _D according to the following equation (9).

In this case, M corresponds to the number of pixels of the signal image SI _D , wherein in particular the number of pixels of the signal image SI _{D is} equal to the number of pixels of the dark images 1 is.

In a similar manner as in step S12 of FIG 8th is discussed in step S24 from the dark images 1 generates a noise image NI _D by expressing the pixel value of each pixel of the noise image NI _D as the standard deviation of the pixel values of the dark images corresponding to that pixel 1 is determined.

In a similar manner as in step S14 of FIG 8th described embodiment, the standard deviation of all pixel values of the pixels of this noise image NI _{D is} determined in step S25 from the noise image NI _D.

In step S26, it is checked for all pixels of the signal image SI _D whether the inequality (3) or the inequality (4) is satisfied for the corresponding pixel or the corresponding pixel value PVSI _{Di of} the pixel in the signal image SI _D , where again, in particular the same factor F = 3 as in the embodiment of 8th is working. If one of the two inequalities (3) or (4) is valid for one pixel in the noise image SI _D , the corresponding pixel of the detector is valid 6 malfunction.

After the step S26, the process ends.

In 10 a further embodiment of a method according to the invention is shown, which in contrast to the 8th and 9 with bright pictures 2 which are detected in a first step S31 of this method.

In the second step S32 of the method, the overall noise image NI ' _{B is selected} from the bright images 2 by generating, by means of Equations (5) and (6), the pixel value of each pixel of the overall noise image NI ' _B as the standard deviation of the pixel values within the light images corresponding to that pixel 2 is determined. The number of pixels of the overall noise image NI ' _B is therefore equal to the number of pixels of each bright image 2 ,

In the third step S33, a predetermined cutout NI _B (10 × 10 pixels) is cut out of the overall noise image NI'B.

In step S34, the average NM or NM _{B of} the 100 (10 × 10) pixel values of the cut-out NI _{B is determined} according to the equation (8), that is, M = 100.

In step S35, the standard deviation NS or NSB of the 100 pixel values of the cut-out NI _{B is determined} according to the equation (7).

In step S36 it is checked whether for one of the 100 pixel values PVNI _Bi the inequality (1) or the inequality (2) is valid, advantageously with the same factor F = 3 as in the methods of 8th and 9 is working. If so, the corresponding pixel of the detector is 6 malfunction.

To all pixels of the detector 6 To check, steps S33 to S36 of the 10 be repeated until each pixel of the detector 6 at least once to the predetermined section NI _{B of} the overall noise image NI ' _B heard.

LIST OF REFERENCE NUMBERS

1

dark image

2

brightly

4

detector

5

control means

6

radiation generator

7

terminal

8th

screen

9

keyboard

10

mouse

11

C-arm

14

DVD

20

X-ray system

S11-S36

step

SI _D , SI ' _D

Signal image of the dark images

NI ' _D , NI' _D

Noise picture of the dark pictures

SI _B , SI ' _B

Signal image of the light images

NI _B , NI ' _B

Noise picture of the hell pictures

Claims (18)

Method for automatic detection of defect pixels of a detector ( 4 ) of an imaging fluoroscopy system ( 20 ), the method comprising the steps of: capturing multiple images ( 1 ; 2 ) by means of the detector ( 4 ), Generating a noise image (NI _D , NI ' _D , NI _B , NI' _B ) from the images ( 1 ; 2 Determining an average of all pixel values of the noise image (NI _D ; NI '_D; NI _B ; NI' _B ), determining a mean deviation of all the pixel values of the noise image (NI _D ; NI '_D; NI _B ; NI' _B ) from the Mean, and detecting a pixel of the detector ( 4 ) as the defect pixels depending on the corresponding pixel value within the noise image (NI _D , NI ' _D , NI _B , NI' _B ), the average of all pixel values, and the average deviation of all pixel values. Method according to claim 1, characterized in that a pixel of the detector ( 4 ) is recognized as a defective pixel if one of the following inequalities applies to the pixel value PVNI corresponding to the pixel in the noise image (NI _D ; NI '_D; NI _B ; NI' _B ): PVNI> NM + F × NS (1) PVNI <NM - F × NS (2), where NM is the mean of all pixel values of the noise image (NI _D ; NI '_D; NI _B ; NI' _B ), F is a factor and NS is the standard deviation of all pixel values of the noise image (NI _D ; NI '_D; NI _B ; NI' _B ) equivalent. Method for automatic detection of defect pixels of a detector ( 4 ) of an imaging fluoroscopy system ( 20 ), the method comprising the steps of: capturing multiple images ( 1 ; 2 ) by means of the detector ( 4 ), Generating a signal image (SI _D , SI ' _D , SI _B , SI' _B ) from the images ( 1 ; 2 ), wherein each pixel of the signal image (SI _D , SI ' _D , SI _B , SI' _B ) is calculated by determining for the respective pixel a pixel value which is substantially equal to an average value of the pixel values of the plurality of images (FIG. 1 ; 2 ), determining an average of all the pixel values of the signal image (SI _D ; SI '_D; SI _B ; SI' _B ), generating a noise image (NI _D ; NI '_D; NI _B ; NI' _B ) from the images (FIG. 1 ; 2 ), Determining an average deviation of all pixel values of the noise image (NI _D ; NI '_D; NI _B ; NI' _B ) from an average of the pixel values of the noise image (NI _D ; NI '_D; NI _B ; NI' _B ), and detecting of a pixel of the detector ( 4 ) as a defect pixel depending on the corresponding pixel value within the signal image (SI _D , SI ' _D , SI _B , SI' _B ), depending on the mean value of all pixel values of the signal image (SI _D ; SI '_D; SI _B ; SI' _B ) and dependent on the mean deviation of all pixel values of the noise image (NI _D , NI ' _D , NI _B , NI' _B ). Method according to Claim 3, characterized in that one pixel of the detector ( 4 ) is recognized as a defective pixel if one of the following two inequalities applies to the pixel value PVSI corresponding to the pixel in the signal image (SI _D , SI ' _D , SI _B , SI' _B ): PVSI> SM + F × NS (3) PVSI <SM - F × NS (4), where SM is the mean of all pixel values of the signal image (SI _D ; SI '_D; SI _B ; SI' _B ), F is a factor and NS is the standard deviation of all pixel values of the noise image (NI _D ; NI '_D; NI _B ; NI' _B ) equivalent. Method according to one of the preceding claims, characterized in that the noise image (NI _D ; NI '_D; NI _B ; NI' _B ) is generated by dividing each pixel of the noise image (NI _D ; NI '_D; NI _B ; NI' _B ) is determined by determining, for the respective pixel, a pixel value which corresponds to a standard deviation of the pixel values corresponding to that pixel within the plurality of images ( 1 ; 2 ) and / or that the mean deviation corresponds to the standard deviation of all pixel values of the noise image (NI _D , NI ' _{D ,} NI _B , NI' _B ). Method according to one of the preceding claims, characterized in that the multiple images are dark images ( 1 ), the dark images ( 1 ) without radiation from a radiation generator ( 6 ) of the fluoroscopy system ( 20 ), or that the multiple images are bright images ( 2 ), where the light images ( 2 ) with radiation from a radiation generator ( 6 ) of the fluoroscopy system ( 20 ) without an object between the radiation generator ( 6 ) and the detector ( 4 ) be generated. Method according to one of the preceding claims, characterized in that the noise image (NI _D ; NI '_D; NI _B ; NI' _B ) only a same predetermined section within the images ( 1 ; 2 ), and / or that the signal image (SI _D , SI ' _D , SI _B , SI' _B ) only corresponds to a same predetermined section within the images (FIG. 1 ; 2 ) corresponds. Method according to one of the preceding claims, characterized in that the signal image (SI _B , SI ' _B ) only a same predetermined section within the light images ( 2 ), and that the predetermined section has an edge length of 8 to 15 pixels. Method according to claim 8, characterized in that the pixel of the detector ( 4 ) is arranged in the middle of the predetermined cutout. Method for automatic detection of defect pixels of a detector ( 4 ) of an imaging fluoroscopy system ( 20 ), the method comprising a first and a second method, wherein the first method comprises at least two methods which are selected from an amount comprising the method according to claim 1 or 2, the method according to claim 3 or 4, and one of the aforementioned The method of the set of any of claims 5-8, and wherein the pixel of the detector ( 4 ) is detected as a defective pixel if the pixel is detected as a defect pixel according to one of the at least two methods. A method according to claim 10, characterized in that the factor in the at least two methods each have the same value. Method according to one of the preceding claims, characterized in that the factor is in a range of 2 to 4. Imaging fluoroscopy system, wherein the fluoroscopy system ( 20 ) a detector ( 4 ), a radiation generator ( 6 ) and control means ( 5 ), wherein the fluoroscopy system ( 20 ) is configured such that the fluoroscopy system ( 20 ) by means of the control means ( 5 ) and the detector ( 4 ) several pictures ( 1 ; 2 ) detects that the control means ( 5 ) a noise image (NI _D , NI ' _D , NI _B , NI' _B ) depending on the images ( 1 ; 2 ) that the control means ( 5 ) determine an average of all the pixel values of the noise image (NI _D , NI ' _D , NI _B , NI' _B ) that the control means ( 5 ) determine an average deviation of all pixel values of the noise image (NI _D , NI ' _D , NI _B , NI' _B ) from the mean value, and in that the control means ( 5 ) one pixel of the detector ( 4 ) as defect pixels depending on the corresponding pixel value within the noise image (NI _D ; NI '_D; NI _B ; NI' _B ), depending on the average of all pixel values and depending on the average deviation of all pixel values. Imaging fluoroscopy system, wherein the fluoroscopy system ( 20 ) a detector ( 4 ), a radiation generator ( 6 ) and control means ( 5 ), and wherein the fluoroscopy system ( 20 ) is configured such that the fluoroscopy system ( 20 ) by means of the control means ( 5 ) and the detector ( 4 ) a plurality of images (1; 2) detects that the control means (1) 5 ) a signal image (SI _D , SI ' _D , SI _B , SI' _B ) depending on the images ( 1 ; 2 ) by the control means ( 5 ) calculate each pixel of the signal image (SI _D , SI ' _D , SI _B , SI' _B ) by the control means ( 5 ) for the respective pixel determine a pixel value which substantially corresponds to an average value of the pixel values of the plurality of images (1; 2) corresponding to this pixel, that the control means ( 5 ) determine an average of all the pixel values of the signal image (SI _D , SI ' _D ) that the control means ( 5 ) a noise image (NI _D , NI ' _D ) depending on the images ( 1 ; 2 ) that the control means ( 5 ) determine an average deviation of all pixel values of the noise image (NI _D , NI ' _D ) from an average of the pixel values of the noise image (NI _D , NI' _D ), and in that the control means determines one pixel of the detector (NI; 4 ) as a defect pixel depending on the corresponding pixel value within the signal image (SI _D , SI ' _D , SI _B , SI' _B ), depending on the mean value of all pixel values of the signal image (SI _D ; SI '_D; SI _B ; SI' _B ) and dependent on the mean deviation of all pixel values of the noise image (NI _D , NI ' _D , NI _B , NI' _B ). Imaging fluoroscopy system according to claim 12 or 14, characterized in that the fluoroscopy system ( 20 ) is designed for carrying out the method according to one of claims 1-11. Transillumination system according to claim 15, characterized in that the fluoroscopy system is an x-ray system ( 20 ). Computer program product comprising a program and stored directly in a memory of programmable control means ( 5 ) of an imaging fluoroscopy system ( 20 ) with program means for carrying out all the steps of the method according to any one of claims 1-12, when the program in the control means ( 5 ) of the fluoroscopy system ( 20 ) is performed. Electronically readable data carrier with electronically readable control information stored thereon, which are designed in such a way that when using the data carrier ( 14 ) in taxation ( 5 ) of an imaging fluoroscopy system ( 20 ) perform the method according to any one of claims 1-12.

Images