JP2006263055A - X-ray image processing system and x-ray image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an X-ray image processing system which can obtain a stable image processing result by executing the image processing corresponding to a tissue structure of a subject and the correction of the image processing result. <P>SOLUTION: In the X-ray image processing system 1, a structural characteristic quantity to show the physical characteristics of the subject and a signal characteristic quantity of an image signal value possessed by a mammary image are calculated by an image processor 30 for the mammary image generated by an image generation apparatus 10. Then from the calculated structural characteristic quantity and the signal characteristic value, a mammary image having an unstable image processing result is discriminated by a linear discrimination method and the image processing or the correction of the image processing result is applied to the mammary image on the basis of the structural characteristic value and the signal characteristic quantity. The mammary image to which the image processing is applied is transmitted to an output apparatus 5 by a control apparatus 4 and is output by the output apparatus 5. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an X-ray image processing system and an X-ray image processing method for performing various image processes on an X-ray image.

In the medical field, digitization of X-ray images obtained by X-ray imaging has been realized.
Some X-ray images are difficult to interpret due to the characteristics of the image signal and the characteristics of the subject (patient). For example, in the case of examination examination of the breast, the breast has a mixture of mammary gland tissue and adipose tissue, and the breast image can be classified into the following four mammary gland types according to the ratio and the distribution status of the mammary gland tissue.

(1) Fatty type. This is a type in which the breast region is almost completely replaced by fat. Since the fat region has a high density on the image and the lesioned part (microcalcification cluster, tumor, etc.) has a low density, the detection is easy if the lesioned part is included in the imaging range.
(2) Mammary gland scattered type. A type in which mammary gland parenchyma is scattered within the breast area replaced with fat. Detection of a lesion is relatively easy.
(3) Non-uniform high concentration type. A type of fat that is mixed in the mammary gland and has a non-uniform concentration. Since the breast tissue has a lower density than fat on the image, there is a risk that the lesion is hidden by the breast tissue.
(4) High concentration type. There is almost no fat mixed in the mammary parenchyma, and the detection rate of the lesion is low.

  Among the above-mentioned mammary gland types, since the mammary gland tissue is uniformly distributed in the mammary image in which the mammary gland type is classified as a high-concentration type, as shown in FIG. The density region is widened and sufficient contrast cannot be obtained. Therefore, it has been difficult to find a lesion such as a microcalcification cluster in which minute low-concentration points are gathered on the mammary gland region. On the other hand, other mammary gland types such as scattered mammary gland types have high-concentration fat regions mixed in the mammary gland region as shown in FIG. Relatively easy.

In the case of a high-density type breast image, the output density of the mammary gland region may be unstable as compared with other breast types. Conventionally, at the time of image processing, a histogram of the image signal value of a breast tissue that is diagnostically important among breasts is determined to determine a reference signal value, and a normalization process and gradation are determined based on the reference signal value. Image processing such as processing is performed (see, for example, Patent Document 1). According to this method, an equivalent image processing result can be obtained without being influenced by the amount of mammary gland tissue having an individual difference for each subject.
JP 2001-238868 A

  However, even though the image is recognized as a mammary gland tissue, the image signal value varies depending on the density of the mammary gland tissue, resulting in a difference in the shape of the created histogram. For example, when the density of the mammary gland tissue is high, the shape is like a histogram A shown in FIG. 9A, and the image signal value is extremely biased toward the low signal side, whereas when the density is low, The shape is like a histogram B shown in FIG. 9B, and the image signal values are distributed with a certain degree of spread from the histogram A.

  This is because histogram A has a high density of mammary gland tissue and the same image signal value, whereas histogram B has a variation in density of the mammary gland tissue and the image signal value also varies. For this reason, for example, even if the reference signal value is determined at a position occupying 10% from the low signal side in each of the histograms A and B, in the case of the histogram A, the reference signal value is determined on the lower signal side than the histogram B. It will be. Therefore, as a result of the image processing, the image processing is performed so that the output density of the mammary gland region is higher than the breast image of the histogram B.

  Thus, the tendency of the image signal value to be biased toward the low signal side is common in a breast image in which a mammary gland tissue is developed such that almost no fat tissue is mixed, that is, a breast image of a high-concentration type breast. However, as described above, since the histogram shape (image signal value distribution state) differs depending on the density of the mammary gland tissue, even if the breast images are classified into the same high density type, the output density of the mammary gland region necessarily increases. Is not limited.

  Therefore, it is not possible to obtain a stable image processing result simply by classifying the mammary gland type as a high-concentration type and performing image processing according to that type, and also consider the distribution state of image signal values of the breast image. Thus, it is necessary to perform image processing.

  An object of the present invention is to provide an X-ray image processing system capable of obtaining a stable image processing result even when a subject has a tissue structure having a large individual difference.

The invention according to claim 1 is an X-ray image processing system,
First extraction means for extracting a structural feature amount relating to the tissue structure of the subject from an X-ray image obtained by photographing the subject;
Second extraction means for extracting a signal feature amount related to an image signal value of a subject area from the X-ray image;
Image processing means for performing image processing on the X-ray image in accordance with the extracted structural feature amount and signal feature amount;
It is characterized by providing.

The invention described in claim 2 is the X-ray image processing system according to claim 1,
The image processing means, when image processing has already been performed on the X-ray image, performs image processing according to the structural feature amount and the signal feature amount and corrects the previous image processing result. To do.

The invention described in claim 3 is the X-ray image processing system according to claim 1 or 2,
Based on the extracted structural feature amount and signal feature amount, the X-ray image includes image discrimination means for discriminating whether or not the X-ray image is a specific image to be subjected to specific image processing,
The image processing means, when it is determined that the X-ray image is a specific image as a result of the determination, performs image processing corresponding to the structural feature amount and the signal feature amount on the X-ray image. To do.

The invention according to claim 4 is the X-ray image processing system according to any one of claims 1 to 3,
The X-ray image is a breast image obtained by photographing a subject's breast,
The first extracting means extracts a feature amount indicating a mammary gland tissue amount in the breast as the structural feature amount based on an image signal value of the X-ray image,
The second extracting means extracts a feature quantity indicating a distribution state of image signal values of mammary gland tissue as the signal feature quantity based on the image signal value of the X-ray image.

The invention according to claim 5 is an X-ray image processing method,
A first extraction step of extracting a structural feature amount related to the tissue structure of the subject from an X-ray image obtained by photographing the subject;
A second extraction step of extracting a signal feature amount related to an image signal value of a subject area from the X-ray image;
An image processing step of performing image processing on the X-ray image in accordance with the extracted structural feature amount and signal feature amount;
It is characterized by including.

The invention according to claim 6 is the X-ray image processing method according to claim 5,
In the image processing step, when image processing has already been performed on the X-ray image, image processing according to the structural feature amount and signal feature amount is performed to correct the previous image processing result. To do.

The invention described in claim 7 is the X-ray image processing method according to claim 5 or 6,
Based on the extracted structural feature quantity and signal feature quantity, the X-ray image includes an image discrimination step for discriminating whether or not the X-ray image is a specific image to be subjected to specific image processing;
In the image processing step, when it is determined that the X-ray image is a specific image as a result of the determination, the X-ray image is subjected to image processing according to the structural feature amount and the signal feature amount. To do.

The invention according to claim 8 is the X-ray image processing method according to any one of claims 5 to 7,
The X-ray image is a breast image obtained by photographing a subject's breast,
In the first extraction step, based on the image signal value of the X-ray image, a feature quantity indicating a mammary gland tissue quantity in the breast is extracted as the structural feature quantity,
In the second extraction step, a feature amount indicating a distribution state of image signal values of mammary gland tissue is extracted as the signal feature amount based on the image signal value of the X-ray image.

  According to the first and fifth aspects of the present invention, the X-ray image is subjected to image processing in accordance with the structural feature amount and the signal feature amount, so that the feature of the tissue structure of the subject and the image signal value of the X-ray image is obtained. Reflected image processing can be performed. Since the features related to the tissue structure such as the amount of the tissue structure vary depending on the subject, the image processing result may not be constant because the output density will be different if the same image processing is performed uniformly. In addition, the image signal value may be biased due to a difference in the density of the tissue structure, and a certain image processing result cannot be obtained by considering only the characteristics of the tissue structure. Therefore, by performing image processing in consideration of the structural feature amount and the signal feature amount as in the present invention, a constant image processing result can be obtained for any X-ray image.

  According to the second and sixth aspects of the invention, correction can be performed so that the structural feature amount and the signal feature amount are reflected also in the X-ray image that has been subjected to image processing. Therefore, even when the image quality is inappropriate, such as when the output density is increased only in an image region of a certain tissue structure, correction can be performed so that the image quality is in a certain range.

  According to the third and seventh aspects of the invention, since specific image processing corresponding to the structural feature amount and the signal feature amount is performed on the X-ray image determined as the specific image, the specific image is applied to all the X-ray images. There is no need to perform processing, and processing efficiency is good.

  According to the fourth and eighth aspects of the present invention, in the case of a breast image, it is possible to perform image processing according to each feature amount of the mammary gland tissue amount and the distribution state of the image signal value of the mammary gland tissue. The mammary gland tissue has large individual differences in amount and density, and if the same image processing is performed on all breast images, the output density of the mammary gland region is not constant. Therefore, as in the present invention, by performing image processing in consideration of the amount of mammary tissue and the distribution state of image signal values of the mammary tissue, a constant image processing result can be obtained regardless of individual differences in the mammary tissue. It becomes.

  In the present embodiment, an example will be described in which image processing is performed on a breast image obtained by X-ray imaging using a breast as an imaging region.

First, the configuration will be described.
FIG. 1 shows a system configuration of an X-ray image processing system 1 in the present embodiment.
As shown in FIG. 1, the X-ray image processing system 1 includes a CR network unit 2 including an image generation device 10, a control device 20, and an image processing device 30, and a RIS (Radiology Information System) / HIS (Hospital Information System) 3. The management device 4 and the output device 5 are configured to be able to transmit / receive information to / from each other via the network bus N1.

First, the CR network unit 2 will be described.
In the CR network unit 2, a plurality of image generation apparatuses 10, a plurality of control apparatuses 20, and an image processing apparatus 30 are connected via a network bus N2. In addition, the installation number of each apparatus 10-30 is not specifically limited.

  The image generation device 10 generates digital image data of a patient's captured image (X-ray image) according to control by any of the control devices 20. As the image generating apparatus 10, an apparatus that generates an X-ray image (X-ray image) by X-ray imaging such as a CR (Computed Radiography), an FPD (Flat Panel Detector), or a cassette-dedicated reader can be applied.

  The image generation apparatus 10 is an apparatus compliant with DICOM (Digital Imaging and Communication in Medicine), which is a standard for X-ray images and communication thereof, and related information (for example, X-ray images are captured) regarding the generated X-ray images. Patient information, imaging information regarding imaging or examination, examination information, etc.) can be input from the outside and can be automatically generated. The image generation apparatus 10 appends the related information to the header of the generated X-ray image, and transmits the X-ray image and the related information to the control apparatus 10 that has controlled the image generation.

  The control device 20 acquires imaging order information (patient information about a patient to be imaged and instruction information about imaging and examination) from the RIS / HIS 3, and controls image generation of the image generation device 10 according to the imaging order information. In addition, the X-ray image generated by the image generation apparatus 10 is acquired and stored as an original image, and distribution of the X-ray image is controlled.

The image processing device 30 performs various types of image processing on the X-ray image transmitted from the control device 20.
FIG. 2 shows the internal configuration of the image processing apparatus 30.
As shown in FIG. 2, the image processing apparatus 30 includes a control unit 31, an operation unit 32, a display unit 33, a communication unit 34, a storage unit 35, an image processing unit 36, and an image memory 37.

  The control unit 31 includes a CPU (Central Processing Unit), a RAM (Random Access Memory), and the like. The CPU reads various control programs from the storage unit 35 and develops them in the RAM, performs various calculations according to the programs and performs processing. Control execution.

  The control unit 31 executes, for example, image determination processing according to the present invention (see FIG. 6) according to the program. The control unit 31 has a determination logic necessary during this processing. Details of processing and determination logic will be described later.

  The operation unit 32 includes a keyboard, a mouse, and the like, generates an operation signal corresponding to a key operation and a mouse operation, and outputs the operation signal to the control unit 31.

  The display unit 33 includes an LCD (Liquid Crystal Display) or the like, and performs various displays such as an operation screen at the time of image processing and an X-ray image after processing in accordance with instructions from the control unit 31.

  The communication unit 34 communicates with external devices on the network buses N1 and N2 in accordance with instructions from the control unit 21. For example, an X-ray image subject to image processing is received from the control device 20, or the processed X-ray image is transmitted to the management device 4 on the network bus N1 as an image for interpretation.

  The storage unit 35 stores a control program executed by the control unit 31 and various image processing programs executed by the image processing unit 36. The storage unit 35 stores parameters necessary for execution of processing by a program, data of processing results, and the like.

  The image processing unit 36 reads various image processing programs stored in the storage unit 35, and performs various image processing such as normalization processing, gradation conversion processing, frequency processing, dynamic range compression processing, etc. on the X-ray image according to the program. .

Hereinafter, each image processing will be described.
The normalization process is a process for correcting fluctuations in X-ray dose caused by variations in patient body shape, imaging conditions, and the like. First, before normalization processing, the image processing unit 36 extracts a region (ROI) that is noticed from diagnosis in an X-ray image, and creates a histogram of image signal values of the extracted region. In the case of a breast image, a subject area is extracted as an ROI. Based on this histogram, reference signal values D1 and D2 are determined. The reference signal values D1 and D2 are set as a reference level of image data by setting a signal value at a predetermined ratio such as 10% from the low signal side and the high signal side of the histogram. The optimal ratio is obtained in advance by statistics by (ROI).

  When the image processing unit 36 sets the signal values D1 and D2 that are predetermined ratios in the histogram (in the example of FIG. 3A, 10% from the low signal side and 15% from the high signal side) as the reference signal values, As shown in FIG. 3B, normalization processing for converting the reference signal values D1 and D2 into desired levels S1 and S2 is performed using the normalization LUT. The LUT is a table of the normalization curve C1 in FIG. In FIG. 3B, the X-axis indicates the signal value and the Y-axis indicates the density level. In this embodiment, the signal value in the range of 0 to 4095 is 0 (in the case of a film, 0 in consideration of the density of the film itself). .2) to 4.0 is shown as an example of normalization.

Next, the image processing unit 36 performs gradation conversion processing on the normalized image that has been normalized.
The gradation conversion process is a process for adjusting the gradation of the X-ray image according to the output characteristics of a monitor, a film, etc., and the output density and contrast at the time of output can be adjusted by this process. In the gradation conversion process, for example, the LUT based on the gradation conversion curve C2 as shown in FIG. 3C is used to convert the normalized image reference levels S1 and S2 into the output density levels T1 and T2. The levels T1 and T2 correspond to predetermined luminance or density at the time of output.
Also, the output density or contrast can be corrected by changing the slope of the gradation conversion curve C2 or rotating or translating the gradation conversion curve C2.

Next, frequency enhancement processing and dynamic range compression processing will be described.
The frequency enhancement process is a process for adjusting the sharpness of an image. In the frequency enhancement process, for example, in order to control the sharpness by the non-sharp mask process shown in Expression (1), the function F is determined by the method shown in Japanese Patent Publication Nos. 62-62373 and 62-62376.

Soua = Sorg + F (Sorg−Sus) (1)
Note that Soua is image data after processing, Sorg is image data before frequency enhancement processing, and Sus is unsharp data obtained by averaging the image data before frequency enhancement processing.

In this frequency enhancement process, for example, F (Sorg-Sus) is set to β × (Sorg-Sus), and β (enhancement coefficient) is substantially linear between the reference values T3 and T4 as shown in FIG. To be changed.
Also, as shown by the solid lines in FIG. 4B, when values “A” and “B” are set to emphasize low luminance, β from the reference value T3 to the value “A” is maximized. The value “B” −the reference value T3 is minimized. Further, β is changed substantially linearly from the value “A” to the value “B”. When emphasizing high luminance, as shown by a broken line, β from the reference value T3 to the value “A” is minimized and maximized from the value “B” to the reference value T4. In addition, from the value “A” to the value “B”, β changes almost linearly. Although not shown, when medium luminance is emphasized, β of values “A” to “B” is maximized. As described above, in the frequency enhancement processing, the sharpness of an arbitrary luminance portion can be controlled by the function F.

  Further, the frequency enhancement processing method is not limited to the non-sharp mask processing, and a technique such as a multi-resolution method disclosed in Japanese Patent Laid-Open No. 9-44645 may be used. In the frequency enhancement process, the frequency band to be enhanced and the degree of enhancement are set based on the imaging region, imaging position, imaging conditions, imaging method, etc., as in the selection of the basic gradation conversion curve in the gradation conversion process. Is done.

  In the dynamic range compression process, the function G is determined by the method disclosed in Japanese Patent Publication No. 266318 in order to perform control to keep the density range easy to see by the compression process shown in Expression (2).

Stb = Sorg + G (Sus) (2)
Note that Stb is processed image data, Sorg is image data before dynamic range compression processing, and Sus is unsharp data obtained by averaging the image data before dynamic range compression processing.

  Here, as shown in FIG. 5 (a), G (Sus) has a characteristic that G (Sus) increases when the unsharp data Sus becomes smaller than the level “La”. The image data Sorg shown in FIG. 5B is image data Stb in which the dynamic range on the low density side is compressed as shown in FIG. 5C. Further, as shown in FIG. 5D, when G (Sus) has such characteristics that G (Sus) decreases when the unsharp data Sus becomes smaller than the level “Lb”, The density is high, and the dynamic range on the high density side of the image data Sorg shown in FIG. 5B is compressed as shown in FIG. In the dynamic range compression processing, the correction frequency band and the degree of correction are set based on the imaging region, the imaging posture, the imaging conditions, the imaging method, and the like.

Here, the reference values T3 and T4 and the values “A” and “B” or the levels “La” and “Lb”, which are processing conditions in the frequency enhancement processing and dynamic range compression processing, are determined by the reference signal values D1 and D2. It is calculated | required by the method similar to the method.
As described above, according to the above-described embodiment, stable gradation processing can be performed without depending on the distribution of image data even when the ratio between the high density region and the low density region changes greatly. Therefore, a radiation image having a density and contrast suitable for diagnosis and the like can always be obtained stably.

  The image data used for determining the processing conditions may be, for example, image data that has been subjected to thinning processing. In this case, since the amount of data is reduced, the processing speed can be improved and the memory capacity can be saved. In this case, it is preferable to use image data that has been thinned so that the effective pixel size of the reduced image by thinning is 0.4 mm to 10.0 mm, preferably 1.0 mm to 6.0 mm.

  The image memory 37 is a memory for storing an X-ray image to be image processed and the processed image.

Next, each device on the network bus N1 will be described.
The RIS / HIS 3 is a system that comprehensively manages information in a radiology department or a hospital. In the RIS / HIS 3, an examination request from a doctor is received and the above-described imaging order information is generated. Then, the generated imaging order information is transmitted to the control device 20 via the network buses N1 and N2.

  The management device 4 stores and manages the image for image interpretation generated by the CR network unit 2 and subjected to image processing together with related information. Also, the distribution destination of the stored image for interpretation is determined, and the image for interpretation is transmitted to the determined distribution destination.

  The output device 5 outputs the image for image interpretation distributed by the management device 4. As the output device 5, a monitor that performs display output, a film output device that performs film output, and the like are applicable.

Next, the operation of the X-ray image processing system 1 will be described.
First, the process of generating a breast image will be described.
When a doctor's examination request is made, imaging order information corresponding to the request is generated in the RIS / HIS 3 and the imaging order information is distributed to each control device 20. The control device 20 displays a list of shooting order information according to the operation of the engineer, and based on the shooting order information selected by the engineer, registration of the recording means (cassette) used for the shooting (cassette ID and shooting order information). And control information (for example, reading pitch) when reading an image from the cassette are set.

  When imaging is performed using a cassette, and the captured cassette is inserted into an arbitrary image generation device 10 among a plurality of image generation devices 10 (reading devices dedicated to cassettes), the image generation device 10 The cassette ID is read. Then, of the control information input from the control device 20, the breast image is read from the cassette that has been imaged according to the control information corresponding to the ID, and the data is generated. Then, when imaging information or the like is written in the header of the breast image as related information of the generated breast image, the imaging order information using the cassette is transmitted to the registered control device 20. That is, regardless of which control device 20 the engineer operates, the breast image generated by the image generation device 10 is always transmitted to the control device 20 operated by the engineer.

In the control device 20, related information such as patient information, imaging information, and examination information is attached to the X-ray image (original image) acquired from the image generation device 10 based on the imaging order information and stored. The duplicate is then generated and transmitted to the image processing device 30 for image processing.
In the image processing apparatus 30, when a breast image to be image-processed is input, a feature amount relating to the breast tissue structure (hereinafter referred to as a structural feature amount) and a feature amount relating to an image signal of the breast image (hereinafter referred to as a signal feature amount). An image determination process for performing specific image processing according to is started. It should be noted that the following image determination processing can be performed on a breast image that has been subjected to image processing once.

Hereinafter, the image determination process will be described with reference to FIG.
In the image determination process shown in FIG. 6, first, a breast region is extracted from a breast image in the image processing unit 17, and then a mammary gland region and a fat region are extracted from the extracted breast region (step S <b> 1).

  The breast region can be extracted using a histogram of the entire image. First, the threshold value is determined from the histogram by a discriminant analysis method (a method for determining a threshold value so that the discriminant ratio between intra-class variance and inter-class variance is maximized in two classes when the histogram is divided into two classes) A breast image is binarized using a threshold value. At this time, a blank region (a region that does not correspond to the subject portion and is directly irradiated with X-rays) in the breast image is “1” due to binarization because of high density, and other subject regions Is expected to be “0”. Therefore, the binarization can divide the breast image into a subject region and other unexposed regions (regions where X-rays are directly irradiated without passing through the subject).

The mammary gland region and the fat region can be extracted by determining a threshold value in the local region in the extracted breast region.
For example, as described in Japanese Patent Application Laid-Open No. 2001-238868, a local region is set in a breast region, a threshold value is set based on a pixel value in the local region, and the breast region is binarized. Are extracted as mammary gland regions. This is sequentially performed for the entire breast region. The threshold value may be a value obtained by subtracting a certain amount from the average value of the pixel values in the local region, or may be a value obtained by multiplying a coefficient in the average value. The fat region is extracted by the same process.
Note that the extraction method of the breast region, the mammary gland region, and the fat region is an example, and may be extracted by other methods.

  When each area is extracted, the ratio of the mammary gland area in the breast area is calculated as the structural feature quantity of the breast by the image processing unit 36, and the mammary gland type of the breast is determined by the control unit 31 based on the ratio (step S2). ). The ratio of the mammary gland region occupying the breast region is obtained from the number of pixels of the mammary gland region relative to the number of pixels of the breast region, that is, the area. If the ratio of the mammary gland region is 90% to 100%, almost the entire breast region is occupied by the mammary gland tissue, so that it is determined as the high concentration type. Moreover, if it is 20 to 90%, it is judged that the mammary gland tissue is scattered in a part of the breast region, and is classified as a heterogeneous high concentration type or a mammary gland scattered type. Further, if it is 1% to 20%, there is almost no mammary gland tissue in the mammary gland region, and it is judged as the fatty type.

  Next, in order to detect the distribution state of the image signal values in the mammary gland region, the image processing unit 36 calculates the variance of the image signal values in the mammary gland region as a signal feature amount (step S3).

  Next, the ratio of the mammary gland region calculated as the structural feature amount and the variance of the image signal value calculated as the signal feature amount are input to the discrimination logic of the control unit 31, and the breast image to be image-processed is based on the output result. Then, the control unit 31 determines whether the image is a specific image that requires specific image processing or correction based on each feature amount (step S4). A specific image is an image that is classified as a high-density type and has a biased image signal value, and the image processing results are likely to be unstable. A specific image processing or correction according to the amount is required.

  Here, the discrimination logic will be described. The discriminating logic outputs a result of discriminating whether or not the breast image is a specific image from each feature amount input by the linear discriminating method, and is constructed from sample data in advance.

Hereinafter, a construction method of the discrimination logic will be described.
First, a plurality of sample images of a breast image P1 that has been classified in advance as a high-concentration type and that is determined to have a bias in the distribution of image signal values in the mammary gland region, and other breast images P2 are prepared. That is, breast images belonging to a class of specific images and a class other than that are prepared. Next, the ratio of the mammary gland area and the variance of the image signal value in the mammary gland area are obtained from each sample image, and plotted with the ratio of the mammary gland area on the X axis and the variance of the image signal value on the Y axis as shown in FIG. . Then, from these sample data (x, y), a linear discriminant that classifies the class of the breast image P1 that is the specific image and the class of the breast image P2 that is not the specific image is calculated. In FIG. 7, the linear discriminant is represented by a line segment C3. As shown in FIG. 7, it can be seen that the breast image P1 group and the breast image P2 group are substantially classified by the line segment C3.

  In the discrimination logic, the distance to the line segment C3 indicated by the linear discriminant is calculated from the ratio of the mammary gland region input for the breast image to be discriminated and the variance (x, y) of the image signal value. Based on the distance, it is determined whether the breast image to be determined belongs to the class of the specific image or the class that is not, and the result is output.

For example, in the case of breast images a, b, and c (corresponding to a, b, and c shown in FIG. 7) having data (x, y) shown in the following table (1), the distance from the line segment C3 is long. In other words, it can be seen that the output density of the mammary gland region tends to increase as the ratio of the mammary gland region increases and the variance of the image signal value decreases.

  When the discrimination logic determines that the breast image is a specific image (step S4; specific image), the image processing unit 36 performs image processing on the breast image according to the structural feature amount and the signal feature amount. Is done. Alternatively, if the breast image to be image-processed has already been subjected to image processing, the image processing result is corrected by performing image processing again with each feature amount (step S5).

  For example, in the above discrimination logic, the distance from the line segment C3 reflects the structural feature amount and the signal feature amount, and the larger the distance from the line segment C, the larger the output density of the mammary gland region of the output image. Tends to be high and the contrast tends to be small.

  Thus, for example, in the case of tone conversion processing, the image processing conditions are changed according to the distance, such as changing the reference signal value of the mammary gland region, rotating the tone conversion curve, or moving the breast transformation image. Image processing for lowering the output density of the mammary gland region is performed. The optimum concentration range in the mammary gland region is 1.2 to 1.59 (according to the image evaluation standard of the Mammography Screening Accuracy Control Central Committee). Therefore, it is preferable to perform image processing or correct the image processing result so as to be within this range.

  As a specific example, if the distance to the line segment C3 is 10 to 15, the position of the reference signal value D1 is moved 5% to the high signal side. If the distance is 3 to 9, the reference signal value D1 is 1%. The output density is adjusted by moving to the high signal side.

  Alternatively, in the mammary gland region, it is output with the same output density and the contrast is expected to be lowered. Therefore, the contrast may be increased by changing the gradient of the gradation conversion curve. Further, the enhancement degree β of the frequency processing may be adjusted to increase the sharpness, and the outline enhancement of the microcalcification cluster may be performed.

  On the other hand, when it is determined by the determination logic that the breast image is not a specific image (step S4; other), normal image processing is performed on the breast image under standard image processing conditions (step S6).

  When image processing or the like is performed on the breast image, the image determination process ends. The processed image after the image processing is displayed on the display unit 33 of the image processing apparatus 30, and the image quality is confirmed by an engineer. When the engineer inputs an operation to confirm the image as an image for interpretation, the processed image is transmitted to the management device 4 via the communication unit 34 as an image for interpretation. The interpretation image transmitted to the management device 4 is output to the output device 5 and used for diagnosis.

  As described above, according to the present embodiment, for the breast image to be image-processed, the structural feature amount of the mammary gland tissue that easily causes individual differences depending on the subject, and the signal feature amount of the image signal value of the mammary gland region in the breast image. Appropriate image processing is performed. As a result, image processing can be performed so as to be within an appropriate image quality range in accordance with the tissue structure of the subject, and a stable image processing result can always be obtained. Therefore, it is possible to achieve uniform image quality that tends to vary depending on the tissue structure characteristics of the subject.

  In addition, since the breast image determined as the specific image from the structural feature amount and the signal feature amount is subjected to image processing according to each feature amount, the specific image processing is performed only on the breast image determined as the specific image. The processing efficiency can be improved by performing normal image processing on the other breast images.

  Even in the case of a breast image that has already undergone image processing, it is possible to correct the previous image processing result by extracting the structural feature amount and the signal feature amount and performing image processing corresponding thereto again. Is possible. Thereby, as a result of confirming the output breast image, if the image quality is inappropriate, such as the output density of the mammary gland region is high, the image quality can be corrected to be in a certain range.

The X-ray image processing system 1 of the present embodiment is a preferred example to which the present invention is applied, and is not limited to this.
For example, in the above description, an example of a breast image has been described, but the present invention can also be applied to an X-ray image of another imaging region (for example, a chest, a head, etc.).

  For example, in the case of an abdominal front image, in the case of an extremely thick subject, the fat part tends to be greatly biased to the lower abdomen, and the histogram tends to be biased to the low signal side as in the case of a high-density type breast image. In this case, since pixels having the same signal value are distributed as a whole, the output image has a low contrast and the output density becomes unstable. Therefore, a feature quantity indicating the size of the body of the subject (for example, a ratio of the subject area included in the entire image) is calculated as the structural feature quantity of the subject, and a feature quantity indicating the distribution state of the signal value as the signal feature quantity ( For example, it is possible to determine whether or not specific image processing is necessary by calculating the variance of signal values in the ROI region and inputting the same into the determination logic in the same manner as the above-described breast image. The corresponding image processing can be performed.

  In the above description, the image processing device 30 determines whether the image is a specific image and performs image processing according to the determination result. However, any device may be responsible for the function. . For example, the control device 20 may make a determination, and the image processing device 30 may perform image processing in response to the determination result. Further, the image processing device 30 may perform standard image processing uniformly, perform the image determination processing in the management device 4, and perform correction only for breast images that require correction of the image processing result.

  In the above embodiment, the ratio of the mammary gland region occupying the breast region is obtained as the structural feature amount. However, the present invention is not limited to this as long as the feature amount indicates the tissue structure feature of the subject. May be applied. As the age is younger, there is a tendency that the amount of mammary tissue tends to be larger. For example, if the age is 20's and the image signal value is biased, it can be determined as a specific image. Age information can be acquired from patient information attached as related information to a breast image.

  Further, the variance of the image signal values in the mammary gland region is obtained as the signal feature amount. However, as long as the feature amount indicates the distribution state of the image signal value, the shape of the breast region histogram may be used.

  Furthermore, the linear discriminant method was used as a method for determining whether or not the image is a specific image using the structural feature amount and the signal feature amount, but other discriminant analysis methods such as Mahalanobis distance and principal component analysis are applied. Alternatively, if it can be determined using the feature value, multivariate analysis such as a neural network or a support vector machine may be performed by adding other factors.

It is a figure which shows the system configuration | structure of the X-ray image processing system in this embodiment. It is a figure which shows the internal structure of the image processing apparatus of FIG. (A) is a figure which shows the example of a histogram of an image signal value, (b) is a figure which shows the example of a characteristic curve used at the time of a normalization process. Further, (c) is a diagram showing an example of a gradation conversion curve used in the gradation conversion process. It is a figure explaining a frequency process. It is a figure explaining a dynamic range compression process. It is a flowchart explaining the image judgment process performed by an image processing apparatus. It is a figure explaining the discrimination method of the specific image by a linear discrimination method. (A) is a figure which shows the example of a high density type breast image, (b) is a figure which shows the breast image example of a mammary gland mixed type. (A) is a figure which shows the example of a histogram when distribution of the image signal value of a mammary gland area | region has a bias | inclination, (b) is a figure which shows the example of a histogram when distribution has a certain extent.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 X-ray image processing system 2 CR network part 10 Image generation apparatus 20 Control apparatus 30 Image processing apparatus 31 Control part 37 Image processing part 4 Management apparatus 5 Output apparatus

Claims (8)

First extraction means for extracting a structural feature amount relating to the tissue structure of the subject from an X-ray image obtained by photographing the subject;
Second extraction means for extracting a signal feature amount related to an image signal value of a subject area from the X-ray image;
Image processing means for performing image processing on the X-ray image in accordance with the extracted structural feature amount and signal feature amount;
An X-ray image processing system comprising:   The image processing means, when image processing has already been performed on the X-ray image, performs image processing according to the structural feature amount and the signal feature amount and corrects the previous image processing result. The X-ray image processing system according to claim 1. Based on the extracted structural feature amount and signal feature amount, the X-ray image includes image discrimination means for discriminating whether or not the X-ray image is a specific image to be subjected to specific image processing,
The image processing means, when it is determined that the X-ray image is a specific image as a result of the determination, performs image processing corresponding to the structural feature amount and the signal feature amount on the X-ray image. The X-ray image processing system according to claim 1 or 2. The X-ray image is a breast image obtained by photographing a subject's breast,
The first extracting means extracts a feature amount indicating a mammary gland tissue amount in the breast as the structural feature amount based on an image signal value of the X-ray image,
The second extraction means extracts a feature quantity indicating a distribution state of image signal values of mammary gland tissue as the signal feature quantity based on an image signal value of the X-ray image. The X-ray image processing system according to any one of the above. A first extraction step of extracting a structural feature amount related to the tissue structure of the subject from an X-ray image obtained by photographing the subject;
A second extraction step of extracting a signal feature amount related to an image signal value of a subject area from the X-ray image;
An image processing step of performing image processing on the X-ray image in accordance with the extracted structural feature amount and signal feature amount;
An X-ray image processing method comprising:   In the image processing step, when image processing has already been performed on the X-ray image, image processing according to the structural feature amount and signal feature amount is performed to correct the previous image processing result. The X-ray image processing method according to claim 5. Based on the extracted structural feature quantity and signal feature quantity, the X-ray image includes an image discrimination step for discriminating whether or not the X-ray image is a specific image to be subjected to specific image processing;
In the image processing step, when it is determined that the X-ray image is a specific image as a result of the determination, the X-ray image is subjected to image processing according to the structural feature amount and the signal feature amount. The X-ray image processing method according to claim 5 or 6. The X-ray image is a breast image obtained by photographing a subject's breast,
In the first extraction step, based on the image signal value of the X-ray image, a feature quantity indicating a mammary gland tissue quantity in the breast is extracted as the structural feature quantity,
In the second extraction step, a feature amount indicating a distribution state of image signal values of mammary gland tissue is extracted as the signal feature amount based on the image signal value of the X-ray image. The X-ray image processing method according to any one of the above.

Images