JP2005052329A - Abnormal shadow candidate detector and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To appropriately treat the mamma according to a difference in its size when detecting an abnormal shadow candidate in a medical radiogram setting the mamma as the subject. <P>SOLUTION: An area calculation means 10 calculates an area R of a subject region in an original image P, a detection threshold determination means 21 automatically determines the detection threshold of the abnormal shadow candidate based on the calculated area R, and an abnormal shadow candidate detection means 30 detects the abnormal shadow candidate based on the determined detection threshold. When the area R is a predetermined value or more, the subject (the mamma) expressed on the original image P is estimated to be the fatty mamma, wherein it is relatively easy to detect the abnormal shadow candidate, on the other hand, when the area R is the predetermined value or less, the subject (the mamma) expressed in the original image P is estimated to be the high concentration mammary gland type mamma, wherein it is relatively hard to detect the abnormal shadow candidate. This constitution lowers the detection sensitivity so as to prevent an erroneous detection of regarding another structure as the abnormal shadow candidate. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to an apparatus for detecting abnormal shadow candidates in a medical radiographic image having a breast as a subject.

  In the medical field, a computer-aided diagnosis system (CAD: Computer Aided Diagnosis) that automatically detects abnormal shadow candidates in an image and highlights the detected abnormal shadow candidates is known.

As a method for detecting an abnormal shadow candidate, for example, image processing using an iris filter is performed on a radiation image (mammography) of a breast, and an output value thereof is subjected to threshold processing, whereby a tumor shadow that is one form of breast cancer or the like ( One form of abnormal shadow) and image processing using a morphological filter are performed automatically, and the output value is thresholded to form a microcalcification shadow that is another form such as breast cancer. A method of automatically detecting a candidate (one form of abnormal shadow) is known (for example, Patent Document 1).
JP-A-8-294479

  By the way, in the case of breast cancer diagnosis, since the internal characteristics differ depending on the size of the breast, the difficulty in detecting abnormal shadow candidates in the radiation image of the breast varies. In other words, small breasts tend to be high-concentration mammary type, and large breasts tend to be fatty, and detection of abnormal shadow candidates is easier with fatty breasts. Has been.

  However, detection of an abnormal shadow candidate according to the difference in internal characteristics depending on the size of the breast has not been performed.

  The present invention has been made in view of such circumstances, and when detecting an abnormal shadow candidate in a medical radiographic image having a breast as a subject, appropriate processing according to the difference in the size of the breast is performed. An object of the present invention is to provide an abnormal shadow candidate detection apparatus and program that enable the above.

  An abnormal shadow candidate detection apparatus according to the present invention includes an area calculation unit that calculates an area of a subject region in an image in an apparatus having an abnormal shadow candidate detection unit that detects an abnormal shadow candidate in a medical radiographic image having a breast as a subject. A detection level determining means for determining a detection level of the abnormal shadow candidate based on the calculated area, and the abnormal shadow candidate detecting means detects the abnormal shadow candidate based on the determined detection level. It is characterized by that.

  Next, a program according to the present invention causes a computer to function as each of the above means.

That is, in a program for causing a computer to function as an abnormal shadow candidate detecting means for detecting an abnormal shadow candidate in a medical radiation image having a breast as a subject,
An area calculation unit that calculates the area of the subject region in the image and a detection level determination unit that determines the detection level of the abnormal shadow candidate based on the calculated area are determined as the abnormal shadow candidate detection unit. The present invention is characterized in that an abnormal shadow candidate is detected based on the detected level.

  Next, details of the image processing apparatus and the program according to the present invention will be described.

  As a specific example of the process of “detecting an abnormal shadow candidate”, an image process using an iris filter is performed, and a threshold value process is performed on the output value of the tumor shadow (a form of abnormal shadow) that is a form of breast cancer or the like. By performing processing that automatically detects candidate regions and image processing using a morphological filter and thresholding the output value, microcalcification shadows that are other forms such as breast cancer (a form of abnormal shadows) For example, a process for automatically detecting candidate areas).

  The “subject region” may be, for example, a region (P−Pc = Pa + Pb) obtained by removing the background region (Pc) from the entire image (P) in the side image of the breast in FIG. It may be a breast region (P−Pc−Pb = Pa) excluding the muscle region (Pb).

  As a specific method of “calculating the area of the subject region”, it is conceivable to first extract the subject region, then count the number of pixels present in the subject region, and use the number of pixels as the area.

  As a specific example of the “detection level”, a detection threshold and a detection rank in the abnormal shadow candidate detection process can be considered. The detection rank refers to the number of candidates detected by the abnormal shadow candidate detection process described above, and the number of abnormalities that are set as detection ranks in descending order of the output value (the possibility of abnormal shadows is high). Detect as a shadow candidate.

  A specific example of “determining the detection level of an abnormal shadow candidate based on the area” is shown below.

  First, when the area is within the predetermined range, the detection threshold is determined to be a predetermined reference value, and when the area is larger than the predetermined range, the area is more abnormal than when the area is within the predetermined range. A detection threshold is determined so that a shadow candidate can be easily detected, and when the area is smaller than the predetermined range, an abnormal shadow candidate is less likely to be detected than when the area is within the predetermined range. It is conceivable to determine the detection threshold. If the area is larger than the predetermined range, it is presumed to be a fat breast. In this case, since other structures such as the mammary gland are not concentrated, false detection is less likely to occur. The detection threshold is determined so that it can be easily detected. If the area is smaller than the predetermined range, it is estimated that the breast is a high-concentration breast type breast. In this case, other structures such as the mammary gland are concentrated. Therefore, since erroneous detection is likely to occur, it means that the detection threshold value is determined so that it is more difficult to detect.

  When the area is within the predetermined range, the number of detection ranks is determined to be a predetermined reference value. When the area is outside the predetermined range, the detection rank is set to a value smaller than the predetermined reference value. It is also possible to determine the number. Here, “outside the predetermined range” includes both cases where the area is larger and smaller than the predetermined range. If the area is larger than the specified range, even if the number of detection ranks is reduced, the detection rank number is determined so that there is a low possibility of detection leakage in practice, so that fewer abnormal shadow candidates are detected. is there. If the area is smaller than the specified range, even if the number of detection ranks is increased, the possibility of increase in false detection is higher than the decrease in detection leakage. Thus, the number of detection ranks is determined. In this case, the abnormal shadow candidate detecting means detects a temporary candidate that can be an abnormal shadow candidate in the image by the same number as the determined detection rank number, and each detected temporary candidate. An index value calculation unit that calculates one or more types of index values indicating the characteristics of the temporary candidate, and a final candidate detection unit that detects a final candidate for an abnormal shadow based on the calculated index value. It is preferable to do. Here, as a specific example of “one or more index values indicating the characteristics of the temporary candidate”, information indicating the shape, density distribution, size, edge state, and the like of the temporary candidate can be considered.

  In the abnormal shadow candidate detection device of the present invention or the computer on which the program of the present invention is mounted, the area calculation means calculates the area of the subject region in the medical radiographic image with the breast as the subject, and the detection level determination means The abnormal shadow candidate detection level is automatically determined based on the calculated area, and the abnormal shadow candidate detection means detects the abnormal shadow candidate in the image based on the determined detection level. It is possible to perform appropriate processing according to the difference in the size of the breast without individually adjusting the detection level, and the diagnostic efficiency and diagnostic accuracy of the diagnostician are improved.

  Further, when the area of the subject region is within a predetermined range, the detection level determining means determines the detection threshold as a predetermined reference value, and when the area is larger than the predetermined range, the area is A detection threshold value is determined so that an abnormal shadow candidate can be detected more easily than when it is within the predetermined range. If the area is smaller than the predetermined range, the area is within the predetermined range. If the detection threshold value is determined so that the abnormal shadow candidate is less likely to be detected than the case, when the area of the subject region is larger than the predetermined range, the detection of the abnormal shadow candidate is prevented. When the area of the subject area is smaller than the predetermined range, it is possible to prevent erroneous detection of abnormal shadow candidates, so that it is possible to appropriately detect abnormal shadow candidates according to the area of the subject area. Become It is effective.

  Further, when the area of the subject region is within a predetermined range, the detection level determining means determines the number of detection ranks to a predetermined reference value, and when the area is outside the predetermined range, the predetermined level is determined. The number of detection ranks is determined to be smaller than the reference value, the temporary candidate detection unit of the abnormal shadow candidate detection means detects the same number of temporary candidates as the number of detection ranks, and the index value calculation unit is detected. If each of the temporary candidates calculates one or more index values and the final candidate detection unit detects a final abnormal shadow candidate based on the calculated index values, the area of the subject region is predetermined. If it is larger than the range, there is practically no detection leakage, and if the area of the subject area is smaller than the predetermined range, false detection is prevented, and in any case, more than necessary temporary candidates Will not be extracted. It is possible to reduce the load, the processing efficiency is improved.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  The abnormal shadow candidate detection apparatus A according to the first embodiment of the present invention is a part of the computer-aided image diagnostic system 101 (see FIG. 1). As shown in FIG. 1, a computer-aided image diagnosis system 101 includes a storage unit 91 that stores image data (hereinafter referred to as original image data) P of a medical radiation image (hereinafter referred to as mammography) of a breast, and the original image data. Area calculation means 10 for calculating an area R of a subject region in an image by P (hereinafter referred to as an original image P), and detection threshold determination for determining detection threshold values T1 and T2 for abnormal shadow candidates based on the calculated area R Based on the means 21 and the detection threshold values T1 and T2 determined by the detection threshold value determining means 21, an abnormal shadow candidate in the original image P is detected, and image data representing the abnormal shadow candidate (hereinafter, local image data P1) An abnormal shadow candidate detecting means 30 for outputting a local image processing means 93 for emphasizing the abnormal shadow candidate represented by the local image data P1, and Whole image processing means 94 for performing image processing such as gradation processing and frequency processing on the image data P, and the whole image and local image processing represented by the original image data P ′ after image processing by the whole image processing means 94 Display means 92 for displaying the abnormal shadow candidate represented by the local image data P1 ′ after the image processing by the means 93 as a visible image. The area calculating unit 10 includes a subject region extracting unit 11 that extracts a subject region in the original image P, and an area counting unit 12 that counts the number of pixels of the detected subject region. The abnormal shadow candidate detecting unit 30 includes The morphological filter processing unit 31 that performs processing by the morphological filter on the original image data P, and the threshold value processing of the output value by the morphological filter based on the detection threshold values T1 and T2 determined by the detection threshold value determination unit 21 The threshold processing unit 32 detects a calcified shadow candidate.

  Here, the area calculating means 10, the detection threshold value determining means 21, the abnormal shadow candidate detecting means 30, the local image processing means 93, and the whole image processing means 94 are a server computer that performs image processing (hereinafter referred to as an image processing server). Has been implemented.

  The storage means 91 includes a database in which mammography image data is stored in association with information on a patient who is a subject and imaging conditions of the image, and the information can be searched as a search key. It is mounted on a server computer (hereinafter referred to as an image management server) having a large-capacity external storage device.

  The display means 92 is mounted on a personal computer (hereinafter referred to as client PC) having a high-definition liquid crystal display.

  An image processing server, an image management server, a client PC, and an image capturing / reading apparatus that captures and acquires mammography are communicable via a network.

  Next, processing performed by the computer-aided image diagnostic system 101 will be described. The storage unit 91 stores original image data P representing the mammography P (see FIG. 2) acquired by the image capturing / reading apparatus and transmitted to the image management server via the network. To do. Further, it is assumed that the original image data P is data of high brightness and high signal level.

  First, the area calculation unit 10 reads the original image data P from the storage unit 91. Specifically, a search request using information for specifying an image to be processed such as a patient ID and imaging date / time as a search key is transmitted to the image management server. In the image management server, the database management system of the storage means 91 searches the database according to this search request, and the original image data P is acquired and transmitted to the image processing server. In the image processing server, the received original image data P is stored in the temporary storage area. The area calculation means 10 reads the original image data P stored in this temporary storage area.

  Then, the subject region extraction unit 11 of the area calculating unit 10 creates a histogram of the original image P, and obtains the minimum luminance Smin1 excluding the background portion (low luminance portion) using the created histogram (see FIG. 3). ), The original image P is binarized using the minimum luminance Smin1 as a threshold value. That is, an area having a luminance higher than this threshold is extracted as a subject area. The subject area is a combined area of the breast area Pa and the pectoral muscle area Pb.

  Next, the area counting unit 12 of the area calculating means 10 counts the number of pixels in the extracted subject area and sets this as the area R of the subject area.

  The detection threshold value determination means 21 evaluates the area R of the subject region acquired by the area counting unit 12, and uses detection threshold values T1 and T2 corresponding to the values of the area R for output values Mo and Mgrad (described later) by the morphology filter processing. decide. Specifically, when the area R is equal to or greater than R1 and equal to or less than R2, the detection threshold values T1 and T2 are determined as predetermined reference values th1 and th2, respectively. When the area R is greater than R2, the detection threshold value T1. And T2 are determined as th1 / c1 and th2 / c2 (c1, c2> 1), respectively, and when the area R is smaller than R1, the detection threshold values T1 and T2 are set to th1 × c1, th2 × c2 (c1 , c2> 1).

The morphology filter processing unit 31 of the abnormal shadow candidate detection unit 30 performs an operation on the original image data P according to the following equation (1) to obtain an output value Mo.

  Here, Bi is M linear structural elements B (i = 1, 2, 3, and 4 in FIG. 4), which is set larger than the microcalcification shadow to be detected.

  In Expression (1), first, a process of searching for a minimum value in a predetermined width centered on the target pixel, which is determined according to the structural element B (erosion process; see FIG. 5B). Then, a process of searching for the maximum value within the predetermined width (dilation process; see FIG. 5A) is performed (opening process; see FIG. 5C). By this opening process, the minute calcified image that is a convex data change portion (image portion that fluctuates in a spatially narrow range) finer than the structural element B is removed. On the other hand, the elongated non-calcified shadow has a length longer than that of the structural element B, and a portion whose inclination (extending direction) coincides with any of the M structural elements Bi remains as it is (formula (1)). Calculation of the second term). Therefore, by removing the smoothed image (image from which the calcified shadow has been removed) obtained by the opening process from the original image data P, an image including only the candidates for the minute calcified image is obtained.

In addition, even in this case, a part of the non-calcified shadow having the same size as the micro-calcified shadow may remain. In such a case, the differential information based on the morphological operation of the following equation (2) is used. The non-calcification image contained in Mo of Formula (1) is further removed.

  In equation (2), the greater the output value Mgrad, the greater the possibility of a microcalcification shadow.

  Next, the threshold processing unit 32, when the output value Mo of the morphology filter processing unit 31 is larger than the threshold T1 and the output value Mgrad is larger than the threshold T2 based on the thresholds T1 and T2 determined by the detection threshold determining unit 21 In this case, regions having the output values Mo and Mgrad are detected as microcalcification shadow candidates. In other cases, the regions having the output values Mo and Mgrad are not detected as tumor shadow candidates.

  Data P1 representing the detected microcalcification shadow candidate is input to the local image processing means 93, and the local image processing means 93 performs image processing so that the abnormal shadow candidate is emphasized with respect to the input local image data P1. Then, the processed local image data P1 ′ is output.

  Further, the whole image processing means 94 reads the original image data P and performs image processing such as gradation processing and frequency processing on the original image data P.

  The local image data P1 ′ obtained by processing by the local image processing means 93 and the original image data P ′ obtained by processing by the whole image processing means 94 are input to the display means 92, and the display means 92 While displaying the image based on the original image data P ′, only the image portion of the abnormal shadow candidate is replaced with the image based on the local image data P1 ′ processed by the local image processing means 93 and displayed. As a result, the display unit 92 displays the image of the abnormal shadow candidate that is emphasized more than the entire image together with the entire image, and is used for diagnosis by a doctor such as a doctor.

  As described above, in the abnormal shadow candidate detecting apparatus A according to the first embodiment of the present invention, the area calculating unit 10 calculates the area R of the subject region in the original image P, and the detection threshold determining unit 21 calculates An abnormal shadow candidate detection threshold is automatically determined based on the determined area R, and the abnormal shadow candidate detection means 30 performs an abnormal shadow candidate detection process based on the determined detection threshold. Here, based on the values th1 and th2 of the detection threshold values T1 and T2 when the area R is equal to or greater than R1 and equal to or less than R2, if the area R is larger than R2, the subject (mammary) represented in the original image P ) Is estimated to be a fatty breast that is relatively easy to detect abnormal shadow candidates, and the detection threshold values T1 and T2 are smaller than th1 and th2, respectively, th1 / c1 and th2 / c2 (c1, c2> 1) Set. As a result, the detection sensitivity is increased, so that the detection of abnormal shadow candidates can be prevented. On the other hand, when the area R is smaller than R1, it is estimated that the subject (breast) represented in the original image P is a high-density breast type breast that is relatively difficult to detect abnormal shadow candidates, and detection thresholds T1 and T2 Are set to values th1 × c1 and th2 × c2 (c1, c2> 1) larger than th1 and th2, respectively. As a result, the detection sensitivity is lowered, and it is possible to prevent erroneous detection of other structures as abnormal shadow candidates. Thus, it is possible for the diagnostician to perform appropriate processing according to the difference in the size of the breast without individually adjusting the detection level, and the diagnostic efficiency and diagnostic accuracy of the diagnostician are improved.

  The abnormal shadow candidate detection apparatus B according to the second embodiment of the present invention forms part of the computer-aided image diagnostic system 102 (see FIG. 6). As shown in FIG. 6, the computer-aided image diagnosis system 101 includes a storage unit 91 that stores image data (hereinafter referred to as original image data) P of a medical radiation image (hereinafter referred to as mammography) of a breast, and the original image data. Area calculation means 10 for calculating the area R of the subject region in the image by P (hereinafter referred to as the original image P), and detection rank determination means 22 for determining the detection rank T of the abnormal shadow candidate based on the calculated area R And an abnormal shadow candidate detecting means 30 for detecting an abnormal shadow candidate in the original image P based on the detection rank T determined by the detection rank determining means 22 and outputting image data P2 representing the abnormal shadow candidate; A region of interest setting means 95 for setting a rectangular region of interest W composed of this abnormal shadow candidate and its neighboring region, and an image (hereinafter referred to as “region of interest”). The) of the region of interest image W and displays with the original image P, and a display unit 92 for displaying the index value for the detected abnormal shadow candidates. The area calculating unit 10 includes a subject region extracting unit 11 that extracts a subject region in the original image P, and an area counting unit 12 that counts the number of pixels of the detected subject region. The abnormal shadow candidate detecting unit 30 includes A temporary candidate detection unit 33 that detects the same number of temporary candidates as the detection rank T determined by the detection rank determination means 22 as potential candidates for abnormal shadows in the original image P by processing using an iris filter; For each provisional candidate, the index value calculation unit 34 calculates an index value indicating the characteristics of the provisional candidate, and the final candidate detection unit 35 detects a final candidate for an abnormal shadow based on the calculated index value. ing.

  Here, the area calculating means 10, the detection rank determining means 22, the abnormal shadow candidate detecting means 30, and the region of interest setting means 95 are mounted on a server computer (hereinafter referred to as an image processing server) that performs image processing.

  The storage means 91 includes a database in which mammography image data is stored in association with information on a patient who is a subject and imaging conditions of the image, and the information can be searched as a search key. It is mounted on a server computer (hereinafter referred to as an image management server) having a large-capacity external storage device.

  The display means 92 is mounted on a personal computer (hereinafter referred to as client PC) having a high-definition liquid crystal display.

  An image processing server, an image management server, a client PC, and an image capturing / reading apparatus that captures and acquires mammography are communicable via a network.

  Next, processing performed by the computer-aided image diagnostic system 102 will be described focusing on differences from the system 101. It is assumed that the original image data P is stored in the storage unit 91 as in the system 101. Further, it is assumed that the original image data P is data of high density and high signal level.

  First, the area calculation unit 10 acquires the area R of the subject region in the same manner as the system 101.

  The detection rank determination unit 22 evaluates the area R of the subject region acquired by the area calculation unit 10 and determines the detection rank T in the iris filter process according to the value of the area R. For example, when the area R is equal to or greater than R1 and equal to or less than R2, the detection rank T is set to 10, and when the area R is greater than R2 or the area R is smaller than R1, the detection rank T is set to 5. Here, the detection rank T is the number of provisional candidates detected by the provisional candidate detection unit 33 of the abnormal shadow candidate detection means 30.

  The temporary candidate detection unit 33 of the abnormal shadow candidate detection unit 30 performs the following processing on the original image data P.

First, with respect to all the pixels constituting the target image, the orientation θ of the gradient vector of the pixel value of the image data is obtained for each pixel m based on the calculation formula shown in the following formula (3).

  Here, as shown in FIG. 7, f1 to f16 are pixel values (image data) corresponding to the pixels on the outer periphery of the mask of 5 vertical pixels × 5 horizontal pixels around the pixel j.

  Next, as shown in FIG. 8, consider radial lines in M types of directions every 2π / M degrees (in FIG. 8, 32 directions every 11.25 degrees are illustrated) with the pixel of interest at the center. For each line, the output value up to the pixel at which the maximum gradient vector concentration is obtained among the pixels on the line is defined as the concentration Cimax in that direction, and the concentration Cimax is averaged in all directions. Then, the degree of concentration C of the gradient vector group for the target pixel is calculated.

Specifically, first, the degree of concentration Ci (n) obtained from the target pixel to the nth pixel on the i-th radial line is obtained by the following equation (4).

  Here, θij is an angle formed by a line segment connecting the j-th pixel on the i-th radial line and the target pixel, and the gradient vector calculated by Equation (3) in the j-th pixel.

  Therefore, for each of the natural numbers n between Rmin and Rmax, the expression (4) is obtained by setting the target pixel as the starting point and the end point as the nth pixel from the target pixel on the i-th radial line. The average value of the components in the pixel direction of the gradient vector in each pixel is calculated as the degree of concentration Ci (n). Here, Rmin and Rmax are the minimum value and the maximum value of the radius of the tumor shadow to be extracted, respectively.

Next, the degree of concentration C of the gradient vector group is calculated by the following equations (5) and (6).

Here, Cimax in Expression (5) is the maximum value of the concentration degree Ci (n) for each radial line obtained in Expression (4), and therefore the concentration degree Ci (n) from the target pixel is the maximum value. The region up to the pixel becomes the candidate region for the tumor shadow in the direction of the line. Then, by calculating equation (5) for all radial lines to determine the area of the tumor shadow on each line, by connecting the area of the tumor shadow on each line with a straight line or a non-linear curve between adjacent lines, The shape of the outer peripheral edge of the region that can be a candidate for the shadow of the tumor can be specified. Note that the coordinates ([x], [y]) on the i-th line and the n-th pixel from the target pixel are expressed by the following formula (7), where the coordinate of the target pixel is (k, l). ).

However, [x] and [y] are maximum integers that do not exceed x and y.

  Further, in Expression (6), the maximum concentration value Cimax given in Expression (5) in this region is averaged in all directions of the radial line (in Expression (6), the case of 32 directions is exemplified). Ask for. This obtained value is the output value I of the iris filter process.

  As described above, the temporary candidate detection unit 33 calculates the iris filter output value I based on the original image P, and then rearranges the output values I in the respective pixels in descending order. The number of detection ranks determined by the determination unit 22 is detected as temporary abnormal shadow candidates.

Next, the index value calculation unit 34 calculates the following feature amount for each detected temporary candidate. First, as the first feature amount, since the contour of the tumor shadow is a shape close to a circle, the circularity Sp (Spreadness) is used. As shown in FIG. 9, an area A and a center of gravity AO of a detected tumor shadow (solid line portion in the figure) and a center of gravity AO are obtained, and a virtual circle having a radius R and having an area equivalent to the area A around the center of gravity AO. Assuming (broken line part in the figure), the degree of circularity is calculated as the occupation ratio of the temporary candidates included in the virtual circle with respect to the area A. That is, assuming that the area of the portion where the virtual circle and the candidate region overlap is A ′, the circularity is calculated by the following equation (8).

Next, the following three feature values are used as the feature values inside the candidate area. That is, a histogram of the temporary candidate density value S is created, and the frequency of the density value S is P (S). Based on the density value S and the frequency P (S), the second value representing the variance var is obtained from the following equation. The feature amount (9), the third feature amount (10) representing the contrast con, and the fourth feature amount (11) representing the angular moment asm are calculated.

  In addition, edge information such as dispersion, bias, correlation value, moment, and entropy for a simultaneous generation matrix generated based on an IFED image (Iris Filter EDge) can also be used as a feature amount (Japanese Patent Laid-Open No. 2003-115041, etc.) reference). Hereinafter, a case where nine feature amounts including these are calculated will be described as an example.

  The final candidate detection unit 35 calculates the Mahalanobis distance based on these feature amounts calculated by the index value calculation unit 34, makes a determination using the likelihood ratio, and detects a final candidate for an abnormal shadow.

First, the Mahalanobis distance Dm1 from the pattern class w1 indicating the non-malignant shadow and the Mahalanobis distance Dm2 from the pattern class w2 indicating the malignant shadow, which are experimentally obtained in advance, are calculated by the following equation (12).

  The feature amounts calculated by the index value calculation unit 34 correspond to the above x1 to x9, respectively, and represent a nine-dimensional space (x1, x2, x3,..., X9). The Mahalanobis distance between the temporary candidate pattern expressed in the nine-dimensional pattern space and the non-malignant shadow pattern is Dm1, and similarly, the Mahalanobis distance between the malignant shadow pattern is Dm2.

  The non-malignant shadow pattern and the malignant shadow pattern are pattern spaces defined by vectors x for each non-malignant shadow and each malignant shadow, which are set based on the results of an experimental investigation of a large number of abnormal shadow candidates in advance. Means. For example, the pattern class w1 formed by the average of the vector x for the non-malignant shadow and the pattern class w2 formed by the average of the vector x for the malignant shadow are shown.

  Here, the final candidate detection unit 35 calculates the Mahalanobis distance using a non-malignant shadow pattern and a malignant shadow pattern in a preset subject area.

  Next, when the candidate area is a malignant shadow, the Mahalanobis distance from the pattern class of the malignant shadow is close (Dm2 indicates a low value), and the Mahalanobis distance from the pattern class of the non-malignant shadow tends to vary. When the candidate area is a non-malignant shadow, the Mahalanobis distance from the pattern class of the non-malignant shadow is close (Dm1 indicates a low value), and the Mahalanobis distance from the pattern class of the malignant shadow tends to vary. Therefore, a likelihood ratio that can significantly distinguish a malignant shadow and a non-malignant shadow according to this tendency is calculated for each candidate region.

  The likelihood ratio is expressed by Dm1 / Dm2, and indicates the inclination on the coordinate plane of FIG. That is, it can be determined that the likelihood ratio is higher as the likelihood ratio is higher, and the possibility that the non-malignant shadow is higher as the likelihood ratio is smaller. For example, when the likelihood ratio is 2 or more, the threshold is set to 2. When it is less than 2, it is determined to be non-malignant.

  As described above, the final candidate detection unit 35 evaluates the feature amount of each temporary candidate, detects what is determined to be malignant as a final candidate for abnormal shadow, and displays image data P2 representing the final candidate and its index Outputs the value (feature value).

  The extracted data P2 representing the final candidate for abnormal shadow is input to the region of interest setting means 95. The region-of-interest setting unit 95 sets a local region having a predetermined shape (for example, a rectangle or a circle) including the final candidate for an abnormal shadow and its neighboring region as the region-of-interest image W. When there are a plurality of detected abnormal shadow final candidates, a plurality of the region-of-interest images W are also set. The region-of-interest setting means 95 outputs image data W representing the set region-of-interest image W.

  The display unit 92 includes the original image data P temporarily stored in the image processing server, the index value (feature value) K for each of the abnormal shadow final candidates output by the final candidate detection unit 35, the region of interest. Image data W representing the region-of-interest image W output by the setting means 95 is acquired, and the original image P, the region-of-interest image W and the index value K are simultaneously displayed on the screen as shown in FIGS. 11 (1) and (2). It is used for diagnosis by an interpreter such as a doctor.

  As described above, in the abnormal shadow candidate detecting apparatus B according to the second embodiment of the present invention, the area calculating unit 10 calculates the area R of the subject region in the original image S, and the detection rank determining unit 22 calculates Based on the determined area R, the abnormal shadow candidate detection rank T is automatically determined, and the temporary candidate detection unit 33 of the abnormal shadow candidate detection means 30 detects the abnormal shadow temporary candidate based on the determined T. I do. Here, based on the detection rank T (= 10) when the area R is R1 or more and R2 or less, if the area R is larger than R2, the subject (breast) shown in the original image S is abnormal. Even if it is estimated that the shadow candidate is a relatively easy fat breast and the detection rank T is reduced, the detection rank T is smaller than the reference because the possibility of detection leakage is low. = 5). On the other hand, if the area R is smaller than R1, the subject (breast) represented in the original image S is estimated to be a high-density breast type breast that is relatively difficult to detect abnormal shadow candidates, and the detection rank T is increased. However, since the possibility of an increase in false detection is higher than the decrease in detection leakage, the detection rank T is set to a value (= 5) smaller than the reference in order to prevent an increase in false detection. Thus, it is possible for the diagnostician to perform appropriate processing according to the difference in the size of the breast without individually adjusting the detection level, and the diagnostic efficiency and diagnostic accuracy of the diagnostician are improved. As a result, the temporary candidate detection unit 33 does not extract temporary candidates more than necessary, so that the processing load of the subsequent index value calculation unit 34 and the final candidate detection unit 35 can be reduced, and the processing efficiency is improved. improves.

  Next, an example will be described in which the present embodiment is modified without departing from the technical idea of the present invention.

  First, the combination of the specific content of the detection level and the specific content of the abnormal shadow candidate detection process is not limited to the above two patterns. For example, the detection threshold value determining unit 21 determines a detection threshold value for the output value of the iris filter according to the area of the subject region, and the abnormal shadow candidate detecting unit 30 uses the iris filter based on the detection threshold value to detect a mass shadow candidate. The detection rank determining means 22 determines the detection rank of the micro calcified shadow according to the area of the subject region, and the abnormal shadow candidate detecting means 30 is based on this detection rank. A temporary candidate for calcification shadow may be detected, and the final candidate may be detected based on a feature amount such as calcification density.

  In the present embodiment, the area of the subject region is classified into three stages, and the detection threshold and the detection rank are determined for each stage. For example, using the reference table that performs the conversion shown in FIG. The threshold value and the detection rank may be changed continuously to some extent according to the area.

  In the present embodiment, each means is distributed and implemented in a plurality of computers, but some or all of these may be implemented in a single computer. This is useful for small systems.

1 is a block diagram showing a configuration of a computer-aided image diagnostic system including an abnormal shadow candidate detection apparatus A according to a first embodiment of the present invention. The figure which shows the side image of the breast by the original image data P The figure explaining the method to detect a to-be-photographed area | region based on the histogram of the original image P A diagram showing the concept of structural elements used in morphological operation processing A graph explaining the basic operation of morphological operation processing The block diagram which shows the structure of the computer-aided image diagnostic system containing the abnormal shadow candidate detection apparatus B used as the 2nd Embodiment of this invention. The figure which shows the mask which calculates the gradient vector in an iris filter process Conceptual diagram showing an iris filter set so that its contour shape changes adaptively The figure which shows the virtual circle which has an area equivalent to the area A of a candidate area | region Diagram explaining likelihood ratio The figure which shows an example of the screen displayed on a screen by the display means in the 2nd Embodiment of this invention. The figure which shows an example of the conversion table which changes a detection threshold value (a) and a detection rank (b) substantially continuously according to the area of a to-be-photographed area | region.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Area calculation means 11 Subject area extraction part 12 Area count part 21 Detection threshold value determination means 22 Detection rank determination means 30 Abnormal shadow candidate detection means 31 Morphological filter processing part 32 Threshold processing part 33 Temporary candidate detection part 34 Index value calculation part 35 Final Candidate detection unit 91 Storage unit 92 Display unit 93 Local image processing unit 94 Whole image processing unit 95 Region of interest setting unit 101 Computer-aided image diagnosis system 102 including the first embodiment of the present invention The second embodiment of the present invention Computer-aided image diagnosis system A including abnormal shadow candidate detection device B according to the first embodiment of the present invention Abnormal shadow candidate detection device according to the second embodiment of the present invention

Claims (6)

In an apparatus provided with an abnormal shadow candidate detecting means for detecting an abnormal shadow candidate in a medical radiographic image having a breast as a subject,
Area calculating means for calculating the area of the subject region in the image;
Detection level determining means for determining a detection level of the abnormal shadow candidate based on the area;
The abnormal shadow candidate detection apparatus, wherein the abnormal shadow candidate detection means detects the abnormal shadow candidate based on the determined detection level. The detection level is a detection threshold of the abnormal shadow candidate;
The detection level determining means determines a detection threshold value as a predetermined reference value when the area is within a predetermined range, and when the area is larger than the predetermined range, the area is equal to the predetermined range. When the detection threshold is determined so that the abnormal shadow candidate is more easily detected than when it is within the range, and when the area is smaller than the predetermined range, the area is within the predetermined range And determining the detection threshold so that the abnormal shadow candidate is less likely to be detected,
The abnormal shadow candidate detection apparatus according to claim 1, wherein the abnormal shadow candidate detection unit detects the abnormal shadow candidate based on the determined detection threshold. The detection level is the number of detection ranks of the abnormal shadow candidate;
The detection level determining means determines the number of detection ranks to a predetermined reference value when the area is within a predetermined range, and the predetermined reference value when the area is outside the predetermined range. Determining the number of detection ranks to a value smaller than
The abnormal shadow candidate detecting means is
A temporary candidate detection unit that detects the number of temporary candidates that can be the abnormal shadow candidates in the image by the same number as the determined number of detection ranks;
An index value calculation unit that calculates one or more types of index values indicating the characteristics of the temporary candidates for each of the detected temporary candidates;
The abnormal shadow candidate detection apparatus according to claim 1, further comprising a final candidate detection unit that detects a final candidate for the abnormal shadow based on the calculated index value. In a program for causing a computer to function as an abnormal shadow candidate detecting means for detecting an abnormal shadow candidate in a medical radiation image having a breast as a subject,
The computer,
Area calculating means for calculating the area of the subject region in the image;
Function as detection level determination means for determining the detection level of the abnormal shadow candidate based on the area;
A program for causing the abnormal shadow candidate detection means to function so as to detect the abnormal shadow candidate based on the determined detection level. The detection level is a detection threshold of the abnormal shadow candidate;
When the area is within a predetermined range, the detection level determining means determines a detection threshold to be a predetermined reference value, and when the area is larger than the predetermined range, the area is the predetermined range. When the detection threshold is determined so that the abnormal shadow candidate can be detected more easily than when the area is within the range, and when the area is smaller than the predetermined range, the area is within the predetermined range The detection threshold is determined so that the abnormal shadow candidate is less likely to be detected,
5. The program according to claim 4, wherein the abnormal shadow candidate detection unit is configured to detect the abnormal shadow candidate based on the determined detection threshold. The detection level is the number of detection ranks of the abnormal shadow candidate;
When the area is within a predetermined range, the detection level determination means determines the number of detection ranks to be a predetermined reference value, and when the area is out of the predetermined range, the predetermined reference value The detection rank number is determined to a value smaller than
The abnormal shadow candidate detection means,
A temporary candidate detection unit that detects the number of temporary candidates that can be the abnormal shadow candidates in the image by the same number as the determined number of detection ranks;
An index value calculation unit that calculates one or more types of index values indicating the characteristics of the temporary candidates for each of the detected temporary candidates;
5. The program according to claim 4, wherein the program is made to function as a final candidate detection unit that detects a final candidate of the abnormal shadow based on the calculated index value.

Images