JP2001222703A - Method and device for detecting and processing abnormal shadow candidate - Google Patents

Abstract

PROBLEM TO BE SOLVED: To adapt the detection level of an abnormal shadow candidate to a user's detection request level in a detection processor for an abnormal shadow candidate. SOLUTION: This abnormal shadow candidate detection processor 10 performing detection processing in which a shadow is detected as an abnormal shadow candidate when an output value K is larger than a threshold stored in a storing means 2 and the shadow is not detected as the abnormal shadow candidate when the output value K is equal to or smaller than the threshold, is provided with a threshold changing means 3 changing the threshold.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a method and apparatus for detecting abnormal shadow candidates such as tumor shadow candidates and microcalcification shadow candidates in radiographic images.

[0002]

2. Description of the Related Art Conventionally, image processing such as gradation processing and frequency processing has been performed on an image signal representing an image obtained by various image acquisition methods to improve image observing and reading performance. Have been done. Especially in the field of medical images such as radiation images of the human body, specialists such as doctors need to accurately diagnose the presence or absence of a patient's disease or injury based on the obtained images. Image processing for improving the image interpretation performance is essential.

In such image processing, the entire image may be processed, but if the purpose of inspection or diagnosis is clear to some extent, only a desired image portion suitable for the purpose is selectively selected. May be emphasized.

Normally, such an image portion is selected manually by an observer while observing an original image before image processing is performed, if necessary. The specified range is determined by the experience of the observer and the level of image reading ability, and may not be objective.

For example, in a radiographic image taken for the purpose of examining breast cancer, it is necessary to extract a tumor shadow or a microcalcification shadow, which is a characteristic of a cancerous portion, from the radiographic image. It is not always possible to specify a range such as a tumor shadow. For this reason, it is required to accurately detect abnormal shadows such as tumor shadows without depending on the skill of the observer.

A computer-assisted image diagnostic process (Japanese Patent Application Laid-Open No. 8-294479) has been proposed as one of the devices that meets this demand. This computer-assisted image diagnosis processing is performed by an iris filter processing (hereinafter, referred to as an iris filter calculation in the present specification) for automatically detecting the above-mentioned tumor shadow candidate using a computer based on the morphological features and the like in the image. ) And morphology calculation processing for automatically detecting the above-mentioned microcalcification shadow candidates, which facilitates objective extraction of these abnormal shadow candidates.

Here, the iris filter processing is performed by Obata et al. Of Tokyo University of Agriculture and Technology, "Detection of tumor shadow in DR image (iris filter)" (Transactions of IEICE D-II Vol.J7
5-D-II No.3 P663-670 March 1992), which has been studied as an effective method to detect candidates for tumor shadows, which is one of the characteristic morphologies in breast cancer. However, the target image is not limited to a tumor shadow in such a mammogram, and an image in which the gradient of an image signal (density or the like) representing the image is concentrated is not limited to any image portion. This is a detection process that can be applied.

Hereinafter, an outline of the process of detecting a tumor shadow candidate using the iris filter will be described.

For example, in a radiographic image (an image represented by a high-density high-signal level image signal) on an X-ray film, it is generally known that a tumor shadow has a slightly lower density value than a surrounding image portion. The density value distribution has a density value gradient such that the density value decreases from the substantially circular periphery toward the center. Therefore, in the tumor shadow, a local gradient of the density value is recognized, and the gradient line is concentrated toward the center of the tumor.

The iris filter calculates the gradient of the image signal represented by the density value as a gradient vector, and outputs the degree of concentration of the gradient vector. Based on this, a candidate for a tumor shadow is detected.

That is, for example, in the mammogram P as shown in FIG. 5A, the gradient vector at an arbitrary pixel within the tumor shadow PJ is directed to the vicinity of the center of the tumor shadow as shown in FIG. In the case of an elongated shadow PK such as a mammary gland, the gradient vector does not concentrate on a specific point as shown in FIG. If the extracted region is extracted, it becomes a candidate considered as a tumor shadow.
It should be noted that, as shown in FIG. 4D, with regard to a shadow PL in which elongated shadows such as a mammary gland intersect, the gradient vector tends to concentrate on a specific point, and may be erroneously detected as a candidate for an abnormal shadow. The above is the basic concept of the iris filter processing. The specific algorithm steps are shown below.

(Step 1) Calculation of Gradient Vector For all the pixels constituting the target image, the direction θ of the gradient vector of the image data is calculated for each pixel j based on the following equation (1). Ask.

[0013]

(Equation 1) Here, as shown in FIG. 6, f _{1 to} f ₁₆ are pixel values (image data) corresponding to pixels on the outer periphery of the mask of 5 × 5 pixels centering on the pixel j.

(Step 2) Calculation of Concentration of Gradient Vector Next, for all the pixels constituting the target image,
For each pixel, the degree of concentration C of the gradient vector having that pixel as the pixel of interest is calculated according to the following equation (2).

[0015]

(Equation 2) Here, N is the number of pixels existing in a circle having a radius R around the pixel of interest, θj is a straight line connecting the pixel of interest and each pixel j in the circle, and the above equation (1) for each pixel j (See FIG. 7). Therefore, the degree of concentration C represented by the above equation (2) becomes a large value when the direction of the gradient vector of each pixel j is concentrated on the target pixel.

By the way, the gradient vector of each pixel j in the vicinity of the tumor shadow is determined regardless of the contrast of the tumor shadow.
Since the pixel of interest is directed substantially toward the center of the tumor shadow, the pixel of interest having a large value of the degree of concentration C can be said to be the pixel at the center of the tumor shadow. On the other hand, in the shadow of a linear pattern such as a blood vessel, the value of the degree of concentration C is small because the direction of the gradient vector is biased in a certain direction. Therefore, the value of the degree of concentration C with respect to the pixel of interest is calculated for each of all the pixels constituting the image, and whether or not the value of the degree of concentration C exceeds a predetermined threshold value is evaluated. Can be detected. In other words, this filter is less susceptible to blood vessels and mammary glands than a normal differential filter,
It has the feature that tumor shadows can be detected efficiently.

Further, in the actual processing, in order to achieve a detection power independent of the size and shape of the tumor, a device for adaptively changing the size and shape of the filter is used. FIG. 8 shows the filter. This filter differs from the filter shown in FIG.
The evaluation of the degree of concentration is performed only by pixels on a radial line in M kinds of directions at every / M degrees (in FIG. 8, 32 directions at every 11.25 degrees are exemplified).

Here, the coordinates ([x], [y]) of the n-th pixel on the i-th line and from the pixel of interest are as follows if the coordinates of the pixel of interest are (k, l): Equation (3),
Given by (4).

[0019]

(Equation 3) Here, [x] and [y] are the maximum integers not exceeding x and y.

Further, the output value up to the pixel at which the maximum concentration is obtained for each line on the radial line is defined as the concentration Cimax in the direction, and the concentration Cimax is averaged in all directions to obtain the target value. The degree of concentration C of the gradient vector group for the pixel is set.

More specifically, first, the degree of concentration Ci (n) obtained from the target pixel to the n-th pixel on the i-th radial line is obtained by the following equation (5).

[0022]

(Equation 4) That is, equation (5) is based on the target pixel as the starting point and the end point as R
The concentration Ci (n) is calculated within the range from min to Rmax. Here, Rmin and Rmax are the minimum and maximum values of the radius of the tumor shadow to be extracted.

Next, the concentration C of the gradient vector group is calculated by the following equations (6) and (7).

[0024]

(Equation 5) Here, since Cimax in Expression (6) is the maximum value of the concentration Ci (n) for each radial line obtained in Expression (5), the concentration Ci (n) is the maximum value from the pixel of interest. The region up to the pixel becomes a candidate region of the tumor shadow in the direction of the line.

Equation (6) is calculated for all the radial lines to determine the area of the tumor shadow on each line, and the area of the tumor shadow on each line is connected between adjacent lines by a straight line or a non-linear curve. The shape of the outer peripheral edge of a region that can be a candidate for a tumor shadow can be specified.

In the equation (7), the maximum value Cimax of the degree of concentration given by the equation (6) in this area is calculated for all directions of the radial line (the equation (7) exemplifies the case of 32 directions). Find the average value. The obtained value is the output value I of the iris filter processing, and this output value I is compared with a predetermined fixed threshold value T1 suitable for determining whether or not the shadow is a tumor, and I ≧ T1 If (or I> T1), the area around the target pixel is an abnormal shadow candidate (tumor shadow candidate), and I <T1 (or I ≦ T1).
If 1), it is determined that it is not a tumor shadow candidate.

The manner in which the size and shape of the area for evaluating the degree of concentration C of the gradient vector group in equation (7) adaptively changes in accordance with the distribution of the gradient vector is enlarged in accordance with the brightness of the outside world. Since the iris of the human eye that shrinks is similar to the appearance, the above-described method of detecting a candidate region of a tumor shadow using the degree of concentration of the gradient vector is referred to as iris filter processing. I have.

The above calculation of the degree of concentration Ci (n) may be performed by using the following equation (5 ') instead of the equation (5).

[0029]

(Equation 6) That is, equation (5 ') calculates the degree of concentration Ci (n) from the pixel corresponding to the minimum value Rmin of the radius of the tumor shadow to be extracted to the end point within the range from Rmin to Rmax. .

According to the above-described steps, the iris filter can effectively detect only a tumor shadow of a desired size from a radiographic image.

By the way, in general, the shadow of a malignant mass has the following morphological characteristics: 1) its edge is irregular; 2) it has a shape close to a circle; and 3) the inside has an uneven density distribution. I have.

Therefore, for a more definitive diagnosis, the image signal of the abnormal shadow candidate obtained by the comparison processing between the iris filter output value I and the threshold value T1 described above is subjected to a shape considering these characteristics. The determination may be further performed. The feature amounts used here are the spread degree (Spreadness), the elongation (Elongation), the roughness of the edge (Roughness), the circularity (Circularity), and the inner roughness (Entropy). The final determination as to whether or not the candidate is a tumor shadow may be made by comparing with another predetermined threshold T2 set in advance. The additional detection processing such as the detection processing based on this morphological feature is not the iris filter processing itself,
Since this is a process for detecting an abnormal shadow candidate by additionally applying the process to the abnormal shadow candidate detection process by the iris filter process, these additional detection processes are also included in the iris filter process, and are described below. Processing ".

On the other hand, the morphological operation processing is a method of detecting a candidate for a microcalcification image, which is a characteristic form of breast cancer, together with a tumor shadow, and uses a multi-scale λ and a structural element (mask) B. It is effective in extracting the calcified image itself, (2) is not easily affected by complicated background information, and (3) the extracted calcified image is not distorted. In other words, this method is different from general differential processing in that the size, shape,
It is possible to perform detection while better maintaining geometric information such as a density distribution. The outline is described below.

(Basic Arithmetic Operation of Morphology) Morphological arithmetic processing is generally developed as set theory in an N-dimensional space. However, for intuitive understanding, a description will be given of a two-dimensional grayscale image.

The point of coordinates (x, y) is represented by a density value f
It is regarded as a space having a height corresponding to (x, y). Here, the density value f (x, y) is a high-luminance high-signal level signal which becomes a larger image signal as the density is lower (the brightness is higher when displayed on a CRT).

First, for simplicity, consider a one-dimensional function f (x) corresponding to the cross section. The structural element g used in the morphological operation is represented by a symmetric function symmetric about the origin as shown in the following equation (8).

(Equation 7) And the value is 0 in the domain, and the domain is represented by the following equation (9).

[0037]

(Equation 8) At this time, the basic form of the morphological operation is as shown in Equation (10).
As shown in (13), the calculation becomes very simple.

[0038]

(Equation 9) That is, the dilation process is a process of searching for a maximum value within a range of ± m (a value determined according to the structural element B) around the target pixel (FIG. 9).
On the other hand, the erosion process is a process of searching for a minimum value within a width of ± m around the pixel of interest (see FIG. 9B). The opening processing corresponds to searching for the maximum value after searching for the minimum value, and the closing processing corresponds to searching for the minimum value after searching for the maximum value.
The opening process is equivalent to smoothing the density curve f (x) from the low brightness side and removing a convex density variation portion (a portion higher in brightness than the surrounding portion) that fluctuates in a space narrower than the mask size 2 m. (See FIG. 9C).
On the other hand, in the closing processing, the density curve f
This corresponds to smoothing (x) and removing a concave density variation portion (a portion having a lower luminance than the surrounding portion) which fluctuates in a space narrower than the mask size of 2 m (FIG. 9D).
reference).

Here, in the case of a signal of a high-density high signal level which becomes a larger value as the density becomes higher, the density value f
Since the magnitude relationship is reversed with respect to the case where the image signal value of (x) is at the high luminance and high signal level, the dilation processing on the signal at the high density and high signal level is performed by the erosion processing at the high luminance and high signal level (see FIG. B)), the erosion processing for the signal with the high density and high signal level is performed by the dilation processing at the high luminance and high signal level (FIG. 9).
(A)), the opening process for a signal with a high density and high signal level matches the closing process for a signal with a high luminance and high signal level (FIG. 9D), , And the opening process at the high luminance and high signal level (FIG. 9C). In this section, a case of an image signal (luminance value) at a high luminance and high signal level will be described.

(Application to Calculated Shading Detection) For the detection of calcified shadows, a difference method of subtracting a smoothed image from an original image can be considered. It is difficult to distinguish between calcified shadows and elongated non-calcified shadows (e.g., mammary glands, blood vessels, and mammary gland supporting tissues) using a simple smoothing method. Obata et al. We propose a morphological operation represented by equation (14) ("Extraction of microcalcification image by morphological filter using multiple structural elements" IEICE Transactions on Electronics, Information and Communication Engineers D-II Vol.J75-D-II No.
7 P1170-1176, July 1992, "Basics of Morphology and Its Application to Mammogram Processing" MEDICALIMAGING TEC
HNOLOGY Vol.12 No.1 January 1994).

[0041]

(Equation 10) Here, Bi (i = 1, 2, 3, 4) are four linear structural elements B shown in FIG. If the structural element B is set to be larger than the calcified shadow to be detected, the calcification is a convex signal change portion (an image portion that fluctuates in a narrow spatial range) finer than the structural element B in the opening process. The image is removed. On the other hand, the elongated non-calcified shadow has a longer length than the structural element B, and if its inclination (extending direction) matches any of the four structural elements Bi, the opening process (the equation (14)) Even if the two-term operation is performed, it remains as it is. Therefore, by subtracting the smoothed image (image from which the calcified shadow has been removed) obtained by the opening process from the original image f, an image including only small calcified image candidates is obtained. This is the idea of equation (14).

As described above, in the case of a signal having a high density and a high signal level, the calcified shadow has a lower density value than the surrounding image portion, and the calcified shadow has a concave signal change with respect to the surrounding portion. Therefore, the closing process is applied instead of the opening process, and the expression (15) is applied instead of the expression (14).

[0043]

[Equation 11] However, even in this case, a part of the non-calcified shadow having the same size as the calcified shadow may remain. In such a case, the differential information based on the morphological operation of the following equation (16) is used. The non-calcified image included in P in equation (14) is further removed.

[0044]

(Equation 12) Here, as the value of Mgrad is larger, the possibility of calcification shadow is larger, and the calcification candidate image Cs can be obtained by the following equation (17).

[0045]

(Equation 13) Here, T1 and T2 are preset thresholds determined experimentally.

However, for non-calcified shadows different from the size of the calcified shadow, P in equation (14) and a predetermined threshold T1
In the case where a non-calcified shadow having the same size as the calcified shadow does not remain since it can be removed only by comparing with the condition (P (i, j) .Gtoreq.T1).

As described above, the morphological operation processing is described for the case of image data having a high luminance and a high signal level. ), The opening operation and the closing operation have the opposite relationship.

In the morphology calculation processing, additional detection processing similar to additional detection processing such as detection processing based on morphological features in the iris filter processing may be further applied. Such additional detection processing is also included in the morphology calculation processing, and is referred to as “processing based on morphology calculation processing”.

[0049]

In the abnormal shadow candidate detection processing such as the processing based on the iris filter processing and the processing based on the morphology operation processing, the image part (region) to be processed is an abnormal shadow. As a final determination as to whether or not the candidate is a candidate, comparison with a predetermined threshold is performed. For example, in the iris filter processing, the magnitude of the output value I of the iris filter processing is compared with a predetermined threshold value T1 or T2, and in the morphology calculation processing, the threshold value T
1 and T2, or a threshold value in a further additional detection process. These thresholds T1 and T2, etc., which are used as criteria for final judgment, are set based on experiments and accumulation of experience, and are very important values for objectively detecting abnormal shadow candidates. I have.

However, these thresholds are only a criterion for performing uniform detection, and deterministic abnormal shadows are 100%
Detecting and erroneously detecting a normal shadow that is not an abnormal shadow does not mean 0%. In other words, in order to prevent a deterministic abnormal shadow from being missed, it is necessary to set a small threshold. In this case, the abnormal shadow is suppressed from being missed, but the normal shadow is erroneously detected as an abnormal shadow. The rate also increases. On the contrary, in order to prevent erroneous detection of a normal shadow, it is necessary to set a large threshold. In this case, although erroneous detection of a normal shadow is suppressed,
On the other hand, detection omission of abnormal shadows also increases. This is generally R
As shown in FIG. 4, as shown in FIG. 4, when emphasis is placed on prevention of detection omission (emphasis on TP (True Possitive)), false detection (FP; False Possitive) also increases. Emphasis), abnormal shadow to be detected (TP)
(Detection of TP decreases).

Therefore, it has been desired that a uniform detection level can be set according to the user's preference and the purpose of detection. However, conventionally, such a change in the detection level has not been possible by the user. That is, the level to be detected as an abnormal shadow candidate by matching with clinical data may differ for each facility (user) such as a hospital or the like, and a mass examination in which a large amount of radiation images must be processed in a short time. It may be different depending on the purpose of use, such as whether to use it for a detailed examination or for a detailed examination of a patient who needs re-examination by a mass examination.

The present invention has been made in view of the above circumstances, and maintains the detection of an abnormal shadow candidate at a uniform detection level, while maintaining this detection level according to the user's preference and purpose of use. It is an object of the present invention to provide an abnormal shadow candidate detection processing method and apparatus that can be flexibly changed.

[0053]

SUMMARY OF THE INVENTION The method and apparatus for detecting an abnormal shadow candidate according to the present invention are capable of changing a detection processing control parameter such as a threshold value for determining whether or not to detect an abnormal shadow candidate, thereby enabling the abnormal shadow candidate to be changed. Can be changed arbitrarily to enable flexible setting of the detection level according to the user's preference and purpose of use.

That is, the abnormal shadow candidate detecting method according to the present invention is the abnormal shadow candidate detecting method for detecting an abnormal shadow candidate in a radiation image based on an image signal representing a radiation image of a subject. It is characterized in that a detection processing control parameter for controlling a detection level of a shadow candidate can be freely changed.

Here, the detection processing control parameter means a parameter whose detection level (detection result) of an abnormal shadow candidate can fluctuate by changing it, and is not limited to a “threshold” described later. It may include various coefficients used in the detection processing, a detection algorithm selected from among a plurality of coefficients, and the like. Further, a method of setting an area where the detectability is changed in the abnormal shadow candidate detection processing, or a method of setting the detectability for each area may be used.

As a process for detecting an abnormal shadow candidate, a process performed by performing a predetermined calculation process on an image signal and comparing the obtained output value with a predetermined threshold value set in advance is applied. Is preferred. In this case, a threshold value can be applied as a detection process control parameter, and it is desirable that the threshold value be changed very easily.

As the predetermined arithmetic processing, it is preferable to apply processing based on iris filter processing or processing based on morphological arithmetic processing. This is because it can be directly used as computer-aided image diagnosis processing. The processing based on the iris filter processing refers to processing including additional detection processing such as iris filter processing and detection processing based on morphological features, which is additionally applied to the iris filter processing. Similarly, the processing based on the morphological operation refers to a process including a morphological operation and a detection process additionally applied to the morphological operation.

When the processing based on the iris filter processing is applied, the tumor shadow candidate is applied as the abnormal shadow candidate, and the output value I of the iris filter processing and the threshold value T1 to be compared are used as the detection processing control parameters.
(If I ≧ T1 (or I> T1), it is determined that the region around the target pixel is a tumor shadow candidate, and if I <T1 (or I ≦ T1), it is not a tumor shadow candidate. The threshold value T1) to be used and the threshold value T2 in the detection processing additionally applied thereafter may be applied.

On the other hand, when the processing based on the morphology calculation processing is applied, the microcalcification shadow candidate is applied as the abnormal shadow candidate, and the output value of the morphology calculation processing and the comparison target are used as the detection processing control parameters. It is preferable to apply the threshold value T1, or T1 and T2 (Equation (17)), or a threshold value in a detection process to be additionally applied thereafter.

It is preferable that the detection processing control parameters be changed in accordance with the user's request level for detecting the abnormal shadow candidate. Therefore, not only the detection processing control parameter itself is directly changed, but also this parameter is indirectly changed, for example, by changing the detection level of the abnormal shadow candidate set corresponding to the detection processing control parameter. You may.

The detection request level of the abnormal shadow candidate is as follows.
For example, the degree of importance of TP (the degree of emphasis on prevention of detection omission) or the degree of emphasis of FP (the degree of emphasis on prevention of erroneous detection) in the above-described ROC characteristic is 0% to 10%.
It may be set continuously in the range of 0%,
The detection level may be simply set to two levels, that is, a detection level that emphasizes TP and a detection level that emphasizes FP.

The abnormal shadow candidate detection processing apparatus of the present invention comprises:
An apparatus for performing the method for detecting an abnormal shadow candidate according to the present invention, wherein the abnormal shadow candidate detecting unit detects an abnormal shadow candidate in the radiation image based on an image signal representing a radiation image of the subject. The abnormal shadow candidate detection processing device provided further comprises parameter changing means for freely changing a detection processing control parameter for controlling a detection level of the abnormal shadow candidate.

Here, as the predetermined arithmetic processing by the abnormal shadow candidate detecting means, it is preferable to apply processing based on iris filter processing, processing based on morphological arithmetic processing, or the like. When a tumor shadow candidate is applied as an abnormal shadow candidate and a process based on morphology calculation processing is applied, a microcalcification shadow candidate may be applied as the abnormal shadow candidate.

It is to be noted that a detection request level input means for receiving an input of a detection request level of an abnormal shadow candidate by a user is further provided, and a parameter changing means is provided in accordance with the detection request level input to the detection request level input means. It is preferable to adopt a configuration in which the detection processing control parameter is changed, because the detection request level is easier to grasp for the user than the detection processing control parameter, so that fine adjustment is possible.

[0065]

According to the method and apparatus for detecting an abnormal shadow candidate of the present invention, the detection processing control parameter for deciding whether or not to detect an abnormal shadow candidate can be changed. The detection level of the candidate can be easily changed, and the detection level of the abnormal shadow candidate can be flexibly changed according to the user's preference and the purpose of use.

[0066]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A specific embodiment of a method and apparatus for detecting abnormal shadow candidates according to the present invention will be described below with reference to the drawings.

FIG. 1 is a schematic block diagram showing an embodiment of an abnormal shadow candidate detection processing apparatus according to the present invention. FIG. 2 is a block diagram of a computer-aided image diagnosis apparatus using the abnormal shadow candidate detection processing apparatus shown in FIG. It is a block diagram showing an example.

The computer-aided image diagnostic apparatus 100 shown in the figure has a storage means 20 for storing an input image signal (hereinafter, referred to as an entire image signal) S, and reads out the entire image signal S stored in the storage means 20. , A whole image processing means 30 for performing image processing such as gradation processing and frequency processing, and a whole image signal S stored in the storage means 20 are read out, and an image signal representing an abnormal shadow candidate Pp of the whole image signal S is read out. An abnormal shadow candidate detection processing device 10 for detecting Sp (hereinafter, referred to as a local image signal); a local image processing means 40 for emphasizing an abnormal shadow candidate on the detected local image signal Sp; The visible image is composed of the whole image represented by the whole image signal S ′ after image processing by the local image processing unit 30 and the abnormal shadow candidate represented by the local image (abnormal shadow candidate) signal Sp ′ after image processing by the local image processing means 40. When It is provided with a display unit 50 for displaying Te configuration.

Here, the whole image signal S input to the computer-aided image diagnostic apparatus 100 is, for example, an excitation light on a stimulable phosphor sheet on which an image P representing a mammogram of a patient as shown in FIG. Is an image signal (a signal of a high density and a high signal level) obtained by photoelectrically reading the stimulated emission light generated by irradiating the image, and then performing digital conversion.

The display means 50 may display the entire image and the abnormal shadow candidate separately on the display surface. In the present embodiment, while displaying the entire image, the abnormal shadow candidate in the entire image is displayed. Only the image part is replaced with the image of the abnormal shadow candidate image-processed by the local image processing means 40 and displayed.

As shown in detail in FIG. 1, the abnormal shadow candidate detection processing device 10 performs predetermined arithmetic processing for detecting an abnormal shadow candidate on the entire image signal S, and outputs the obtained arithmetic processing. By comparing the value K with the threshold value T as a predetermined detection processing control parameter stored in the threshold value storage means 2, the abnormal shadow candidate detection processing means 1 for detecting and processing the abnormal shadow candidate Pp in the whole image P is processed. The apparatus further includes a threshold value changing unit 3 as a parameter changing unit for changing a predetermined threshold value T.

Here, the content of the threshold processing by the abnormal shadow candidate detection processing means 1 is such that if the processing output value K for a predetermined area in the entire image P is larger than the threshold T (K>
T), the region is detected as an abnormal shadow candidate Pp, and if the calculation processing output value K is equal to or smaller than the threshold value T (K ≦
T), the region is not detected as an abnormal shadow candidate Pp,
That is.

The threshold changing means 3 changes the minimum threshold Tmin to the maximum threshold Tmax (Tmi) in the ROC characteristic curve shown in FIG.
threshold value Tin to be changed in the range up to n <Tmax)
(Tmin.ltoreq.Tin.ltoreq.Tmax) and a threshold value T stored in advance in the threshold value storage means 2.
And means for changing to.

Next, the operation of the computer-aided image diagnosis apparatus 100 including the abnormal shadow candidate detection processing apparatus 10 according to the present embodiment will be described.

The whole image signal S representing the image P representing the mammogram is input to the computer-aided image diagnostic apparatus 100, and this whole image signal S is temporarily stored in the storage means 20. Here, the whole image processing means 30 reads out the whole image signal S stored in the storage means 20, and performs image processing such as gradation processing and frequency processing on the whole image signal S.

On the other hand, the abnormal shadow candidate detection processing device 10 also reads out the whole image signal S stored in the storage means 20, and this whole image signal S is inputted to the abnormal shadow candidate detection means 1.

Here, when the image interpreter attempts to perform the TP-oriented detection in order to suppress the detection omission of the abnormal shadow, the threshold Tin (Tin) smaller than the threshold T stored in the threshold storage means 2 in advance. <T) is input to the threshold changing unit 3. The threshold changing unit 3 changes the threshold T stored in the threshold storing unit 2 to the input threshold Tin. The abnormal shadow candidate detecting means 1 performs an abnormal shadow candidate detecting process on the read whole image signal S, and when performing a threshold process on whether or not to finally detect the abnormal shadow candidate, the threshold storing means 2 The stored threshold value Tin is read out, and threshold value processing is performed based on the read-out changed threshold value Tin and the output value K of the abnormal shadow candidate detection processing.

The abnormal shadow candidate detection processing means 1 outputs the output value K
Is larger than the threshold value, it is detected as an abnormal shadow candidate, and if the output value K is equal to or smaller than the threshold value, it is not detected as an abnormal shadow candidate. Becomes smaller than the value T stored in advance, so that the output value K often exceeds the threshold value Tin more than the threshold value T,
As a result, the area detected as an abnormal shadow candidate increases, and detection omission of an abnormal shadow is suppressed.

On the other hand, in order to prevent the image reader from erroneously detecting a normal shadow as an abnormal shadow candidate, an FP emphasis detection is performed, which is larger than a threshold value T stored in the threshold value storage means 2 in advance. When the threshold value Tin (T <Tin ') is input to the threshold value changing means 3, the threshold value changing means 3 replaces the threshold value T stored in the threshold value storing means 2 with the inputted threshold value T
in ′, and the abnormal shadow candidate detecting means 1 performs threshold processing based on the changed threshold value Tin ′ read from the threshold storing means 2 and the output value K of the abnormal shadow candidate detecting processing.
Since the threshold value Tin ′ to be compared with the output value K becomes larger than the value T stored in advance, the output value K
Is smaller than the threshold Tin than when the threshold T is exceeded. As a result, the area detected as an abnormal shadow candidate is reduced, and erroneous detection of a normal shadow as an abnormal shadow candidate is suppressed.

The local image signal Sp representing the image of the abnormal shadow candidate detected according to the detection request level is input to the local image processing means 40, and the local image processing means 40 receives the input local image signal Sp. Is subjected to image processing for clarifying the interpretation of the abnormal shadow candidate, and is output as the processed local image signal Sp ′.

The whole image signal S 'outputted from the whole image processing means 30 and the local image signal Sp' outputted from the local image processing means 40 are inputted to the display means 50, and the whole image represented by the whole image signal S 'is The image of the abnormal shadow candidate represented by the local image signal Sp 'is displayed on the display means 50. In this display, as described above, while displaying the entire image, only the image portion of the abnormal shadow candidate in the entire image is replaced with the image of the abnormal shadow candidate image-processed by the local image processing means 40 and displayed. is there.

More specifically, in the above-described embodiment, the abnormal shadow candidate detection processing means 1 includes an iris filter processing means for detecting a tumor shadow candidate by processing based on the iris filter processing described above and a microscopic processing by processing based on morphological operation processing. By applying the morphological operation processing means for detecting and processing the calcified shadow candidate, it is possible to detect a tumor shadow candidate or a microcalcified shadow candidate which is a characteristic form of breast cancer, and the threshold value changing means 3 performs the iris filter processing. Output value K (same as output value I described above)
And T1 (and T2) (see Equation (17)) to be compared with the output value of the morphological operation process and the threshold value T to be compared with, so that the tumor shadow candidate and the minute These abnormal shadow candidates can be detected according to the detection request level of the calcified shadow candidates.

The above embodiment is an example in which a predetermined threshold is applied as a detection processing control parameter in the present invention, and a threshold changing means is applied as a parameter changing means, but the present invention is limited to the above embodiment. Not a thing,
In particular, various parameters other than the threshold can be applied to the detection processing control parameters.

As shown in FIG. 11, it is preferable to adopt a more user-friendly configuration. That is, the apparatus further comprises a detection request level input means 3a for inputting the detection request level of the abnormal shadow candidate by the user, and sets the threshold value changing means 3 'to a threshold value corresponding to the detection request level input to the detection request level input means 3a. By having the function of changing the threshold value stored in the threshold value storage means 2 after the conversion into the threshold value, the user can change the threshold value in a practical manner.

In general, detection control parameters such as thresholds are set to be the same for each facility, each imaging device, each radiologist, and each patient (subject). Therefore, by storing (storing) the changed parameters for each facility, each imaging device, each image reading doctor, and each patient, it is possible to save the trouble of resetting these parameters in the subsequent processing. Further, by storing at least the image data targeted for the abnormal shadow candidate detection process and this parameter in association with each other, if the abnormal shadow candidate detection process needs to be performed again on the image data later, the parameter Can be referred to, and the parameters of the abnormal shadow candidate detection processing performed on the new image data can be referred to by referring to the parameters of the abnormal shadow candidate detection processing performed on the other image data.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram showing an embodiment of an abnormal shadow candidate detection processing device according to the present invention.

FIG. 2 is a block diagram showing an example of a computer-aided image diagnosis apparatus using the abnormal shadow candidate detection processing apparatus shown in FIG.

FIG. 3 is a diagram showing a mammogram represented by an image signal S input to the computer-aided image diagnostic apparatus shown in FIG. 3;

FIG. 4 is a diagram showing ROC characteristics.

FIG. 5 is a conceptual diagram showing the degree of concentration of a concentration gradient in a mammogram.

FIG. 6 is a diagram showing a mask for calculating a gradient vector in iris filter processing.

FIG. 7 is a diagram showing the concept of the degree of concentration of a gradient vector for a target pixel.

FIG. 8 is a conceptual diagram showing an iris filter set so that a contour shape is adaptively changed.

FIG. 9 is a graph illustrating a basic operation of the morphological operation processing;

FIG. 10 is a diagram showing the concept of a structural element used in morphological operation processing.

FIG. 11 is a diagram showing another embodiment in which a part of the configuration of the abnormal shadow candidate detection processing device of the embodiment shown in FIG. 1 is changed.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Abnormal shadow candidate detection processing means 2 Threshold storage means 3 Threshold change means 10 Abnormal shadow candidate detection processing apparatus 20 Storage means 30 Overall image processing means 40 Local image processing means 50 Display means 100 Computer-aided image diagnostic apparatus

Claims (10)

[Claims] 1. An abnormal shadow candidate detection method for detecting an abnormal shadow candidate in a radiation image based on an image signal representing a radiation image of a subject, wherein: a detection process control for controlling a detection level of the abnormal shadow candidate A method for detecting abnormal shadow candidates, wherein parameters are freely changeable. 2. A process in which the abnormal shadow candidate detection process is performed by performing a predetermined arithmetic process on the image signal and comparing the obtained output value with a predetermined threshold value set in advance. 2. The abnormal shadow candidate detection processing method according to claim 1, wherein the detection processing control parameter is the predetermined threshold. 3. The abnormal shadow candidate detecting method according to claim 1, wherein the predetermined arithmetic processing is processing based on iris filter processing, and the abnormal shadow candidate is a tumor shadow candidate. 4. The abnormal shadow candidate detecting method according to claim 1, wherein the predetermined arithmetic processing is processing based on morphological arithmetic processing, and the abnormal shadow candidate is a microcalcification shadow candidate. . 5. The detection processing control parameter according to claim 1, wherein the detection processing control parameter is changed according to a user's request level for detecting the abnormal shadow candidate.
The abnormal shadow candidate detection processing method according to the paragraph. 6. An abnormal shadow candidate detection processing device comprising abnormal shadow candidate detection means for detecting an abnormal shadow candidate in the radiation image based on an image signal representing a radiation image of the subject, wherein the abnormal shadow candidate is detected. An abnormal shadow candidate detection processing device further comprising parameter changing means for freely changing a detection processing control parameter for controlling a level. 7. The processing for detecting the abnormal shadow candidate is performed by performing predetermined arithmetic processing on the image signal and comparing the obtained output value with a predetermined threshold set in advance. The abnormal shadow candidate detection processing apparatus according to claim 6, wherein the detection processing control parameter is the predetermined threshold. 8. The abnormality according to claim 6, wherein said predetermined arithmetic processing by said abnormal shadow candidate detecting means is processing based on iris filter processing, and said abnormal shadow candidate is a tumor shadow candidate. Shadow candidate detection processing device. 9. The method according to claim 6, wherein the predetermined arithmetic processing by the abnormal shadow candidate detection means is processing based on morphological arithmetic processing, and the abnormal shadow candidate is a microcalcification shadow candidate. Abnormal shadow candidate detection processing device. 10. A detection request level input means for receiving an input of a detection request level of the abnormal shadow candidate by a user, wherein the parameter change means changes the parameter according to the detection request level input to the detection request level input means. The abnormal shadow candidate detection processing device according to any one of claims 6 to 9, wherein the detection processing control parameter is changed.