JP2002074326A - System for detecting abnormal shadow candidate - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide support information for improving diagnosis performance when a person such as a doctor interpreting the radiogram of a picture diagnoses whether the candidate area of a detected abnormal shadow is malignant or not. SOLUTION: A candidate area detection means 10 detects the candidate area of the abnormal shadow in the picture based on original picture data P inputted from an original picture memory 90 receiving and storing original picture data P. A feature quantity calculation means 20 calculates the feature quantity of the candidate area detected by the candidate area detection means 10. A judgment means 30 judged whether the candidate area is malignant or not based on calculated feature quantity and at the same time, a feature quantity picturing means 40 makes the value of feature quantity into picture data. A picture output means 50 is provided with three CRTs and outputs original picture data P stored in the original picture memory 90, judgment result information by the judgment means 30 and picture data of feature quantity onto the picture of each CRT.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an abnormal shadow candidate detection processing system, and more particularly, displays support information for diagnosing whether or not an abnormal shadow candidate area detected from a radiation image of a subject is malignant. The present invention relates to an abnormal shadow candidate detection processing system.

[0002]

2. Description of the Related Art Conventionally, in the medical field, it has been necessary to interpret a radiographic image of a subject, find a lesion, and observe the state of the lesion to diagnose the presence or absence of a disease and the progress of the disease. Generally done. However, interpretation of radiographic images is influenced by the experience of the interpreter and the level of image interpretation ability, and is not always objective.

[0003] For example, in mammography (diagnosis radiation image of a breast as a subject) photographed for the purpose of examining breast cancer, a tumor shadow, a microcalcification shadow, or the like, which is one of the features of a cancerous portion, is obtained from the image. It is necessary to detect the abnormal shadow, but it is not always possible for the reader to specify the range of the abnormal shadow accurately. For this reason, it has been required to accurately detect abnormal shadows such as tumor shadows and microcalcification shadows without depending on the skill of the reader.

In order to meet this demand, abnormal shadow candidate detection is performed in which, based on image data of a subject acquired as a diagnostic image, abnormal shadow candidates in an image represented by the image data are automatically detected using a computer. A processing system (computer-assisted image diagnostic apparatus) has been proposed (Japanese Patent Laid-Open No. 8-29)
No. 4479, JP-A-8-287230, etc.).

This abnormal shadow candidate detection processing system automatically detects abnormal shadow candidates using a computer based on the density distribution characteristics and morphological characteristics of abnormal shadows, and mainly detects tumor shadows. An abnormal shadow candidate area is detected by using iris filter processing suitable for performing the processing, or morphological filter processing suitable for mainly detecting microcalcification shadow.

[0006] The iris filter processing compares a iris filter output value representing the maximum value of the degree of concentration of the density gradient of the image signal with a predetermined threshold value to obtain a tumor, which is one of the characteristic forms of breast cancer in the image. On the other hand, the morphological filter processing is a method effective for detecting a candidate region of a shadow. On the other hand, the morphological filter processing is based on a morphological operation processing using a structural element having a size larger than the microcalcified shadow to be detected on the image signal. By comparing the output value with a predetermined threshold value, this is an effective method for detecting a candidate region of microcalcification shadow, which is one of the characteristic forms of breast cancer, in an image.

The abnormal shadow candidate detection processing system sets an abnormal shadow candidate area detected by iris filter processing or morphology filter processing and a local area including a nearby area as an area of interest (hereinafter referred to as ROI). Specialized image processing such as enhancement processing is performed on the ROI image in accordance with the attribute (whether a tumor shadow or microcalcification shadow) of the abnormal shadow candidate area included in the ROI image, while the original image is processed. The predetermined image processing is performed also on the entire image for improving the image interpretation performance. Further, the abnormal shadow candidate detection processing system includes a whole image that has been subjected to image processing and a ROI that has been subjected to the specialized image processing.
The image is output as one image to an image display unit such as a CRT or a printing unit such as a laser printer.

Specifically, when mammography is applied as a subject, for example, as shown in FIG.
Is displayed on the right side of the display surface of the target image in such a manner that the entire image of the right breast and the ROI image W including the detected abnormal shadow candidate P ₃₀ ′ and its neighboring area P ₃₁ are superimposed, and the abnormal shadow is detected on the left side when viewed. One layout image information laid out so as to display the entire left breast image that has not been
Output to RT.

On the other hand, among the abnormal shadow candidate regions obtained by the iris filter processing and the morphology filter processing, normal shadows and benign shadows (hereinafter, for simplicity, normal shadows and benign shadows are collectively referred to as non-malignant shadows). In order to exclude only those malignant shadows by excluding the ones that have been erroneously detected, the candidate area is determined using the feature amount representing the shape, internal, and peripheral features of the detected candidate area, and the candidate area is determined to be malignant There has been proposed a method of detecting only the selected candidate region as a final abnormal shadow candidate (Japanese Patent Laid-Open No. 9-167238).

As a method of detecting only the malignant shadow, a density histogram inside the candidate area is obtained for the candidate area of the tumor shadow detected by the iris filter processing, and a plurality of feature quantities based on the density histogram, that is, a variance, are calculated. Values, contrasts, angular moments, etc., and each feature value is defined by a predetermined weighting function to calculate a new evaluation function value. Based on the calculated evaluation function value, whether the candidate area is malignant is determined. And a method of detecting only the candidate area determined to be malignant. In addition, as the feature amount, in addition to the one related to the above-described density histogram, the variance value, the bias, the correlation value, the moment, the entropy, which is edge information indicating the feature of the edge of the candidate region, and the feature of the shape of the candidate region are indicated. Circularity or the like is used.

The Mahalanobis distance can be used as the evaluation function value. The Mahalanobis distance means Dmi defined by the following equation (1), and is a distance measured from the center of the distribution by weighting a hyperellipsoid represented by a covariance matrix Σ.

[0012]

(Equation 1) According to equation (1), a Mahalanobis distance Dm1 with a pattern class (i = 1) indicating a non-malignant shadow obtained in advance by experiment and a Mahalanobis distance Dm2 with a pattern class (i = 2) indicating a malignant shadow are calculated. Then, Dm1 and Dm2 are compared to determine whether the candidate area is malignant. For example, Mahalanobis distance D with a pattern class indicating a non-malignant shadow
When m1 is closer to the Mahalanobis distance Dm2 with the pattern class indicating a malignant shadow, that is, when Dm1 <Dm2, the pattern class is a non-malignant shadow and indicates a malignant shadow from the Mahalanobis distance Dm1 with the pattern class indicating the non-malignant shadow. If the Mahalanobis distance Dm2 is short, that is, Dm1>
In the case of Dm2, it is determined that the image is a malignant shadow.

[0013]

However, the abnormal shadow candidate detection processing system detects abnormal shadow candidates, and finally diagnoses whether or not the detected abnormal shadow candidate is a malignant shadow. Is an image interpreter such as a doctor who observes and interprets the image of the abnormal shadow candidate output on the display screen or the like. Therefore, it is desired to provide the image interpreter with various information effective for making a final diagnosis.

In view of the above circumstances, the present invention provides support information that can improve diagnostic performance when an image reader finally diagnoses whether or not a detected abnormal shadow candidate area is a malignant shadow. It is an object to provide an abnormal shadow candidate detection processing system to be provided.

[0015]

An abnormal shadow candidate detection processing system according to the present invention includes a candidate area detecting means for detecting an abnormal shadow candidate area in a radiation image based on radiation image data representing a radiation image of a subject; In an abnormal shadow candidate detection processing system including a feature amount calculating unit that calculates a feature amount of a detected candidate region, and an image output unit that outputs radiation image data as a visible image, based on a value of the calculated feature amount. Further comprising a feature amount imaging unit that converts the feature amount into image data, wherein the image output unit converts the position information indicating the position of the detected candidate region and the image data of the feature amount of the candidate region into a visible image together with the radiation image data. It is characterized by outputting.

Here, the abnormal shadow candidate area is determined by the detection processing used by the candidate area detection means, that is, the iris filter processing for detecting a tumor shadow or the morphology filter processing for detecting microcalcification shadows. Indicates the detected shadow, not only malignant shadow,
A feature that is detected because the feature on the image shows a feature similar to a tumor shadow or a microcalcification shadow, and includes a shadow that is not clear whether it is an abnormal shadow or a benign tumor or the like. .

The feature amount of the candidate area includes, for example, a variance value, a contrast, and an angular moment representing a feature of a density histogram of the candidate area, a variance value, a bias, a correlation value, and a moment representing a feature of an edge of the candidate area. , Entropy, and a circularity representing the characteristic of the shape of the candidate area, and is a value that is effective for determining whether the candidate area is malignant and for determining whether the candidate area is malignant.

The conversion of a feature value into image data means that, for example, the closer the feature value is to the lowest possible value, the higher the possibility that the candidate area is a malignant shadow, and the closer the feature value is to the highest value, the higher the candidate area is. When it can be determined that the region is highly likely to be a non-malignant shadow, the highest image signal value is set to 0, the lowest image signal value is set to 1024 using 10-bit data, and the feature amount is calculated. Values within the obtained range are linearly assigned to image signal values from 0 to 1024, and the feature amount is converted into image data having the assigned signal values.

That is, for example, the circularity Sp (Spreadness), which is one of the features of the tumor shadow, is obtained by calculating the area A of the candidate area and its center of gravity AO as shown in FIG. Assuming a virtual circle having a radius R having the same area as the area A, the occupancy of the area of the candidate area included inside the virtual circle with respect to the area A of the candidate area, and a portion where the virtual circle and the candidate area overlap Is the area of A ′, the circularity is calculated by the following equation (2).

[0020]

(Equation 2) The value of the circularity value is in the range of 0.0 to 1.0.However, since the shape of a tumor shadow is more likely to be non-malignant as its shape is closer to a circle, the value of the circularity value is closer to 1.0. It is non-malignant, and it can be said that the closer to 0.0, the more likely it is. Therefore, circularity
The circularity is linearly assigned to the signal value so that the signal value becomes 0 and the circularity 0.0 becomes the signal value 1024, and is converted into image data. Is converted to an image having a higher signal value as the image is more malignant.
As a result, for example, in an image represented by an image signal with a high density and high signal level, the feature amount image is displayed as an image with a higher density as the degree of malignancy is increased, and is displayed as a lighter image as the degree of non-malignancy is increased.

Further, the abnormal shadow candidate detection processing system according to the present invention comprises a candidate area detecting means for detecting a candidate area of an abnormal shadow in a radiation image based on radiation image data representing a radiation image of a subject; A feature amount calculating unit for calculating a feature amount of the region, a local region extracting unit for extracting local image data of a local region including the detected candidate region, and a candidate region indicating at least the candidate region in the extracted local image data An abnormal shadow candidate detection process including local image enhancement means for performing predetermined image enhancement processing on image data, and image output means for outputting the enhanced local image data and radiation image data as visible images The system further includes a feature amount imaging unit configured to convert the feature amount into image data based on the calculated value of the feature amount,
The image output means displays the image data of the feature amount of the candidate region included in the local region based on the local image data and the position information indicating the position of the candidate region together with the local image data and the radiation image data subjected to the enhancement processing to the visible image. May be output.

Here, the predetermined image enhancement processing refers to, for example, gradation processing, frequency processing, enlargement processing, and the like.
In addition to the above, various enhancement processes are possible as long as the enhancement processing improves the reading performance of a local image based on the local image data, particularly, a candidate area image based on the candidate area image data.

The local region including the candidate region means a region near the candidate region including the candidate region, and various shapes such as a rectangle, a circle, and an ellipse can be adopted as the peripheral shape thereof. .

Further, the position information may be an arrow pointing to the position of the candidate area. Further, the position information may be a frame indicating a rectangular area including the candidate area. In addition, a frame other than a rectangle, such as a circle or an ellipse, may be used according to the peripheral shape of the local region.

Further, the abnormal shadow candidate detection processing system of the present invention requires a storage means for storing position information indicating the position of a candidate area and image data of a characteristic amount of the candidate area, and a request for displaying a characteristic amount at a desired position. The image output means receives a request for displaying a characteristic amount at a predetermined position by the position indicating means, and refers to the position information stored in the storage means to determine the characteristic amount at the position. May be output.

Further, the image processing apparatus may further include a determination unit that determines whether the candidate area is malignant based on the characteristic amount, and the image output unit may further output the determination result of the candidate area by the determination unit.

It is preferable that the characteristic amount is a plurality of characteristic amounts calculated by the characteristic amount calculating means.

The radiation image is preferably a breast radiation image.

The image output means refers to an image display means such as a CRT or a printing means such as a printer.

The abnormal shadow candidate detection processing system according to the present invention includes a candidate area detecting means for detecting a candidate area of an abnormal shadow in a radiation image by a filtering process based on radiation image data representing a radiation image of a subject; An abnormal shadow candidate detection processing system including an image output unit that outputs radiation image data as a visible image, wherein the image output unit outputs an output result of the filter processing as a visible image together with the radiation image data. It may be a thing.

The above-mentioned filter processing refers to iris filter processing for mainly detecting a tumor shadow and morphological filter processing for mainly detecting a microcalcification shadow.

[0032]

According to the abnormal shadow candidate detection processing system of the present invention configured as described above, the characteristic amount is obtained along with the radiation image data of the subject and the position information indicating the position of the candidate region of the abnormal shadow in the subject. Is output to a display screen or the like, so that an image interpreter such as a doctor interprets a radiation image of a subject and diagnoses whether or not the candidate area is a malignant shadow. It is possible to easily refer to a feature amount image obtained by imaging the characteristics of the feature amount, and it is expected that the diagnostic ability of a radiogram interpreter will be improved. That is, since the feature amount of each candidate area is an index value useful for determining whether or not the candidate area is malignant, the doctor finally determines the malignancy of the abnormal shadow detected as the candidate area. It is often useful for a doctor or the like to diagnose an abnormal shadow by making it possible to refer to a feature image in which the malignancy of each feature can be seen at a glance.

In the case where the apparatus further comprises a judgment means for judging whether or not the candidate area is malignant by using a plurality of feature values, a judgment factor of the candidate area judged to be malignant by the judgment means is provided. Can be confirmed from the feature image, and in particular, when the image reader has a doubt about the candidate area determined not to be malignant, the overlooking of the malignant shadow can be performed by referring to the feature image. This can be prevented. That is, since the determination means determines a candidate area based on an evaluation function value obtained by defining a plurality of feature amounts using a predetermined weighting function, for example, one of nine feature amounts protrudes. If the other feature value indicates a low value even if the value indicates a high value, the evaluation function value becomes a low value, and as a result, it can be determined that the image is not malignant. However, unlike a candidate region in which all nine feature values indicate low values, a candidate region having such characteristics may be malignant, so that the image interpreter makes a final diagnosis for each candidate region. At this time, by referring to the image of each feature amount, it is possible to make a diagnosis without overlooking a candidate area that may be a malignant shadow.

[0034]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of an abnormal shadow candidate detection processing system according to the present invention will be described below with reference to the drawings.
FIG. 1 is a diagram showing a specific embodiment of the abnormal shadow candidate detection processing system according to the present invention.

The abnormal shadow candidate detection processing system according to the present embodiment is based on an original image memory 90 for inputting and storing original image data P from an image reader or the like, and based on the original image data P input from the original image memory 90. A candidate area detecting means 10 for detecting a candidate area of an abnormal shadow in an image (hereinafter simply referred to as a candidate area for simplicity); a feature quantity calculating means 20 for calculating a feature quantity of the detected candidate area; Determining means 30 for determining whether the candidate area is malignant based on the amount,
A feature amount imaging unit 40 for converting the value of the feature amount into image data,
An image output means 50 for outputting the original image data P stored in the original image memory 90, the determination result information by the determination means 30, and the image data of the feature amount. Image output means 50
Is composed of three CRTs and has input means such as a mouse and a keyboard.

Next, the operation of the abnormal shadow candidate detection processing system according to this embodiment configured as described above will be described.

The original image data P of the mammography (see FIG. 2A), which is a radiographic image of the breast, is input from an image reading device or the like to the original image memory 90 and stored.

The candidate area detecting means 10 receives the original image data P from the original image memory 90, performs an iris filter process on the original image data P, and detects a candidate area of a tumor shadow.

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 known that the density value of the tumor shadow part is slightly lower than that of the surrounding image part. Therefore, inside the tumor shadow, a gradient of the density value is seen in which the density value decreases from the periphery of the substantially circular shape 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 region of a tumor shadow is detected.

Hereinafter, description will be made with reference to the mammography shown in FIG. Gradient vector at an arbitrary pixel inside the tumor pattern P ₁ in the original image data P is directed to the vicinity of the center of masses shadow P ₁ as shown in FIG. 2 (2), an elongated shadow P ₂ such as a blood vessel shadows and mammary gland 2 As shown in (3), the gradient vector does not concentrate on a specific point. Therefore, to evaluate the distribution of orientation of locally gradient vector, by detecting the area are concentrated in a specific point, it is the tumor pattern P ₁ in the candidate region. Also, shading P ₃ which is elongated shadows between the mammary gland and the like, as shown in FIG. 2 (4) crossed can be detected as a pseudo-candidate region because there is a tendency that the gradient vector is concentrated on a specific point.

Hereinafter, a specific algorithm of the iris filter processing will be described.

First, for all the pixels constituting the target image, the direction θ of the gradient vector of the image data is obtained for each pixel j based on the calculation formula shown in the following equation (3).

[0044]

(Equation 3) Here, as shown in FIG. 3, f _{1 to} f ₁₆ are pixel values (image data) corresponding to pixels on the outer periphery of a mask having a size of, for example, 5 pixels × 5 pixels around the pixel j. It is.

Next, for all the pixels constituting the target image, the degree of concentration C of the gradient vector using the pixel as the pixel of interest is calculated according to the following equation (4).

[0046]

(Equation 4) Here, N is the number of pixels existing in a circle having a radius R around the target pixel, θj is a straight line connecting the target pixel and each pixel j in the circle, and the above equation (3) for each pixel j is used. This is the angle formed by the calculated gradient vector (see FIG. 4). Accordingly, the degree of concentration C expressed by the above equation (4) 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 in which the degree of concentration C takes a large value can be said to be the pixel at the center of the tumor shadow. On the other hand, the value of the degree of concentration C is small in an elongated shadow such as a blood vessel shadow because the direction of the gradient vector is biased in a certain direction. Therefore, for all the pixels that make up the image,
The value of the degree of concentration C for each pixel of interest is calculated,
By evaluating whether or not the value of the degree of concentration C exceeds a preset threshold, a tumor shadow can be detected. That is, this filter is less affected by blood vessels, mammary glands, and the like than a normal difference filter, and has a feature that a tumor shadow can be detected efficiently.

Further, in actual processing, in order to achieve a detection power that is not affected by the size or shape of the tumor, a measure is taken to adaptively change the size and shape of the filter. FIG. 5 shows the filter. This filter is different from the filter shown in FIG.
The above-mentioned degree of concentration is evaluated only by pixels on a radial line in M types of directions (in FIG. 5, 32 directions every 11.25 degrees) adjacent at an angular interval of degrees.

Here, the coordinates ([x], [y]) of the n-th pixel on the i-th line and from the pixel of interest are given by the following equation, assuming that the coordinates of the pixel of interest are (k, l). It is given by (5) and (6).

[0050]

(Equation 5) Here, [x] and [y] are the maximum integers not exceeding x and y.

Further, the output value up to the pixel where the maximum concentration is obtained for each line on the radiation line is defined as the concentration Cimax in that direction, and the concentration Cimax is averaged in all directions. Let the value be the degree of concentration C of the gradient vector group for the pixel of interest.

More specifically, first, the degree of concentration Ci (n) obtained from the pixel of interest to the n-th pixel on the i-th radial line is obtained by the following equation (7).

[0053]

(Equation 6) That is, in equation (7), the starting point is set to the target pixel, and the ending point is set to Rmi.
n within the range from n to Rmax,
(N) is calculated.

Here, Rmin and Rmax are the minimum and maximum values of the radius of the tumor shadow to be detected.

Next, the concentration C of the gradient vector group is calculated by the following equations (8) and (9).

[0056]

(Equation 7) Here, Cimax in equation (8) is the maximum value of the degree of concentration Ci (n) for each radial direction line obtained in equation (7).
The area from the pixel of interest to the pixel at which the degree of concentration Ci (n) is the maximum value is the area of the tumor shadow in the direction of the line.

Equation (8) is calculated for all radial direction lines to determine the edge points of the tumor shadow area on each line, and the adjacent edge points of the tumor shadow area on each line are calculated as follows: By connecting with a straight line or a non-linear curve, it is possible to specify the contour of a candidate area that can be a candidate for a tumor shadow.

In the equation (9), the maximum value Cimax of the degree of concentration given by the equation (8) in this area is calculated in all directions of the radial direction line (the equation (9) exemplifies the case of 32 directions). An average value is determined for. The obtained value is an iris filter output value I, and this output value I is compared with a predetermined threshold value T suitable for determining whether or not the shadow is a tumor shadow, and I ≧ T (or If I> T), a region centered on the target pixel is a candidate region, and I <T
(Or ≦ T), it is determined that the area is not a candidate area,
Detect candidate areas.

The above-mentioned calculation of the degree of concentration Ci (n) is as follows.
The following equation (7 ′) may be used instead of equation (7).

[0060]

(Equation 8) That is, equation (7 ') shows that the concentration corresponding to the minimum value Rmin of the radius of the tumor shadow to be detected is set as the starting point, and the concentration point Ci (n) within the range where the ending point is from Rmin to Rmax.
Is calculated.

The candidate area detecting means 10 acquires both the image data of the candidate area detected from the original image data P and the position data indicating the position of the candidate area on the original image.

The feature amount calculating means 20 is provided with the candidate area detecting means 10
, Image data of the candidate area is input, and based on the image data, the shape, the inside, and the edge of the candidate area are calculated.

As the first feature amount of the candidate area, the circularity Sp is used because the outline of the tumor shadow is a shape close to a circle. As shown in FIG. 6, the area A of the candidate region and its center of gravity AO are obtained, and a virtual circle having a radius R centered on the center of gravity AO and having the same area as the area A is assumed, and candidates included in the virtual circle are included. The circularity is calculated as the occupation ratio of the area to the area A. That is, assuming that the area of the portion where the virtual circle and the candidate area overlap is A ', the circularity is calculated by the following equation (2).

[0064]

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

[0065]

(Equation 10) In addition, IFED (Iris Fil
ter Edge) Five feature amounts calculated from the image are used. Hereinafter, the calculation method will be described in detail.

The candidate area detected by the iris filter processing, that is, the tumor shadow P representing the breast cancer in the radiation image
About ₁ and pseudo-abnormal pattern P _3, it detects an image portion including the vicinity thereof, for example, as a square area, creating a peripheral edge image using the iris filter processing for the detected square area (IFED image).

That is, in equation (8) of the iris filter processing, the position of the point giving the maximum value of the concentration Ci (n) on the i-th line extending radially from the target pixel is obtained. However, this process does not limit the value of n that gives the maximum value.

As a result, if the target pixel is the candidate area P ₁ or P ₃
When n is within the range, n when the equation (8) takes the maximum value
Indicates the pixel to which the i-th line intersects the edge B of the candidate regions P ₁ and P _3. For example, as shown in FIG. 7, pixels B1, B2, B3, and B4 are designated for the target pixel 1, and pixels B2, B5, B6, and B7 are provided for the target pixel 2.
Instruct.

On the other hand, when the target pixel is outside the candidate areas P ₁ and P ₃ , the maximum value of the expression (8) is obtained when the target pixel itself is specified. That is, in FIG. 7, for the target pixel 3 outside the candidate regions P ₁ and P ₃ , the value of Expression (8) becomes the maximum when the target pixel 3 itself is indicated.

As described above, all the pixels in the square area including the candidate area are sequentially set as the target pixels, and the pixel having the maximum value in the equation (8) is counted. This is illustrated schematically in FIG.

That is, the count values for pixels outside the candidate areas P ₁ and P ₃ are all “1”, and the count values for pixels inside the candidate areas P ₁ and P ₃ are all “0”. count value for the pixels on the edge B of the area P ₁ and P ₃ is an image composed of all value of 1 or more is obtained. The image of this count value is defined as an IFED image.

Next, the following processing is performed on this IFED image to create a simultaneous generation matrix.

That is, as shown in FIG.
The center of gravity AO of ₁ and P ₃ is obtained, a radial line is extended from the center of gravity AO, an arbitrary point on this line is set as i, and a point perpendicular to this line and separated from the point i by two pixels is defined as j.

The count value of the point i in the IFED image and the count value of the point j are counted up into a matrix as shown in FIG. Specifically, when the point i is outside the candidate areas P ₁ and P ₃ , the count value of the point i in the IFED image is “1”, and at that time the point j is also in the candidate areas P ₁ and P ₃ . If it is outside, the count value at point j is also “1”. In this case, in the matrix of FIG. 10, “1” is counted in the column where “1” in the vertical direction i and “1” in the horizontal direction j intersect. Is done.

On the other hand, the point i is the candidate area P₁ , P_Three Inside
Yes, and point j is also a candidate area P₁ , P _Three If inside
Since the count value is 0 at both the i point and the j point,
In the column where “0” in direction i and “0” in horizontal direction j intersect
Count "1".

Further, when the point i is the candidate area P₁ , P_Three Margin of
B, j point is also a candidate area P₁ , P _Three On the edge B of
If the count value at point i is “5” and the count value at point j is
When the event value is “3”, “5” in the vertical direction i and the horizontal direction
“1” is counted in the column where “3” in the direction j intersects.
The count value that counts up in this matrix is cumulative
Is what you do. That is, the count value is “5” again.
When scanning the i-point and the j-point whose count value is “3”,
“5” in the vertical direction i and “3” in the horizontal direction j of the matrix
"2" which added "1" to the original "1" in the crossing column
Is stored.

Since the point i is an arbitrary point in the IFED image, the matrix is completed by scanning a radial line so that all the pixels of the IFED image become the point i, and scanning the point i on the line. Let it. The matrix of the IFED image is called a simultaneous generation matrix Pg (x, y).

Here, when the candidate area is a tumor shadow, the shape characteristics of the tumor shadow that the edge of the tumor shadow is substantially circular, and the fact that the points i and j are extremely close to each other are considered. , I point is on the periphery (the count value of the IFED image has a large value of 1 or more), and the j point may be on the periphery (the count value of the IFED image has a large value of 1 or more). Is extremely high.

On the other hand, when the candidate area is a pseudo abnormal shadow, it is extremely rare that the pseudo abnormal shadow has a circular edge like the above-mentioned intersection of two blood vessels.
Even if the point and the point j are close to each other, just because the point i is on the edge does not necessarily mean that the point j is also on the edge. Rather, the possibility that the point j is on the edge is extremely low. Become.

Therefore, the simultaneous generation matrix Pg (x, y)
Is significantly different depending on whether the candidate area is a tumor shadow or a pseudo abnormal shadow. The characteristic value of this simultaneous generation matrix is edge information, and this edge information is used as a feature value. That is, from the following expression, the fifth feature value (1
3), a sixth feature (14) representing the deviation dfe (difference entropy), a seventh feature (15) representing the correlation value cor (correlation), and a moment idm (inverse difference m).
oment), the eighth feature (16), entropy se
A ninth feature (17) representing (sum entropy) is obtained.

[0081]

[Equation 11] The feature amount calculating means 20 inputs the position data of the candidate area on the original image together with the image data of the candidate area from the candidate area detecting means 10 and associates the nine feature quantities in the candidate area with the position data. calculate. That is, when a plurality of candidate areas are detected by the candidate area detection means 10, nine feature values are calculated for each of the candidate areas in association with the position data.

The determining means 30 receives the values of the above nine characteristic amounts and the position data from the characteristic amount calculating means 20, and determines whether or not the candidate area is malignant using the Mahalanobis distance.

First, according to the following equation (1), a pattern class (i =
The Mahalanobis distance Dm1 from 1) and the Mahalanobis distance Dm2 from the pattern class (i = 2) indicating the malignant shadow are calculated.

[0084]

(Equation 12) The nine feature amounts obtained by the feature amount calculation means 20 correspond to the above-mentioned x1 to x9, respectively, (x1, x2, x
3,..., X9). The Mahalanobis distance between the pattern of the candidate area expressed on the 9-dimensional pattern space and the pattern of the non-malignant shadow is Dm1.
Similarly, the Mahalanobis distance to the malignant shadow pattern is Dm2.

Here, the non-malignant shadow pattern and the malignant shadow pattern are defined as a vector x for each of the non-malignant shadows and each of the malignant shadows, which is set in advance based on the result of experimentally investigating a large number of abnormal shadow candidates. Means the defined pattern space. For example, a pattern class w1 formed by an average of the above-mentioned vector x for a non-malignant shadow, and a pattern class w formed by an average of the above-mentioned vector x for a non-malignant shadow, respectively.
Indicated by 2.

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

The likelihood ratio is represented by Dm1 / Dm2, and is shown in FIG.
1 shows an inclination on a coordinate plane. In other words, it can be determined that the larger the likelihood ratio is, the higher the possibility of a malignant shadow is, and the smaller the likelihood ratio is, the higher the possibility of a non-malignant shadow is. Therefore, for example, a threshold value is set to 2, and when the likelihood ratio is 2 or more, Is determined to be non-malignant when less than 2.

The determining means 30 acquires the determination result data in association with the position data of the candidate area.

The feature amount image forming means 40 includes the feature amount calculating means 20
, And the position data of the candidate area having the characteristic amounts are input, and each is converted into an image signal value based on the value of each characteristic amount. That is, the value of each feature value is more malignant as it approaches the highest value (or the lowest value) of the range that the value of the feature value can take, and it is more non-malignant as it approaches the lowest value (or the highest value). , Because it has the property of being able to distinguish malignant or non-malignant likelihood by its value, in a specific range that includes all possible values of each feature, of the lowest or highest value in that range The values of the feature amounts are linearly assigned to the image signal values so that the value that can be identified as non-malignant has a signal value of 0, and the value that can be identified as malignant has a signal value of 1024. The image data is converted into image data having a signal value based on each value.

For example, the range that the value calculated as the circularity Sp can take is included in the range of 0.0 to 1.0 according to the above equation (2). A tumor shadow has the characteristic that the closer it is to a circle, the more likely it is to be non-malignant.
It can be said that the closer the circularity is to 1.0, the less aggressive, and the closer the circularity is to 0.0, the more likely it is to be malignant.
Therefore, a circularity of 1.0 indicates a signal value of 0 and a circularity of 0.0 indicates a signal value of 10.
By assigning the value of the circularity linearly to the signal value so as to be 24 and converting the signal value into image data, the non-malignant image is converted into an image with a low signal value, and the malignant image is converted into an image with a high signal value.

Similarly, the second feature value (the above equation (10)) representing the variance var of the density histogram of the candidate area, the third feature value (the above equation (11)) representing the contrast con, and the angular moment asm are A fourth feature quantity (the above equation (12)), a fifth feature quantity (the above equation (13)) representing the variance var of the simultaneous generation matrix which is the feature quantity of the edge of the candidate area, and the bias dfe , The sixth feature value (Expression (14) above), and the correlation value co
r, the seventh feature quantity (formula (15) above), moment id
The eighth feature quantity (the above equation (16)) representing m, the entropy
Regarding the value of the ninth feature amount (the above equation (17)) representing se,
Based on the characteristics of each feature, the image is converted into an image having a higher signal value as the possibility of being malignant is higher, and a signal value as being more likely as being non-malignant.

The image output means 50 outputs the image data of the original image, the determination result data of the candidate area detected by the candidate area detection means 10 by the determination means 30 and the position data of the candidate area to be determined, Input image data and the position data of the candidate area having the characteristic amount by the converting means 40, and display the original image, the determination result, the arrow indicating the position of the candidate area, and the characteristic amount image on the CRT screen. Display above.

For example, an original image is displayed on the first CRT screen shown in FIG. On the displayed original image P ′, there are shadows P ₁₀ and P ₂₀ which can be candidate regions for the tumor shadow, but since these shadows P ₁₀ and P ₂₀ are not explicitly specified, image interpretation by the image reader is performed. It may not be detected depending on the ability.

On the second CRT screen shown in FIG. 13, among the candidate areas detected by the candidate area detecting means 10,
Shadow P ₁₀ it is determined that the malignancy by determination means 30 is clearly with reference to the positional data in the appropriate position on the original image P'as a candidate region P ₁₀ '. That is, the original image P '
With displays arrow M ₁₀ to demonstrate the candidate region P ₁₀ ', the image reading person regardless of the interpretation ability, it is possible to confirm the presence of the candidate region by referring to this arrow. On the other hand, no arrow is displayed on the candidate area P ₂₀ ′, which is the shadow P ₂₀ that has been detected as a candidate area by the candidate area detection means 10 but has been determined not to be malignant by the determination means 30, so that the candidate area P ₂₀ ′ May be detected or not detected depending on the ability of the image reader, as in the case of the first CRT.

On the third CRT screen shown in FIG.
Complementary area P_Ten´, P₂₀'Feature amount image C_Ten, C₂₀But each
Candidate area P having the feature amount_Ten´, P₂₀´ position data
With reference to the data, it is displayed at a corresponding position on the original image P '.
At this time, first, the feature amount image of circularity, which is the first feature amount,
An image in which the image has a signal value corresponding to the value of the feature amount
C _Ten, C₂₀As the candidate area P on the original image_Ten´, P₂₀of
Position, and each time the mouse is clicked, a ninth
The feature amount is displayed.

The feature amount images C ₁₀ and C ₂₀ are displayed as image signals of high density and high signal level. That is, the feature amount images C ₁₀ and C ₂₀ are displayed as darker images as the feature amounts of the candidate regions P ₁₀ ′ and P ₂₀ ′ have more malignant elements, and the more the feature amounts have non-malignant elements, It is displayed as an image with a low density (bright).

[0097] For example, although FIG. 14 is obtained by displaying the feature quantity image representing a circularity, the candidate region P ₁₀ 'because the feature amount image C ₁₀ of being displayed in a darker image, the candidate region P _10'
It can be understood that the value of the circularity indicated that the degree of malignancy was high. On the other hand, since the feature amount image C ₂₀ of the candidate area P ₂₀ ′ is displayed as a thin image, it can be seen that the value of the circularity has a low possibility of being malignant, but the signal value is 0.
Therefore, it is necessary to consider it a little, and it is possible to inform the image reader that the candidate area P ₂₀ ′ exists at this location.

That is, since the characteristic amount of the malignant abnormal shadow tends to be generally high over all the characteristic amounts, the characteristic amount of the malignant abnormal shadow is located at the position of the candidate area P ₁₀ ′ which is clearly indicated as malignant on the second CRT screen. corresponding to the third CRT position on the screen of, but most of the feature image C ₁₀ of the nine feature quantity is displayed as dark images, corresponding to the candidate region P ₂₀ 'which has been determined not to be malignant in the third position of the CRT screen of the most of the feature image C ₂₀ it is displayed as a thin image. On the other hand, even in a candidate area determined not to be malignant, when only a value of a certain feature amount is extremely high,
In fact, it can be malignant. These candidate areas are:
As in the candidate region P ₂₀ ′, the first CRT and the second CRT
Since it is not specified on the screen of the RT, there is a high possibility that it is usually overlooked. However, feature amount images of all candidate regions are displayed on the screen of the third CRT irrespective of the result of the determination means 30, and particularly feature amount images that can cause malignancy are displayed as dark images. By referring to the third CRT, the image interpreter can make a diagnosis without overlooking a candidate area that can be a malignant shadow among candidate areas overlooked by the determination unit 30.

Next, a second specific embodiment of the present invention will be described. FIG. 15 is a diagram illustrating a configuration of the abnormal shadow candidate detection processing system according to the present embodiment. The first
The description of the same elements as those of the embodiment is omitted unless otherwise required.

The abnormal shadow candidate detection processing system according to the present embodiment is based on an original image memory 90 for inputting and storing original image data P from an image reading device or the like, and based on the original image data P input from the original image memory 90. A candidate area detecting means 10 for detecting a candidate area of an abnormal shadow in an image; a local area extracting means 60 for extracting image data of a local area including the detected candidate area; Local image enhancement means 70 for performing predetermined image enhancement processing, feature amount calculation means 20 for calculating the feature amount of the detected candidate area, and determining whether the candidate area is malignant based on the calculated feature amount. A determining means 30 for determining, and a feature amount imaging means 40 for converting the value of the feature amount calculated by the feature amount calculating means 20 into image data
And the original image data P stored in the original image memory 90,
An image output unit 50 outputs the local image emphasized by the local image enhancement unit 70, the determination result information by the determination unit 30, and the image data of the feature amount.

Next, the operation of the abnormal shadow candidate detection processing system according to this embodiment configured as described above will be described.

The local area extracting means 60 is a candidate area detecting means.
From 10, input the original image data P including the detected candidate area and the position data of the candidate area on the original image, and include the candidate area in the input original image data P based on the position data of the candidate area. A nearby area is specified according to a preset processing procedure, and local image data is extracted.

The local image enhancing means 70 receives the extracted local image data and its position data, performs image processing such as gradation processing, frequency processing, and enlargement processing on the local image data. get.

The image output means 50 receives the local image data, its position data and the emphasized image data from the local image enhancing means 70 and, as shown in FIG. 16, displays the candidate area detecting means 10 on the second CRT screen. Region P detected by
_The local image consisting of ₃₀ ′ and its neighboring area P _{31 is} transformed into the original image P
'And refer to the position data at the corresponding position on the
₃₀ displays the position with the rectangular frame M ₃₀ demonstrates a ', to display the ROI image W based on the emphasis screen data subjected to the enlargement processing on the partial image data at the same time. Note that the local image enhancement means 70 temporarily stores the acquired enhanced image data, and outputs the enhanced image data to the image output means 50 to display the ROI image W only when a signal indicating a predetermined display request is received. It is desirable to do. In other words, since it is usual for the radiologist to observe the entire image first and recognize the information of the entire image to some extent, and then observe the detailed structure, the observer first observes the entire image, and after or during the observation By inputting a display request with a simple operation at the desired time,
It is desirable to be able to refer to the ROI image immediately.

Similarly, the result of the judgment by the judging means 30 is temporarily stored in the judging means 30, and is output to the image output means 50 when the interpreter inputs a request for displaying the judgment result (not shown). ). That is, in the present embodiment, all the candidate areas are specified together with the rectangular frame indicating the position on the second CRT screen, and the determination result by the determination means 30 can be referred to by the request of the image interpreter. Judgment means
Stored in 30.

The other elements are the same as in the first embodiment.

Note that the arrow or rectangular frame for specifying the candidate area may indicate only the candidate area determined to be malignant as in the first embodiment, or may be the one for the second embodiment. As described above, all the candidate areas including the candidate area determined not to be malignant may be displayed so that the determination result can be referred to. Also, the shape thereof can be various shapes such as a circular shape and an elliptical shape, in addition to an arrow and a rectangular frame.

Further, the form of the image display means 50 is not limited to the above-mentioned form, and the screen may be switched and displayed by one CRT. At this time, the feature image corresponding to the position is displayed on the screen only when the reader observing the original image P 'inputs a request signal requesting the display of the feature image at the desired position from the input means. It can also be displayed. In addition to the method of displaying the feature amount image as a shaded image, a method of changing the color according to the value of the feature amount, a method of changing the color for each feature amount and displaying as a shaded image of each color, etc. That is, any image may be used as long as the image is such that the degree of malignancy can be determined at a glance from the feature amount, such as displaying a malignancy scale (see FIG. 17) on the screen. In addition to the method of displaying the nine feature values, a method of automatically displaying the feature amounts in addition to the method of displaying the nine feature amounts by clicking the mouse as described above, and of course, the display order may be any order.

The display mode of the original image P 'is not limited to a set of left and right breasts displayed at the same time back to back, and one set of left and right breasts may be displayed alternately.

Note that the method of assigning a feature value to an image signal value is not limited to a mode in which an image signal value is assigned to a specific range including all possible values of each feature value as described above. It is also possible to assign only the value of the range in which the likelihood of malignancy or non-malignancy that can be discriminated from the value of each feature amount can be significantly identified to the image signal value. The image signal value may be assigned in any form as long as the image clearly indicates the malignancy or non-malignancy that can be determined.

In the abnormal shadow candidate detection processing system according to the present embodiment, the processing for detecting a tumor shadow candidate using the iris filter has been described. However, the processing is not limited to the iris filter processing, and the microcalcification shadow candidate can be detected using a morphology filter. The method can be applied to any method for detecting an abnormal shadow, such as a process for performing the processing.

The feature values are not limited to the above nine feature values. For example, a value of a feature amount is determined by imaging the likelihood of malignancy determined from the Mahalanobis distance from a malignant pattern and / or a non-malignant pattern and the likelihood ratio of the Mahalanobis distance from a malignant pattern and / or a non-malignant pattern. The image data may be converted into image data by the same method and referred to as a feature amount image, or an image obtained by converting the density of calcification into image data may be used. That is, any value may be used as the value to be converted into image data as long as it is a value that is effective in determining the likelihood of malignancy of the candidate area.

In the above-described embodiments, the description has been given of the case where the feature amount is converted into an image. However, the output of various filter processes in the candidate region detecting means such as the iris filter process and the morphology filter process is formed as an image. Can be used as a diagnosis support image. For example, in the microcalcification shadow detection processing, an image in which calcification is extracted by morphology filter processing (from the original image)
In the tumor shadow detection processing, the image obtained by performing the iris filter processing on the original image may be directly referred to as the diagnosis support image in the tumor shadow detection processing.

[0114] As a method of determining whether or not the candidate area is malignant, a determination method using the Mahalanobis distance is used, but the determination method is not limited to this.

When the simultaneous generation matrix is created, the point j is defined as a point perpendicular to the radial line where the point i exists and separated by two pixels from the point i (see FIG. 9). ,
The angle of intersection with the radial line, the pixel separated from point i, and the like can be appropriately changed according to the size of the tumor shadow to be detected.

The present invention is not limited to breast CAD, but can be applied to chest CAD and the like.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an abnormal shadow candidate detection processing system according to an embodiment of the present invention;

FIG. 2 is a diagram illustrating an operation of an iris filter process.

FIG. 3 is a diagram illustrating a mask having a size of 5 pixels vertically and 5 pixels horizontally centered on a target pixel j;

FIG. 4 is a view for explaining an angle between a target pixel and a gradient vector at each pixel j.

FIG. 5 is a conceptual diagram showing an iris filter set so that a contour shape is adaptively changed.

FIG. 6 is a diagram showing a virtual circle having an area equal to the area A of a candidate area;

FIG. 7 is a diagram illustrating an operation of forming an IFED image;

FIG. 8 is a conceptual diagram showing an IFED image.

FIG. 9 is a diagram illustrating an operation of creating a simultaneous generation matrix based on an IFED image;

FIG. 10 is a diagram showing a simultaneous generation matrix.

FIG. 11 is a diagram showing the operation of a judgment method based on the Mahalanobis distance.

FIG. 12 shows a first CRT in the embodiment shown in FIG.
Showing the information output to the

FIG. 13 is a view showing information output to a second CRT in the embodiment.

FIG. 14 is a diagram showing information output to a third CRT in the embodiment.

FIG. 15 is a configuration diagram of an abnormal shadow candidate detection processing system according to another embodiment of the present invention;

FIG. 16 is a diagram showing information output to a second CRT in the embodiment.

FIG. 17 is a diagram illustrating an example of a feature image.

[Explanation of symbols]

 10 Candidate area detection means 20 Feature amount calculation means 30 Judgment means 40 Feature amount imaging means 50 Image output means 60 Local area extraction means 70 Local image enhancement means 90 Original image memory

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 4C093 AA05 CA15 DA06 FD05 FF16 FF17 FF19 FF20 FF28 FF37 FF43 FG13 5B057 AA08 CA08 CB08 CE06 DA20 DB09 DC09 DC23 5L096 AA06 BA06 FA37 FA59 GA55 JA11

Claims (9)

[Claims] 1. A candidate area detecting means for detecting a candidate area of an abnormal shadow in a radiation image based on radiation image data representing a radiation image of a subject, and a candidate area detecting means for detecting the candidate area detected by the candidate area detecting means. In an abnormal shadow candidate detection processing system comprising: a feature amount calculating unit that calculates a feature amount; and an image output unit that outputs the radiation image data as a visible image, the value of the feature amount calculated by the feature amount calculating unit The image output means further includes a feature amount imaging unit configured to convert the feature amount into image data based on the position information indicating the position of the detected candidate region and image data of the feature amount of the candidate region. An abnormal shadow candidate detection processing system, which outputs a visible image together with the radiation image data. 2. A candidate area detecting means for detecting a candidate area of an abnormal shadow in a radiation image based on radiation image data representing a radiation image of a subject; and a candidate area detecting means for detecting the candidate area detected by the candidate area detecting means. A feature amount calculating unit that calculates a feature amount; a local region extracting unit that extracts local image data of a local region including the detected candidate region; and at least one of the local image data extracted by the local region extracting unit. Local image enhancement means for performing a predetermined image enhancement process on the candidate area image data indicating the candidate area; and a visible image of the local image data and the radiation image data which have been enhanced by the local image enhancement means. An abnormal shadow candidate detection processing system comprising: an image output unit that outputs the feature amount; Further comprising a feature amount imaging unit that converts the feature amount into image data based on the local image data, wherein the image output unit is based on the local image data together with the enhanced local image data and the radiation image data. An abnormal shadow candidate detection processing system, characterized in that image data of a feature amount of a candidate area included in a local area and position information indicating the position of the candidate area are output as a visible image. 3. The abnormal shadow candidate detection processing system according to claim 1, wherein the position information is an arrow pointing to the position of the candidate area. 4. The abnormal shadow candidate detection processing system according to claim 1, wherein the position information is a frame indicating a rectangular area including the candidate area. 5. A storage device for storing position information indicating a position of the candidate region and image data of a feature amount of the candidate region, and a position instructing unit for requesting display of a feature amount at a desired position. When the image output unit receives a request for displaying a feature amount at a predetermined position by the position instruction unit, the image output unit refers to the position information stored in the storage unit and converts the image data of the feature amount at the position. The abnormal shadow candidate detection processing system according to any one of claims 1 to 4, wherein the system outputs the abnormal shadow candidate. 6. A determination unit that determines whether or not the candidate area is malignant based on the feature amount, wherein the image output unit further outputs a determination result of the candidate area by the determination unit. The abnormal shadow candidate detection processing system according to any one of claims 1 to 5, wherein: 7. The abnormal shadow candidate detection processing system according to claim 1, wherein said characteristic amount is a plurality of characteristic amounts calculated by said characteristic amount calculating means. 8. The abnormal shadow candidate detection processing system according to claim 1, wherein the radiation image is a breast radiation image. 9. A candidate region detecting means for detecting a candidate region of an abnormal shadow in the radiation image based on radiation image data representing a radiation image of a subject by a filter process, and outputting the radiation image data as a visible image. An abnormal shadow candidate detection processing system comprising an image output unit, wherein the image output unit outputs an output result of the filter processing as a visible image together with the radiation image data.