JP4513365B2 - Medical image processing apparatus and medical image processing program - Google Patents

Description

  The present invention relates to a medical image processing apparatus and a medical image processing program, and more particularly to a medical image processing apparatus and a medical image processing program for improving the efficiency of diagnosis when a doctor interprets a medical image.

  In the medical field, digitization of medical images has been realized, and medical image data generated by a CR (Computed Radiography) device, a CT (Computed Tomography) device, an MRI (Magnetic Resonance Imaging) device, etc. is converted into a CRT (Cathode Ray Tube). ) And the like, and a doctor interprets a medical image displayed on the image display device, and diagnoses by observing a state of a lesioned part and a change over time.

  Computer-aided diagnosis (Computer-Aided Diagnosis) that analyzes image data using digital image processing technology and automatically detects abnormal shadow candidates such as breast cancer and lung cancer for the purpose of reducing the burden on doctors' interpretation. CAD) has been developed (see, for example, Patent Document 1).

  Examples of a method for determining whether a shadow image to be detected is an abnormal shadow image or a normal image include discriminant analysis performed by extracting feature amounts of the detection target shadow image and multivariate analysis such as similar image detection. . In discriminant analysis, feature amounts of a plurality of reference images classified in advance into an abnormal shadow image and a normal image are stored, a boundary is calculated, and the feature amount of the detection target shadow image is determined as an abnormal shadow image and a normal image. It is determined which of the two is closer. In the similar image detection, the feature amount of the detection target shadow image is compared with the feature amount of the reference image, and the reference image having the most similar points is detected.

  In order to improve discrimination performance by such multivariate analysis, it is effective to use more feature quantities. However, when the feature quantity is small relative to the number of data of the reference image, the discrimination performance is improved by adding the feature quantity. However, if the feature quantity becomes too large relative to the number of data of the reference image, the statistical reliability Decreases, and the discrimination performance decreases. In order to prevent this, it is important to collect more reference images along with the addition of feature amounts.

In addition, when detecting an abnormal shadow candidate by a conventional method, even if the size, position, and direction of the same image are changed, different feature amounts may be displayed. Therefore, an abnormal shadow image corresponding to any size, position, and direction must be stored as a reference image.
JP 2002-112985 A

  However, since the number of abnormal shadow images is much smaller than that of normal images, it is difficult to actually collect many abnormal shadow images as reference images. Furthermore, it is not practical to create an abnormal shadow image corresponding to any size, position, and direction because it takes a lot of time and cost.

  The present invention has been made in view of the above-described problems in the prior art. Whether the detection target shadow image is an abnormal shadow image or a normal image with high accuracy while reducing costs by using fewer reference images. It is an object of the present invention to provide a medical image processing apparatus and a medical image processing program capable of discriminating the above.

The invention according to claim 1 is a comparison that compares a reference image that has been classified into an abnormal shadow image and a normal image in advance and a detection target shadow image extracted from a medical image obtained by photographing a subject. Means for determining whether the detection target shadow image is an abnormal shadow image or a normal image based on a comparison result by the comparison means, the size of the detection target shadow image in the medical image processing apparatus is the size of the corresponding position or direction in the reference image, and corrects the position or direction, including an image correction means for outputting a detection target shadow image the amendment to the comparison means, the image correction means, said detection A principal axis angle is obtained using a moment or a geometric feature of the target shadow image, and the direction of the detection target shadow image is corrected based on the obtained principal axis angle .

  According to a second aspect of the present invention, in the medical image processing apparatus according to the first aspect, there are a plurality of the reference images, and the size, position, or direction of each reference image is unified.

  According to a third aspect of the present invention, in the medical image processing apparatus according to the first or second aspect, the medical image is a mammogram.

According to a fourth aspect of the present invention, in the medical image processing apparatus according to any one of the first to third aspects, the image correction unit performs enlargement or reduction, parallel movement, or rotational movement using affine transformation. Thus, the size, position, or direction of the detection target shadow image is corrected.

The invention according to claim 5 is the medical image processing device according to any one of claims 1 to 4 , wherein the determination as to whether the detection target shadow image is an abnormal shadow image or a normal image is This is performed based on the comparison result by the comparison means and the analysis result of the detection target shadow image before correction.

A sixth aspect of the present invention is the medical image processing apparatus according to any one of the first to fifth aspects, wherein the amount of parallel movement in correcting the position of the detection target shadow image determined to be the abnormal shadow image A rotation angle in correcting the direction of the detection target shadow image determined to be the abnormal shadow image, a detection target shadow image determined to be the abnormal shadow image, or detection detected to be the abnormal shadow image A display means for displaying size information of the target shadow image is provided.

According to the seventh aspect of the present invention, a reference image previously classified into an abnormal shadow image and a normal image and a detection target shadow image extracted from a medical image acquired by photographing a subject are stored in a computer. In a medical image processing program for realizing a comparison function for comparison and a discrimination function for discriminating whether the detection target shadow image is an abnormal shadow image or a normal image based on a comparison result in the comparison function, When the computer realizes an image correction function that corrects the size, position, or direction of the detection target shadow image to a size, position, or direction corresponding to the reference image, and realizes the comparison function, use is compared with the detected shadow image the corrected image, when realizing the image correction function, a moment or geometric characteristics of the detection target shadow image Te let seek main axis angle, characterized in that to correct the direction of the detection target shadow image based on the main axis angle obtained.

According to an eighth aspect of the present invention, in the medical image processing program according to the seventh aspect , there are a plurality of the reference images, and the size, position, or direction of each reference image is unified.

The invention according to claim 9 is the medical image processing program according to claim 7 or 8 , wherein the medical image is a mammogram.

According to a tenth aspect of the present invention, in the medical image processing program according to any one of the seventh to ninth aspects, when the image correction function is realized, the affine transformation is used to enlarge or reduce, translate, or The size, position, or direction of the detection target shadow image is corrected by performing rotational movement.

The invention according to claim 11 is the medical image processing program according to any one of claims 7 to 10 , wherein when the discrimination function is realized, the comparison result in the comparison function and the detection target shadow before correction The detection target shadow image is determined to be an abnormal shadow image or a normal image based on an analysis result of the image.

According to a twelfth aspect of the present invention, in the medical image processing program according to any one of the seventh to eleventh aspects, the computer corrects the position of the detection target shadow image determined to be the abnormal shadow image. The amount of parallel movement, the rotation angle in the correction of the direction of the detection target shadow image determined to be the abnormal shadow image, the detection target shadow image determined to be the abnormal shadow image, or the abnormal shadow image A display function for displaying size information of the determined shadow image to be detected is realized.

According to the inventions described in claims 1 and 7 , the reference image is compared with the detection target shadow image in which the size, position or direction is corrected to the size, position or direction corresponding to the reference image, and this comparison is performed. Since it is determined whether the detection target shadow image is an abnormal shadow image or a normal image based on the result, it is possible to match the size, position, or direction of the detection target shadow image with the reference image and then compare with the reference image. it can. Therefore, since it is not necessary to prepare a reference image corresponding to any size, position, or direction, whether the detection target shadow image is an abnormal shadow image with high accuracy while reducing costs by using a smaller number of reference images. It is possible to determine whether the image is a normal image.
In addition, the principal axis angle is obtained using the moment or geometric feature of the detection target shadow image, and the direction of the detection target shadow image is corrected based on the obtained main axis angle, so the direction of the detection target shadow image is used as the reference image. After matching, it can be compared with the reference image.

According to the second and eighth aspects of the present invention, since the size, position, or direction of each reference image is unified, the size, position, or direction of the detection target shadow image can be easily set to a size corresponding to the reference image. It can be corrected to the position or the direction.

According to the third and ninth aspects of the invention, it is possible to determine whether the detection target shadow image is an abnormal shadow image or a normal image with respect to a mammogram.

According to the fourth and tenth aspects of the present invention, the size, position, or direction of the shadow image to be detected is corrected by performing enlargement or reduction, translation, or rotation using affine transformation. Correction of the size, position, or direction of the shadow image can be performed by a single conversion.

According to the inventions of claims 5 and 11 , the detection target shadow image is an abnormal shadow image based on the comparison result of comparing the reference image with the corrected detection target shadow image and the analysis result of the detection target shadow image before correction. Therefore, it is possible to accurately determine whether the detection target shadow image is an abnormal shadow image or a normal image while reducing costs by using fewer reference images. .

According to the sixth and twelfth aspects of the present invention, the parallel movement amount in the correction of the position of the detection target shadow image determined to be an abnormal shadow image, the detection target shadow image determined to be the abnormal shadow image, and the like. The rotation angle in the direction correction, the detection target shadow image determined to be the abnormal shadow image, or the size information of the detection target shadow image determined to be the abnormal shadow image are displayed. Can let you know.

[First Embodiment]
The medical image processing apparatus 1 according to the first embodiment of the present invention will be described in detail with reference to FIGS. However, the scope of the invention is not limited to the illustrated examples.

  First, a schematic configuration of the medical image processing apparatus 1 will be described. As shown in FIG. 1, a medical image processing apparatus 1 includes a CPU (Central Processing Unit) 2, an I / F (InterFace) unit 3, an operation unit 4, a display unit 5, a communication unit 6, and a ROM (Read Only Memory) 7. , A RAM (Random Access Memory) 8, a storage unit 9, a reference image database 10, and the like, and each unit is connected by a bus 11.

  In accordance with various instructions input from the operating means 4 and data input from the I / F unit 3 or the communication unit 6, the CPU 2 stores a program designated from various programs stored in the ROM 7 in the work area of the RAM 8. And the various processes are executed in cooperation with the program, and the processing results are stored in a predetermined area of the RAM 8 and displayed on the display means 5.

  The CPU 2 extracts a detection target shadow image from the image data, corrects the size, position, or direction of the detection target shadow image to the size, position, or direction corresponding to the reference image, creates a corrected image, The correction image is compared, and based on the comparison result, it is determined whether the detection target shadow image is an abnormal shadow image or a normal image, and one or more types of abnormal shadow candidates are detected (FIG. 3). reference).

  The I / F unit 3 is an interface for connecting to the image generation apparatus G, and inputs image data generated by the image generation apparatus G to the medical image processing apparatus 1.

  As the image generation device G, a laser digitizer that scans a laser beam on a film on which a medical image acquired by photographing a subject is recorded to acquire image data, or a photoelectric conversion element such as a CCD (Charge Coupled Device) A film scanner or the like that acquires image data recorded on a film by a sensor consisting of the above can be applied.

  Also, instead of reading a medical image recorded on a film, an imaging device that captures a medical image using a stimulable phosphor, and a radiation detection element and a capacitor that generate charges according to the intensity of irradiated radiation A flat panel detector or the like can be connected, and the method for inputting image data is not particularly limited.

  The operation means 4 includes a keyboard composed of cursor keys, numeric keys, and various function keys, and outputs a press signal generated by each key operation to the CPU 2. Note that the operation unit 4 may include a pointing device such as a mouse or a touch panel, and other input devices as necessary.

  The display means 5 is composed of an LCD (Liquid Crystal Display) or the like, and displays various display information such as medical images and detection results of abnormal shadow candidates under the control of the CPU 2.

  The communication unit 6 includes a communication interface such as a network interface card, a modem, and a terminal adapter, and transmits and receives various types of information to and from external devices on the communication network. For example, it is good also as a structure which receives image data from the image generation apparatus G via the communication part 6, or it connects with the server in a hospital and the medical terminal installed in each clinic via the communication part 6, and is abnormal. It is good also as a structure which transmits the detection result of a shadow candidate.

  The ROM 7 is composed of a nonvolatile semiconductor memory and stores various programs executed by the CPU 2.

  The RAM 8 is composed of a rewritable semiconductor element. The RAM 8 is a storage medium in which data is temporarily stored, and a program area for developing a program to be executed by the CPU 2 and a data area for storing various processing results by the CPU 2 are formed. .

  The storage unit 9 is configured by a magnetic or optical recording medium, a semiconductor memory, or the like, and includes image data input from the image generation device G via the I / F unit 3, a detection target shadow image extracted from the image data, A correction image created by correcting the detection target shadow image, a feature amount extracted from the correction image, and the like are stored.

The reference image database 10 stores the feature amounts of a plurality of reference images that serve as a reference when determining abnormal shadows. As shown in FIG. 2, each reference image is generated by the cutout region units of a predetermined size, the main shaft length L _1, L ₂ in the cutout region, · · ·, L _n, the center of gravity position G _1, G ₂ ,..., G _n and the spindle angle are unified. Here, they are unified so that the principal axis angle is 0 degree. The reference image is classified in advance into an abnormal shadow image and a normal image.

The principal axis length is the length of the maximum length line (main axis) that can be drawn in the target graphic, and is one of the feature quantities characterizing the size of the object. By finding the distance in pixel units for all combinations of two points on the boundary of the object and obtaining the two points having the maximum distance, the end points (x ₁ , y ₁ ) and (x ₂ , y ₂ ) of the main axis Is obtained. The spindle length L is given by equation (1).

The principal axis angle θ is an angle formed between the x-axis of the image and the principal axis, and is obtained using a geometric feature as shown in the equation (2). The principal axis angle θ is one of the feature quantities that characterize the direction of the object.

ただし、ｆ（ｘ，ｙ）は（ｘ，ｙ）における画像データである。 The principal axis angle θ can also be obtained using the moment of the target graphic. The (p + q) next moment m _pq is defined by equation (3).

Here, f (x, y) is image data at (x, y).

ここで、

である。 The (p + q) next moment around the center of gravity (central moment) μ _pq is defined by equation (4).

here,

It is.

The principal axis angle θ obtained from the moment is expressed by Equation (6).

Next, the operation in the first embodiment will be described.
FIG. 3 is a flowchart for explaining the abnormal shadow candidate detection process 1 executed by the medical image processing apparatus 1. Hereinafter, it is assumed that the subject of the processing of each step is the CPU 2 unless otherwise specified. Here, an example in which a mammogram obtained by photographing a breast is used as medical image data will be described, but the target region is not limited to this, and may be the chest or abdomen.

  As shown in FIG. 3, first, image data is input from the image generation device G via the I / F unit 3 (step S <b> 1) and stored in the storage unit 9.

  Next, the image data stored in the storage unit 9 is subjected to image analysis, and a detection target shadow image that may be an abnormal shadow such as a mass shadow or a microcalcification cluster is extracted (step S2).

  Next, the size, position, and direction of the detection target shadow image are corrected to the size, position, or direction corresponding to the reference image, and a corrected image is created (step S3).

The corrected image creation process will be described with reference to FIG.
First, the principal axis length of the detection target shadow image is calculated (step S11), and the enlargement ratio or reduction ratio is determined so as to have a size corresponding to the reference image (step S12).

  Further, the barycentric position of the detection target shadow image is calculated (step S13), and the parallel movement condition is determined so as to be a position corresponding to the reference image (step S14).

  Further, the principal axis angle of the detection target shadow image is calculated (step S15), and the rotational movement condition is determined so as to be in the direction corresponding to the reference image (step S16).

  Then, based on the determined enlargement ratio or reduction ratio, parallel movement condition, and rotational movement condition, an affine transformation is applied to the detection target shadow image to create a corrected image (step S17).

  FIG. 5 shows an example of the corrected image. FIG. 5A is an example of a corrected image created by enlarging based on the spindle length. FIG. 5B is an example of a corrected image created by performing parallel movement based on the position of the center of gravity. FIG. 5C is an example of a corrected image created by performing rotational movement based on the principal axis angle.

  Next, as shown in FIG. 3, feature amounts are extracted from the corrected image (step S4). Here, a feature amount related to texture is used as the feature amount.

  FIG. 6 shows an example of a local pattern used for texture analysis. 6 is a feature amount indicating the value (pixel value) of the image data of the center pixel when the image is analyzed in units of 3 × 3 pixel blocks. f1 represents a value obtained by doubling the value of the image data of the center pixel, and f2 represents a value obtained by multiplying the value of the image data of the center pixel by three. f3 represents a value obtained by multiplying the value of the image data of the center pixel and the pixel on the right side of the 3 × 3 pixel block. f34 represents a value obtained by multiplying the image data of the center pixel of the 3 × 3 pixel block, the upper left pixel, and the upper right pixel. The same applies to other patterns.

  When the image data of each pixel has the value shown in FIG. 7, when attention is paid to the 3 × 3 pixel block F, f0 is 2, f1 is 2 × 2 = 4,..., F34 is 1 × 2 × 3. = 6. In this way, 35 feature amounts are calculated. The 3 × 3 pixel block is shifted pixel by pixel with respect to the image data, and feature amounts are calculated for all regions.

  Next, as shown in FIG. 3, the feature quantity extracted from the corrected image is compared with the feature quantity of the reference image stored in the reference image database 10 using multivariate analysis, and the detection target shadow image is compared. Is an abnormal shadow image or a normal image (step S5).

Discriminant analysis and similar image detection can be used as a method of discriminating the detection target shadow image. In discriminant analysis, based on a reference image belonging to a group of normal images and a reference image belonging to a group of abnormal shadow images prepared in advance, a normal image and an abnormal shadow image on a feature amount space with the feature amount as an axis The boundary line is provided, and it is determined which group the detection target shadow image belongs to. An example of discriminant analysis using the Mahalanobis distance will be described with reference to FIG. The Mahalanobis distance is a distance from the center of each group in consideration of data dispersion to the discrimination data (detection target shadow image). FIG. 8 is a plot of the reference image with feature quantity 1 and feature quantity 2 as two variables. As shown in FIG. 8, two groups are prepared: a group A consisting of reference images a _{1 to} a ₄ classified as abnormal shadow images, and a group B consisting of reference images b _{1 to} b ₄ classified as normal images. to advance, obtaining the Mahalanobis distance D _a from the center G _a group a for discriminating data X, the group B the Mahalanobis distance D _b between the center G _b respectively. Then, based on the distance ratio D _a / D _b , it is determined which group the determination data X belongs to. When the distance ratio D _a / D _b is equal to or less than the threshold value, the determination data X is determined to be an abnormal shadow image close to the group A, and when the distance ratio D _a / D _b is greater than the threshold value, the determination data X is determined to be a normal image close to the B group. FIG. 8 shows a boundary line L between the abnormal shadow image and the normal image based on the Mahalanobis distance ratio.

ここで、Ｚ_akは検出対象陰影画像のｎ個の特徴量のうちのｋ番目の特徴量、Ｚ_bkは類似画像検出に用いる基準画像のｎ個の特徴量のうちのｋ番目の特徴量である。 In the similar image detection, the Euclidean distance D between the detection target shadow image and the reference image in the feature amount space represented by the equation (7) is calculated, and the reference image with the short distance D is detected as the similar image.

Here, Z _ak is the k-th feature amount of the n feature amounts of the detection target shadow image, and Z _bk is the k-th feature amount of the n feature amounts of the reference image used for similar image detection. is there.

  Detecting a reference image with the shortest distance D, and determining whether the detection target shadow image is an abnormal shadow image or a normal image based on whether the reference image is an abnormal shadow image or a normal image It is also possible to detect a plurality of reference images with a short distance D and determine whether the detection target shadow image is an abnormal shadow image or a normal image based on the ratio of the abnormal shadow image to the normal image. Also good.

Then, the detection target shadow image determined to be an abnormal shadow image is detected as an abnormal shadow candidate (step S6), and the detection result is displayed on the display means 5 (step S7). Specifically, the abnormal shadow candidate on the image data is indicated by a marker or the like, and the parallel movement amount in the correction of the position of the detection target shadow image determined to be an abnormal shadow image, the rotation angle in the direction correction, or The size information of the detection target shadow image determined to be an abnormal shadow image is displayed. Here, the size information refers to the size of the detection target shadow image before correction or the size of the cut-out image when the detection target shadow image is extracted.
Thus, the abnormal shadow candidate detection process 1 is completed.

  As described above, according to the medical image processing apparatus 1, the size, position or direction of the detection target shadow image is corrected to the size, position or direction corresponding to the reference image, and the reference image and the corrected image are compared. Therefore, since it is determined whether the detection target shadow image is an abnormal shadow image or a normal image based on the comparison result, the size, position, or direction of the detection target shadow image is matched with the reference image and Can be compared. Therefore, since it is not necessary to prepare a reference image corresponding to any size, position, or direction, whether the detection target shadow image is an abnormal shadow image with high accuracy while reducing costs by using a smaller number of reference images. It is possible to determine whether the image is a normal image.

  In addition, since the size, position, or direction of each reference image is unified, the size, position, or direction of the detection target shadow image can be easily corrected to the size, position, or direction corresponding to the reference image. .

  In addition, when creating a corrected image, the size, position, or direction of the detection target shadow image can be corrected by a single conversion by performing enlargement or reduction, translation, and rotation using affine transformation. it can.

  Further, by displaying the detection target shadow image (abnormal shadow candidate) determined to be an abnormal shadow image, its size information, the parallel movement amount in position correction, or the rotation angle in direction correction, these pieces of information are displayed. Can inform the user.

  In the first embodiment, as shown in FIG. 2, the case where the cutout areas of the respective reference images are unified to a predetermined size has been described. However, the size of the cutout area of each reference image is as follows. It does not necessarily have to be unified.

[Second Embodiment]
Next, a second embodiment to which the present invention is applied will be described.
The medical image processing apparatus according to the second embodiment has the same configuration as that of the medical image processing apparatus 1 shown in the first embodiment. Description is omitted. Hereinafter, a configuration and processing characteristic of the second embodiment will be described.

  The CPU 2 extracts a detection target shadow image from the image data, corrects the size, position, or direction of the detection target shadow image to the size, position, or direction corresponding to the reference image, creates a corrected image, Comparing with the corrected image, and determining whether the detection target shadow image is an abnormal shadow image or a normal image based on the comparison result and the analysis result of the detection target shadow image before correction; Types of abnormal shadow candidates are detected (see FIG. 9).

Next, the operation will be described.
FIG. 9 is a flowchart for explaining the abnormal shadow candidate detection process 2 executed by the medical image processing apparatus according to the second embodiment. Hereinafter, it is assumed that the subject of the processing of each step is the CPU 2 unless otherwise specified.

  As shown in FIG. 9, first, image data is input from the image generation device G via the I / F unit 3 (step S <b> 21) and stored in the storage unit 9.

  Next, the image data stored in the storage unit 9 is subjected to image analysis, and a detection target shadow image that may be an abnormal shadow such as a mass shadow or a microcalcification cluster is extracted (step S22).

  Next, the size, position, and direction of the detection target shadow image are corrected to the size, position, or direction corresponding to the reference image, and a corrected image is created (step S23). Then, a feature amount is extracted from the corrected image (step S24). The correction image creation method and feature amount extraction are the same as those in the first embodiment, and thus the description thereof is omitted.

  Also, feature values such as circularity, contrast with the background image, intensity component of density gradient from the periphery to the center of the shadow are extracted from the detection target shadow image before correction (step S25).

  Then, using the feature amount extracted from the corrected image and the feature amount extracted from the detection target shadow image before correction as input data for multivariate analysis, the detection target shadow image is an abnormal shadow image or a normal image Is determined (step S26). The feature amount showing the feature that varies depending on the size, position, and direction of the image is determined by comparing the feature amount extracted from the corrected image with the feature amount of the reference image stored in the reference image database 10, Other feature quantities are determined using the feature quantities extracted from the detection target shadow image before correction.

Then, the detection target shadow image determined to be an abnormal shadow image is detected as an abnormal shadow candidate (step S27), and the detection result is displayed on the display means 5 (step S28).
Thus, the abnormal shadow candidate detection process 2 ends.

  According to the medical image processing apparatus in the second embodiment, the size, position or direction of the detection target shadow image is corrected to the size, position or direction corresponding to the reference image, and the reference image and the corrected image are compared. Since it is determined whether the detection target shadow image is an abnormal shadow image or a normal image based on the comparison result and the analysis result of the detection target shadow image before correction, the cost is reduced by using fewer reference images. Therefore, it is possible to accurately determine whether the detection target shadow image is an abnormal shadow image or a normal image.

  The description in each of the above embodiments is an example of a suitable medical image processing apparatus according to the present invention, and the present invention is not limited to this. The detailed configuration and detailed operation of each part constituting the medical image processing apparatus can be changed as appropriate without departing from the spirit of the present invention.

It is a block diagram which shows schematic structure of the medical image processing apparatus 1 in 1st Embodiment. It is a figure which shows the example of a reference | standard image. It is a flowchart which shows the abnormal shadow candidate detection process 1 performed by the medical image processing apparatus 1 in 1st Embodiment. It is a flowchart which shows a correction image creation process. It is a figure which shows the example of a correction | amendment image. It is a figure which shows the example of the local pattern used for a texture analysis. It is a figure which shows the example of the image data used for a texture analysis. It is a conceptual diagram of Mahalanobis distance in discriminant analysis. It is a flowchart which shows the abnormal shadow candidate detection process 2 performed by the medical image processing apparatus in 2nd Embodiment.

Explanation of symbols

1 medical image processing apparatus 2 CPU
3 I / F section 4 Operating means 5 Display means 6 Communication section 7 ROM
8 RAM
9 Storage Unit 10 Reference Image Database 11 Bus

Claims (12)

Comparing means for comparing a reference image classified in advance into an abnormal shadow image and a normal image and a detection target shadow image extracted from a medical image obtained by photographing a subject, the comparison means In the medical image processing apparatus configured to determine whether the detection target shadow image is an abnormal shadow image or a normal image based on a comparison result,
Image correction means for correcting the size, position or direction of the detection target shadow image to a size, position or direction corresponding to the reference image, and outputting the corrected detection target shadow image to the comparison means ;
The image correcting means obtains a principal axis angle using a moment or a geometric feature of the detection target shadow image, and corrects the direction of the detection target shadow image based on the obtained principal axis angle. Medical image processing apparatus.   The medical image processing apparatus according to claim 1, wherein there are a plurality of the reference images, and the sizes, positions, and directions of the reference images are unified.   The medical image processing apparatus according to claim 1, wherein the medical image is a mammogram. Wherein the image correcting means, enlarged or reduced by using the affine transformation, by performing movement or rotational movement parallel claim 1-3, characterized in that to correct the size, position or orientation of the detected shadow image The medical image processing apparatus according to any one of the above. 2. The determination as to whether the detection target shadow image is an abnormal shadow image or a normal image is performed based on a comparison result by the comparison unit and an analysis result of the detection target shadow image before correction. The medical image processing apparatus according to any one of to 4 . The amount of parallel movement in the correction of the position of the detection target shadow image determined to be the abnormal shadow image, the rotation angle in the correction of the direction of the detection target shadow image determined to be the abnormal shadow image, and the abnormal shadow image it is with the determined detection target shaded image, or the abnormality any one of claims 1 to 5 in which size information of the determined detection target shadow image as the shadow is an image, further comprising a display means for displaying, characterized in The medical image processing apparatus according to Item. A comparison function for comparing a reference image previously classified into an abnormal shadow image and a normal image on a computer with a detection target shadow image extracted from a medical image obtained by photographing a subject, and the comparison function In a medical image processing program for realizing a discrimination function for discriminating whether the detection target shadow image is an abnormal shadow image or a normal image based on the comparison result in
In the computer,
Realizing an image correction function for correcting the size, position or direction of the detection target shadow image to the size, position or direction corresponding to the reference image;
When realizing the comparison function, the reference image and the corrected shadow image to be detected are compared ,
When realizing the image correction function, the principal axis angle is obtained using the moment or geometric feature of the detection target shadow image, and the direction of the detection target shadow image is corrected based on the obtained principal axis angle. A medical image processing program characterized by that. The medical image processing program according to claim 7 , wherein there are a plurality of the reference images, and the sizes, positions, and directions of the reference images are unified. 9. The medical image processing program according to claim 7 , wherein the medical image is a mammogram. When realizing the image correction function, the size, position, or direction of the shadow image to be detected is corrected by performing enlargement or reduction, translation, or rotation using affine transformation. The medical image processing program according to any one of claims 7 to 9 . When realizing the discrimination function, it is discriminated whether the detection target shadow image is an abnormal shadow image or a normal image based on the comparison result in the comparison function and the analysis result of the detection target shadow image before correction. The medical image processing program according to any one of claims 7 to 10 . In the computer,
The amount of parallel movement in the correction of the position of the detection target shadow image determined to be the abnormal shadow image, the rotation angle in the correction of the direction of the detection target shadow image determined to be the abnormal shadow image, and the abnormal shadow image any one of claims 7 to 11, characterized in that to realize the display function of displaying the size information of a judged to be the detection target shaded image, or the abnormal shadow detection target shadow images it is judged that image The medical image processing program according to item.

Images