JP6987342B2 - Image processing equipment, methods and programs - Google Patents

Description

The present invention relates to an image processing apparatus for processing a medical image, a method and a program thereof.

In diagnosis using a medical image inspection device represented by an X-ray CT (X-ray Computed Tomography) device or an MRI (Magnetic Response Imaging) device, the captured three-dimensional medical image is reproduced as a continuous two-dimensional cross section. It is common to construct and observe the two-dimensional cross-sectional image for interpretation.

With the sophistication of imaging equipment, the three-dimensional resolution of the generated three-dimensional medical image is also improved, and the data size tends to increase. In particular, it has become possible to make the generation interval of the two-dimensional cross section described above finer, and it has become possible to observe the lesions appearing on the medical image in more detail. The number of sheets is also increasing.

Further, in particular, in a CT apparatus, it has become possible to take a three-dimensional medical image with a low dose and high image quality, and the opportunity to take a CT image is increasing.

For these reasons, a technology called CAD (Computer Aided Detection) is being developed to reduce the burden on doctors and technicians when interpreting a huge amount of medical three-dimensional medical images and mainly to prevent oversight of lesions. .. CAD aims to automatically or semi-automatically detect and measure the size of shadows, identify normal / abnormal shadows, and distinguish lesion types of abnormal shadows by applying image processing technology.

Here, CAD for the purpose of presenting a shadow with a high suspicion of a lesion due to the characteristics of a medical image will be described. With the development of image processing technology, especially in the field of image processing, CAD has improved the accuracy. On the other hand, the image features of lesion images in medical images are diverse even if they have the same disease name, and when identifying abnormal shadows mainly using machine learning, the accuracy required due to insufficient learning data is obtained. There was a problem that it could not be achieved.

In order to solve these problems, Patent Document 1 uses a plurality of classifiers that classify input data into two or more classes based on different image features as a method for improving the identification accuracy of facial images. A method of classifying into classes and adjusting the parameters of the classifier based on multiple attribute information has been proposed.

Japanese Unexamined Patent Publication No. 2017-102671

In the automatic detection of a suspicious lesion shadow by CAD, a definition formula that defines the suspicion of a lesion is generally set using a feature amount obtained from an image, and a shadow with a high suspicion of the lesion is presented. When setting the definition formula here, when referring to the information of the past image that has already been diagnosed as a lesion, the characteristics of the image are diverse even if the disease name is the same, and the past that has been diagnosed. There is a problem that the accuracy of the definition formula becomes low because the amount of images in the image is not sufficient. Further, in the technique of labeling a plurality of image data, the techniques proposed so far as a countermeasure against the problem that the appearance changes greatly even if the learning data has the same label are based on different image features. Although it is possible to adjust the parameters of the classifier based on the multiple attribute information set in, the attribute information used as the basis for classification may not be the optimum setting in all cases. There are challenges.

An object of the present invention is to provide a medical image processing device, a medical image display method, and a program capable of setting a highly accurate definition formula of suspicion of a lesion even when the amount of past images to be referred to is insufficient. To provide.

In particular, we provide an effective technique when the detection rate of lesions is low in the setting of the definition formula by the conventional technique, for example, when the nodule detection rate is less than about 60% when the radiologist interprets the image.

In order to solve the above problems, the present invention is an image processing device that processes a medical image, and displays a plurality of suspected lesion regions and a plurality of feature quantities of the suspected lesion region in the medical image. Provided is an image processing apparatus having a configuration including an information display unit and a lesion suspected area classification unit that classifies a lesion suspected area according to a feature amount selected from a plurality of displayed feature amounts.

Further, in order to solve the above-mentioned problems, the present invention is an image processing method for processing a medical image by an image processing apparatus, wherein the image processing apparatus has a plurality of suspected lesion regions and suspected lesion regions in the medical image. Provided is an image processing method for displaying a plurality of feature quantities of the above on a monitor and classifying the suspected lesion region according to the feature quantities selected from the plurality of displayed feature quantities.

Further, in order to solve the above-mentioned problems, in the present invention, an image processing program for causing an image processing device to process a medical image, wherein the image processing device has a plurality of suspected lesion regions and suspected lesion regions in the medical image. Provided is an image processing program that displays a plurality of feature quantities of the above on a monitor and classifies the suspected lesion region according to the feature quantities selected from the displayed plurality of feature quantities.

According to the present invention, it is possible to set a highly accurate definition formula for suspicion of a lesion even when the amount of past medical images to be referred to is insufficient.

It is a figure which shows the system configuration example including the medical image processing apparatus which concerns on Example 1. FIG. It is a figure which shows the structural example of the lesion suspected area classification part and the lesion likelihood setting part which concerns on Example 1. FIG. It is a figure which shows an example of the lesion suspected area classification type definition processing flow which concerns on Example 1. FIG. It is a figure which superposed and displayed the feature amount of each lesion suspected area and the image of the corresponding lesion suspected area which concerns on Example 1. FIG. It is a figure which superposed and displayed the other feature amount of each lesion suspected area which concerns on Example 1. on the image of a human body model. It is a figure which shows the screen display example for determining the necessity of classification from the display by the lesion suspicion information display part which concerns on Example 1. FIG. It is a figure which shows an example of the lesion likelihood calculation formula which concerns on Example 1. FIG. It is a figure which shows an example of the lesion likelihood calculation formula definition processing flow which concerns on Example 1. FIG. It is a figure which shows an example of the lesion likelihood calculation formula which concerns on Example 1. FIG. It is a figure which shows an example of the lesion likelihood calculation processing flow which concerns on Example 1. FIG. It is a figure for demonstrating the problem of the present medical image processing technology. It is a figure for demonstrating the evaluation of the evaluation data at the time of using the medical image processing apparatus which concerns on Example 1. FIG.

Hereinafter, embodiments of the image processing apparatus of the present invention will be sequentially described with reference to the drawings. The image processing apparatus of the present invention processes a medical image to classify a suspected lesion area on the medical image, and the suspected lesion area is the medical knowledge of an image interpreter and the medical basis (evidence) for disease diagnosis. It means a point and area with a high suspicion of lesion, which is judged based on. Here, it is highly possible that the lesion to be diagnosed can be determined from the difference in brightness and the difference in its distribution from the surrounding area, that is, the area where the suspicion of the lesion is low, when it appears on the medical image.

The first embodiment is an image processing apparatus for processing a medical image, which includes a plurality of suspected lesion regions in the medical image, a suspected lesion information display unit for displaying a plurality of feature quantities of the suspected lesion region, and a plurality of displayed parts. This is an example of an image processing device having a configuration including a suspected lesion area classification unit for classifying suspected lesion areas according to the feature amount selected from the feature amounts of the above, and the amount of past medical images to be referred to is insufficient. This is an example of a medical image processing apparatus, a method, and a program capable of setting a highly accurate definition formula of suspicion of a lesion and detecting a suspicious part of a lesion with high accuracy.

In this embodiment, a reconstructed three-dimensional medical image obtained by a CT medical imaging device will be described, but this technique can also be applied to data obtained by another medical imaging device. For example, even if the data is obtained by an MRI imaging device or the like, a three-dimensional image that can be expressed as a stack of a plurality of two-dimensional cross sections is obtained, and if it is said that lesion features appear in the pixel distribution, it should be applied. Can be done.

As described above, the lesion suspicious area in this example is a point and area where there is a high suspicion of a lesion, which is determined based on the medical knowledge of the image interpreter and the medical basis (evidence) for the diagnosis of the disease. Refers to the part. Here, it is highly possible that the target lesion can be determined from the difference in brightness and the difference in distribution from the surrounding area, that is, the area where the suspicion of the lesion is low, when it appears on the medical image. For example, in the case of a lung nodule, it is generally known that its CT value appears on a CT image as a region containing many pixels higher than the surrounding air region. Other objects that contain a lot of high-intensity pixels on the chest CT image are blood vessels and bones, which are distinguished from them according to the distribution shape of the high-intensity values, and are suspected to be lesions, that is, lung nodules in this example. It is said that the height can be determined.

FIG. 1 shows an example of a system configuration including the medical image processing apparatus according to the present embodiment. As shown in the figure, the system of this embodiment is obtained from a medical image processing device 11, an input device 10 that receives an operator's input and the like and transmits the input to the medical image processing device 11, and a medical image processing display device 11. It is composed of a medical image and a monitor 12 for displaying suspected lesion information. This system configuration can be configured by an ordinary computer, and the medical image processing apparatus can be realized by a central processing unit (CPU) that executes various processing programs and a memory that stores those processing programs and data.

The medical image processing device 11 includes a medical image storage unit 20 that stores medical images, a lesion suspicious information storage unit 21 that stores lesion suspicious information corresponding to each medical image, and a plurality of medical images and lesion suspicious information at once. It is composed of a lesion suspicion information display unit 22 displayed on the monitor 12, a lesion suspicion area classification unit 23 for classifying the lesion suspicious area, and a lesion likelihood setting unit 24 for setting the lesion likelihood for the classified lesion suspicion area. There is.

FIG. 2 shows an example of the system configuration of the lesion suspected region classification unit 23 and the lesion likelihood setting unit 24. As shown in the upper part of the figure, the lesion suspected region classification unit 23 receives the input from the input device 10 and defines the lesion suspected region classification formula, and the lesion suspected region classification formula definition unit 30 and the lesion suspicion that stores the formula. It is composed of a region classification type storage unit 31 and a lesion suspected area classification execution unit 32 that classifies the input lesion suspected area according to the lesion suspected area classification formula.

As shown in the lower part of FIG. 2, the lesion likelihood setting unit 24 includes a lesion likelihood calculation formula definition unit 33 that defines a lesion likelihood calculation formula for each classified group from information on the classified suspected lesion region. It is composed of a lesion likelihood calculation formula storage unit 34 that stores the formula and a lesion likelihood calculation unit 35 that calculates the lesion likelihood of the input lesion suspected region according to the lesion likelihood calculation formula.

Next, the flow of processing by the medical image processing apparatus of this embodiment having the system configuration shown in FIGS. 1 and 2 will be described. Here, the medical image processing device 11 that receives the input from the system or the instruction from the user stores the lesion suspected area classification formula in the lesion suspected area classification formula storage unit 31 and the lesion suspected region classification formula definition processing from the system. The medical image processing device 11 that receives the input of the above or the instruction from the user divides into the lesion likelihood calculation formula definition process that stores the lesion likelihood calculation formula in the lesion likelihood calculation formula storage unit 34, and each of them will be described separately. ..

First, the lesion suspected region classification formula definition process will be described using the process flow of FIG. The lesion suspicion information display unit 22 acquires a plurality of medical images and corresponding lesion suspicion information from the medical image storage unit 20 and the lesion suspicion information storage unit 21 by input from the system or instructions from the user, and is used for medical purposes. Each of the plurality of feature quantities of the suspected lesion region in the image is calculated and displayed on the monitor 12 in an intuitive and easy-to-understand format (step 101).

Next, the lesion suspicious area classification formula definition unit 30 of the lesion suspicious area classification unit 23 acquires whether or not there is a need for classification for each of the plurality of feature quantities by input from the user to the input device 10 (step 102). A lesion suspicious area classification formula is defined based on the above (step 103), and is stored in the lesion suspicious area classification formula storage unit 31 (step 104).

Next, an example of a method of displaying the suspected lesion region and the feature amount by the lesion suspicion information display unit 22 will be described with reference to FIGS. 4 and 5. FIG. 4A shows the brightness of each suspected lesion area, and FIG. 4B calculates the size of each suspected lesion area as a feature amount, and a histogram of the brightness or size of these feature amounts and a histogram. The image of the representative suspected lesion area corresponding to the value of is superimposed and displayed.

Further, (a) of FIG. 5 features each suspected lesion region and a two-dimensional position in the image, and (b) of FIG. 5 features each suspected lesion region and the axial position in the volume data of the human body as a subject. It is calculated as, and superimposed and displayed on the cross-sectional image of the human body model. With the display screens as shown in FIGS. 4 and 5, the user can grasp what kind of characteristic the suspected lesion region exists for each feature amount and at what ratio. The dotted line in FIGS. 4 and 5 indicates a threshold value for classifying the suspected lesion region, which will be described later.

FIG. 6 is a diagram showing an example of a display screen for the user to determine the necessity of classification from the display on the monitor 12 by the lesion suspicion information display unit 22 of the medical image processing device 11. As shown in (a), (b), (c), and (d) of FIG. 6, on the display screen of the monitor 12, for example, a check box 34 beside the superimposed display image shown in FIGS. 4 and 5. Is displayed. Then, the user inputs 35 to the feature amount that is determined to require classification. The lesion suspicious area classification formula definition unit 30 of the lesion suspicious area classification unit 23 acquires the presence or absence of the feature quantity check, and reflects the information in the classification necessity judgment table showing an example in (e) of FIG. , Acquire information on the necessity of classification and define the lesion suspected area classification formula.

Here, the lesion suspected region classification formula of the lesion suspected region classification formula definition unit 30 will be described with reference to a specific example. This lesion suspected region classification formula is a formula for classifying a plurality of lesion suspected regions into a plurality of groups according to their feature amounts. For example, when a plurality of suspected lesion regions ROI [n] (n = 0 to N) can be represented by the center position (Cx [n], Cy [n]) of the region and the radius R [n]. When the feature amount determined to be in need of classification in step 102 of 3 is a two-dimensional position in the image as shown in the classification necessity determination table of FIG. 6 (e), the lesion suspected region is defined. The case of classifying into two groups of G0 and G1 will be described.

FIG. 7 shows an example of the lesion suspected region classification formula defined in this case and stored in the lesion suspected region classification formula storage unit 31 as Equation 1. In this equation 1, the vertical axis in the two-dimensional image increases toward the bottom of the image, and the lesion suspected area classification execution unit 32 has the center position in the image above the threshold value (threshold). At G0, a suspected lesion area equal to or below the threshold is meant to be classified as G1. This threshold value may be determined in advance, or may be automatically obtained by the average value or median value of the feature amount (here, Cx [n]) calculated from the suspected lesion region, or by a method such as a discriminant analysis method. Alternatively, it may be adjusted and decided by the user. In FIGS. 4 and 5, the threshold value is shown by a dotted line, but if the user interface (UI) is set so that the user can move the dotted line on the screen, the user can easily adjust the threshold value using the display screen of the monitor 12. Can be done.

Next, with reference to FIG. 8, the lesion likelihood calculation formula definition process in the lesion likelihood calculation formula definition unit 33 of the lesion likelihood setting unit 24 will be described. The medical image processing device 11 acquires one or more suspected lesion regions and medical images corresponding to the regions from the suspected lesion information storage unit 21 and the medical image storage unit 20 by input from the system or instructions from the user. Then, a suspected lesion area with image information is generated (step 201).

The lesion suspected region classification execution unit 32 classifies these lesion suspected region with medical image information into a plurality of groups according to the lesion suspected region classification formula as shown in FIG. 7 (step 202).

Next, the lesion likelihood calculation formula definition unit 33 grasps the feature amount indicating the image feature for the lesion suspected region with medical image information classified into groups, and defines the lesion likelihood calculation formula for each classified group ( Step 203), store in the lesion likelihood calculation formula storage unit 32 (step 204).

The lesion likelihood calculation formula defined by the lesion likelihood calculation formula definition unit 33 will be described with reference to a specific example. In the lesion likelihood calculation formula, the lesion likelihood Lprev [n] of the suspected region is calculated for each of a plurality of groups classified by the lesion suspect region classification process using the lesion likelihood region whose lesion likelihood is known in advance. It is a formula to calculate. In the case where a plurality of suspected lesion regions ROI [n] are classified into two groups of G0 and G1, the group G0 will be described.

Here, the suspected lesion region classified as G0 is set to ROI_0 [n (0 to N_0)], the lesion likelihood of ROI_0 [n] is known in advance, and the definite lesion likelihood L [n] is 0 or 1. It is assumed that it is represented by the two values. The definite lesion likelihood L [n] represents a lesion suspected region where L [n] = 1 was a lesion according to a doctor's diagnosis, and L [n] = 0 represents another, that is, a lesion suspected region which was not a lesion.

FIG. 9 shows an example of the lesion likelihood calculation formula in this case as Equation 2. In Equation 2, π is the pi, R [n] is the radius of the suspected lesion region ROI [n], P [n] is the average brightness of the suspected lesion region ROI [n], and Amax is the area that ROI [n] can take. The maximum value of, Pmax, indicates the maximum value of the average brightness that ROI [n] can take. According to this equation 2, the suspected lesion area has a larger area and higher brightness, which is more likely to be a lesion.

In this equation 2, the lesion likelihood Lprev [n] takes a continuous value, but a multi-valued one using a threshold value can be used as the final lesion likelihood. Here, for the sake of explanation, a simple lesion likelihood calculation formula is given as an example, but the design of this formula is based on the so-called rule base, and the lesion likelihood region that was truly a lesion and the lesion likelihood suspicion that was not. It is possible to judge the difference in the feature amount of the region and manually design from each possible range, or it is possible to design by using machine learning such as SVM (Support Vector Machine) or Deep Learning.

By the processing of the medical image processing apparatus 11 described above, it is possible to set a highly accurate definition formula of suspicion of a lesion, that is, a lesion likelihood calculation formula even when the amount of past images to be referred to is insufficient. become.

Next, using the lesion likelihood calculation processing flow of FIG. 10, the process in which the lesion likelihood is calculated by the lesion likelihood calculation formula set in this way and stored in the lesion likelihood calculation formula storage unit 34, that is, the system. The lesion likelihood calculation process for calculating the lesion likelihood by the lesion likelihood setting unit 24 by the medical image processing apparatus 11 receiving the input from or instructed by the user will be described.

The medical image processing device 11 acquires a suspected lesion area to be processed and a medical image corresponding to the area from the suspected lesion information storage unit 21 and the medical image storage unit 20 by input from the system or an instruction from the user. Then, it is set as a suspected lesion area with image information (step 301).

Next, the lesion suspicious area classification execution unit 32 determines the group to which the lesion suspicious area with image information to be processed belongs according to a preset medical image classification formula (step 302).

Next, the lesion likelihood calculation unit 35 calculates the lesion likelihood according to the lesion likelihood calculation formula corresponding to the group to which the lesion suspicious region with image information belongs (step 303). The lesion likelihood calculation formula illustrated in FIG. 9 exists as many as the number of groups classified by the lesion suspicious region classification unit 23. In step 303, the plurality of lesion likelihoods obtained from these plurality of calculation formulas are converted into one lesion likelihood by logical sum or logical product, and then output.

Here, a specific example of the problem solved by using the medical image processing apparatus 11 of the present embodiment described above will be described. First, problems that occur in the current medical image processing apparatus will be described with reference to FIG.

In the figure, figures such as circles, triangles, and squares schematically show lesion-suspicious areas of learning data and evaluation data. In the learning data shown in FIG. 11A, the suspected lesion region indicated by the dotted circle is defined as the region that is truly a lesion. Considering the difference between the area indicated by the dotted circle and the area not indicated by the dotted line in this training data, it can be considered that only the area where the shape is a circle and the color is black is truly a lesion. Therefore, it is considered that an equation is designed in which the region having a round shape and a black color has a high nodule likelihood.

When the lesion likelihood of the evaluation data shown in FIG. 11 (b) is calculated using this lesion likelihood calculation formula, the result is that the lesion likelihood in the region where the shape is round and the color is black is high. .. This result is shown as an evaluation by a classifier. On the other hand, the figure shown as an evaluation by a doctor is a result assuming that the same evaluation data is evaluated by a doctor. As described above, the result is different from the evaluation by the discriminator, and there is a region indicated by a dotted circle in the region other than "the shape is a circle and the color is black". As described above, with the current technology, if the variation of the learning data is not sufficient, there is a high possibility that a lesion likelihood calculation formula that correctly captures the characteristics of the nodule cannot be designed.

Next, a case where the medical image processing apparatus 11 of this embodiment is used will be described with reference to FIG. It is assumed that (a) and (b) of FIG. 12 use the same learning data and evaluation data as in the case of FIG. Further, in the example described here, the lesion suspected region classification formula described in the formula 1 of FIG. 7 is used. The dotted line in the training data of FIG. 12A indicates the threshold value described in FIG. 7. Therefore, G0 is a set of suspected lesion regions above the threshold value, and G1 is a set of suspected lesion regions below the threshold value. When the lesion likelihood calculation formulas are designed for each of G0 and G1, it is considered that the result is that the lesion likelihood of the suspected lesion region is high in the case of G0, which is “round shape” and in the case of G1, which is “black in color”.

Here, the lesion likelihood calculation unit 35 using the lesion likelihood calculation formula designed by G0 is referred to as the discriminator 0, and the lesion likelihood calculation unit 35 using the lesion likelihood calculation formula designed by G1 is referred to as the discriminator 1. .. The left part of FIG. 12B shows the evaluation results for the evaluation data assumed by the discriminator 0 and the discriminator 1. Here, the logical sum of the results obtained by a plurality of classifiers is used as the final output. The final identification result is shown in the right part of FIG. 12 (b). In this way, by incorporating the user's knowledge that the medical image processing apparatus of this embodiment should set different lesion likelihood calculation formulas for the two-dimensional position in the image of the suspected lesion region, here, the upper part and the lower part of the image. , It is possible to obtain a highly accurate identification result that matches the identification result by the user.

According to the medical image processing apparatus of Example 1 described in detail above, the past to be referred to when referring to the information of the past image that has already been diagnosed as a lesion in the automatic detection of the suspected lesion shadow. It is possible to provide a medical image processing device, a method, and a program capable of setting a highly accurate definition formula of suspicion of a lesion even when the amount of images in the image is insufficient. In particular, it is possible to provide an effective technique when the detection rate of lesions is low in the setting of the definition formula by the conventional technique, for example, when the nodule detection rate is less than about 60% when the radiologist interprets the image.

The processing technique of the medical image processing apparatus 11 of the first embodiment can also be used as an initial adjustment of accuracy before the operation of CAD. In this case, the lesion suspected area classification formula and the lesion likelihood calculation formula are set in advance to the accuracy considered to be empirically appropriate, and at the time of CAD delivery, according to the needs of the user at that time, the interpretation policy of the facility, etc. It is supposed to be used for adjustment. The initial values of the suspected lesion area classification formula and the lesion likelihood calculation formula can be set manually, or the available medical image database can be used to define the suspected lesion area classification formula and the lesion. It can also be set by executing the likelihood calculation formula definition process.

Here, the lesion suspicious area classification formula definition process and the lesion likelihood calculation formula definition process set before adjustment, that is, at the time of shipment, can be performed using the medical image database acquired at the delivery destination facility, or can be performed. If it is possible to use a medical image database acquired at another facility that has an image interpretation policy close to that of the delivery destination, it may be used. Alternatively, as a result of adjustment at another facility having an image interpretation policy close to the delivery destination facility, the lesion suspected area classification formula and the lesion likelihood calculation formula may be used.

When the processing technique described above is used as the initial adjustment, clinical data taken in the past can be used as the medical image used for presenting and adjusting the certainty. In this case, it is possible to perform more efficient adjustment by classifying in advance according to the type of lesion suspected area information and showing a lesion suspected shadow such that the feature amount used in the certainty calculation formula is significantly different. .. Alternatively, the medical image used here does not necessarily have to be an actual clinical image. For example, phantom data obtained by photographing a phantom having a shape and material close to the human body can be used, or a lesion suspected position can be simulated from the clinical image. It is also possible to use the simulated data duplicated and generated.

Further, it can be used not only for the initial adjustment before the operation of the CAD but also for the adjustment to be performed at any time during the operation of the CAD. When performing initial adjustment as described above, past images, phantom images, simulated images, etc. are used as the data to be used, but when using for adjustment to be performed at any time during operation, the shooting device and shooting policy of the facility concerned. It is possible to make adjustments using the images taken according to the above, and there is an effect that detection can be performed with an accuracy closer to the user's desire.

The present invention is not limited to the above-described embodiment, and includes various modifications. For example, the above-described embodiment has been described in detail in order to explain the present invention in an easy-to-understand manner, and is not necessarily limited to the one including all the described configurations. As described above, the present invention can be applied not only to reconstructed three-dimensional medical images obtained by a CT medical imaging device but also to other medical image data obtained by an MRI imaging device or the like. Further, it is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. Further, it is possible to add / delete / replace a part of the configuration of each embodiment with another configuration. In Example 1, the medical image processing device 11 does not include a medical image taking device (not shown), but the medical image processing device 11 may include a medical image taking device, and the medical image processing device 11 is for medical use. It may function as part of an imaging device.

Further, for each of the above-mentioned configurations, functions, medical image processing devices, etc., an example of creating a processing program that realizes a part or all of them has been described, but a part or all of them is designed, for example, by an integrated circuit. Needless to say, it may be realized by hardware. That is, all or part of the functions of the image processing apparatus may be realized by, for example, an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array) instead of the program.

As is clear from the description of the specification described in detail above, a large number of inventions are included in the present specification in addition to the inventions defined in the claims. Some of them are listed below.

[List 1]
An image processing method for processing medical images with an image processing device.
The image processing device is
A plurality of suspected lesion areas in the medical image and a plurality of feature quantities of the suspected lesion area are displayed on the monitor.
The suspected lesion area is classified according to the feature amount selected from the displayed plurality of feature amounts.
The feature amount includes the brightness and size of the suspected lesion area.
An image processing method characterized by that.

[List 2]
The image processing method according to the list 1.
The image processing device is
A histogram of the feature amount is displayed on the monitor.
An image processing method characterized by that.

[List 3]
The image processing method according to the list 1.
The feature amount includes the two-dimensional position of the suspected lesion region in the medical image.
An image processing method characterized by that.

[List 4]
The image processing method according to the list 1.
The feature amount includes the body axis position of the subject in the volume data of the suspected lesion region.
An image processing method characterized by that.

[List 5]
An image processing program that causes an image processing device to process medical images.
The image processing device
A plurality of suspected lesion areas in the medical image and a plurality of feature quantities of the suspected lesion area are displayed on the monitor.
The suspected lesion area is classified according to the feature amount selected from the displayed plurality of feature amounts.
The feature amount includes the brightness and size of the suspected lesion area.
An image processing program characterized by this.

[List 6]
The image processing program described in Listing 5
The image processing device
Operate to display the histogram of the feature amount on the monitor.
An image processing program characterized by this.

[List 7]
The image processing program described in Listing 5
The feature amount includes the two-dimensional position of the suspected lesion region in the medical image.
An image processing program characterized by this.

[List 8]
The image processing program described in Listing 5
The feature amount includes the body axis position of the subject in the volume data of the suspected lesion region.
An image processing program characterized by this.

[List 8]
The image processing program described in Listing 5
The image processing device
An image processing program characterized in that the lesion likelihood of the suspected lesion region belonging to the group is calculated for each group of the suspected lesion region classified.

[List 9]
The image processing program described in Listing 8
The image processing device
Image processing characterized in that a lesion suspected region classification formula is defined using the selected feature amount, and a plurality of the lesion suspected regions are classified into the group based on the lesion suspected region classification formula. program.

[List 10]
The image processing program described in Listing 8
The image processing device ...
A lesion likelihood calculation formula is defined for each group based on the characteristic amount of the lesion suspected region classified into the group, and the lesion likelihood of the lesion suspected region is calculated based on the lesion likelihood calculation formula. An image processing program characterized by operating like,.

10 Input device 11 Medical image processing device 12 Monitor 20 Medical image storage unit 21 Lesion suspicious information storage unit 22 Lesion suspicious information display unit 23 Lesion suspicious area classification unit 24 Lesion likelihood calculation unit 30 Lesion suspicious area classification formula definition unit 31 Lesion suspicion Area classification formula Storage unit 32 Lesion likelihood calculation formula Definition unit 33 Lesion likelihood calculation formula Storage unit 34 Check box 35 User input

Claims (7)

An image processing device that processes medical images.
A plurality of suspected lesion areas in the medical image, a lesion suspected information display unit for displaying a plurality of feature quantities of the suspected lesion area, and a lesion information display unit.
A lesion suspected area classification unit that classifies the lesion suspected area according to the feature amount selected from the displayed plurality of feature amounts, and a lesion suspected area classification unit.
For each group classified by the suspected lesion region classification unit, a lesion likelihood setting unit for calculating the lesion likelihood of the suspected lesion region belonging to the group is provided.
An image processing device characterized by this. The image processing apparatus according to claim 1.
The feature amount includes the brightness and size of the suspected lesion area.
An image processing device characterized by this. The image processing apparatus according to claim 2.
The lesion suspicion information display unit displays a histogram of the feature amount.
An image processing device characterized by this. The image processing apparatus according to claim 2.
The feature amount includes the two-dimensional position of the suspected lesion region in the medical image.
An image processing device characterized by this. The image processing apparatus according to claim 2.
The feature amount includes the body axis position of the subject in the volume data of the suspected lesion region.
An image processing device characterized by this. An image processing method for processing medical images with an image processing device.
The image processing device is
A plurality of suspected lesion areas in the medical image and a plurality of feature quantities of the suspected lesion area are displayed on the monitor.
The suspected lesion area is classified according to the feature amount selected from the displayed plurality of feature amounts.
For each group of the suspected lesion regions classified, the lesion likelihood of the suspected lesion region belonging to the group is calculated.
An image processing method characterized by that. An image processing program that causes an image processing device to process medical images.
The image processing device
A plurality of suspected lesion areas in the medical image and a plurality of feature quantities of the suspected lesion area are displayed on the monitor.
The suspected lesion area is classified according to the feature amount selected from the displayed plurality of feature amounts.
For each group of the suspected lesion regions classified, the lesion likelihood of the suspected lesion region belonging to the group is calculated .
An image processing program characterized by this.

Images