JP7429437B2 - Image processing device, image processing method, and image processing program - Google Patents

Description

The present invention relates to an image processing apparatus such as a nuclear magnetic resonance imaging apparatus (hereinafter referred to as an MRI apparatus), an image processing method, and an image processing program, and more particularly, the present invention relates to a method for dividing image regions of an MRI image of a subject and its method. The present invention relates to a method for supporting condition evaluation in areas that attract attention.

In MRI images, images with various contrasts can be obtained, such as T1-weighted images and T2-weighted images. In recent years, it has become possible to take diffusion-weighted images (also known as DWI), which focus on the diffusion restriction of hydrogen atoms (protons), and are used to visualize cerebral infarctions and detect tumors. In particular, diffusion-weighted imaging (Body DWI or DWIBS), which images the whole body, was developed to detect tumors throughout the body, and has come to be used to visualize and evaluate metastatic bone tumors caused by prostate cancer and breast cancer. It is also possible to calculate the diffusion coefficient (ADC) from the two diffusion-weighted images obtained by modulating the gradient magnetic field strength (b value), and this strength is said to be related to cell density, making it possible to judge the malignancy of a tumor. ing.

In such a method that contributes to diagnosis using MRI images, there have been several problems in identifying regions and conditions. For example, in a diffusion-weighted image taken with a high gradient magnetic field strength, the signal of a tumor is generally high, and the signals in other areas are erased, so the signal is displayed low. At this time, tumors often have low levels of diffusion coefficient (ADC) in areas with high signal values. However, the red pulp, which is a normal tissue (a normal hematopoietic tissue found in infants, normally changes to a lipid-rich yellow pulp as they grow up, but due to various factors, the yellow pulp becomes yellow as they become adults). The yellow pulp is completely adipose tissue, and the red pulp is a mixture of adipose tissue and water.) has a signal as high as that of a tumor, and has an ADC as high as that of a tumor known to be at low levels.
For this reason, even if an attempt is made to search for a tumor using diffusion-weighted images, it is difficult to distinguish and identify the red pulp, which is normal tissue, and the tumor.

In contrast, a discrimination method using a "water/fat signal cancellation method (phase difference method, WFS method, DIXON method, etc.)", which is a fat suppression method for MRI, has been proposed (see Non-Patent Document 1 below). This is a method for determining minute amounts of fat in tissues, and is often used in the field of diagnostic imaging to differentiate between adrenal tumors and adrenal adenomas. This is because tumors generally do not contain fat in their tissues, so this method is used based on the concept that ``if fat can be detected, adrenal gland tumors can be ruled out.'' It is said that by applying this method to metastatic bone tumors, it becomes possible to differentiate between tumors and red pulp.

An MRI apparatus has the advantage of being able to simultaneously acquire a plurality of contrast images of different types in a single examination, and has greatly contributed to the advancement of image diagnostic technology in recent years (see Non-Patent Document 2 below). For example, in general, there are T2-weighted images that emphasize both fat and water, T1-weighted images that emphasize only fat, and fat-suppressed images and water-suppressed images that reduce fat and water signals from each image. 2. Description of the Related Art A method is known in which an image or the like can be obtained by taking a single image, and a diagnosis is supported using a plurality of images of a region of interest of a subject.

Study on CT value and MRI signal evaluation of adrenal adenoma: Journal of Saitama Medical University, Volume 33, No. 2, April 2006 Tsushima Y, Ishizaka H, Matsumoto M. Adrenal masses: differentiation with chemical shift, fast low-angle shot MR imaging. Radiology 1993;186:705 - 9.

By the way, the above-mentioned "water/fat signal cancellation method" utilizes the fact that the signal of protons, which are hydrogen atoms, changes depending on the surrounding conditions (uniformity). It takes advantage of the fact that the signal values are different between a uniform state such as "only fat" or "only fat" and a non-uniform state such as "water and fat mixed." In other words, a tumor can be interpreted as having only water surrounding it, and a red pulp can be interpreted as having a mixture of water and fat, so the signal states between the two are different. You can distinguish between the two.

In the above-mentioned "water/fat signal cancellation method," the "in phase" image, which is the moment when the atomic spin rotation periods of water and fat match, and the moment when the rotation periods are exactly opposite, the most discrepant. The signal values are different from the "out of phase" image, which is the moment when the signal changes. In other words, in the in-phase, the spins of water and fat match, so the signal values of both are emphasized, whereas in the out-of-phase, the spins do not match, so if water and fat coexist in the same pixel, the signals of each other will be emphasized. is canceled out and decreases. By comparing these, it is thought that the presence of fat will be revealed and it will be possible to determine whether it is a tumor or a red pulp. In order to determine this state, the MRI apparatus acquires two types of image data, an "in phase image" and an "out of phase image", of the same region of interest at one time.

However, as mentioned above, since the signal states of the in-phase image and the out-of-phase image are different from each other, it has been possible to determine whether the target site is a tumor or the red pulp, but it is difficult to visually Because there are no objective indicators for cancer, diagnosis, such as how far a tumor has spread, has always been left to the subjective visual judgment of doctors. For this reason, there was a fear that accurate diagnosis would become difficult.
The present invention was made in view of the above circumstances, and provides an image processing device, an image processing method, and an image processing device that can make a clear and objective diagnosis based on image data without relying solely on a doctor's subjective visual judgment. Its purpose is to provide an image processing program.

Further, the image processing device of the present invention includes:
a difference image calculation unit that obtains difference image data by calculating difference values between corresponding pixels for two types of MRI image data taken of the same subject part;
a statistical information calculation unit that calculates statistical information based on each pixel value of the difference image data calculated by the difference image calculation unit;
a threshold calculation unit that calculates a threshold for dividing each pixel of the difference image into two groups based on the statistical information calculated by the statistical information calculation unit;
a pixel group classification unit that classifies each pixel of the difference image into two groups based on the threshold calculated by the threshold calculation unit;
a pixel value replacement unit that gives a value unique to each group as a pixel value to each pixel included in the group for each divided group, and obtains a pixel value replacement image;
Based on the signal value or region information obtained from a type of predetermined medical image data different from the two types of MRI image data, and the signal value or region information obtained from the data of the pixel value replaced image, a subject part state determination unit that determines the state of the subject part for each pixel;
It is characterized by having the following.
Note that the "area information" described above and below can be an image area according to various definitions, and refers to an arbitrarily specified image area.

In this case, the predetermined medical image data may be diffusion weighted image data or area information specified by the diffusion weighted image.
Furthermore, in any of the image processing apparatuses described above, the two types of MRI image data can be in-phase and out-of-phase image data in a water/fat signal cancellation method, which is an MRI fat suppression method.

Furthermore, the image processing method of the present invention includes:
For two types of MRI image data taken of the same subject part, calculate the difference value between corresponding pixels to obtain difference image data,
Calculating statistical information based on each pixel value of the obtained difference image data,
Based on the calculated statistical information, calculate a threshold for dividing each pixel of the difference image into two groups,
Based on the calculated threshold, each pixel of the difference image is divided into two groups,
For each divided group, a value specific to that group is given to each pixel included in that group as a pixel value, and a pixel value replaced image is obtained;
Based on the signal value or region information obtained from a type of predetermined medical image data different from the two types of MRI image data, and the signal value or region information obtained from the data of the pixel value replaced image, determining the state of the subject part for each pixel;
It is characterized by this.

Furthermore, the image processing program of the present invention includes:
to the computer,
A function to obtain difference image data by calculating a difference value between corresponding pixels for two types of MRI image data taken of the same subject part;
a function of calculating statistical information based on each pixel value of the obtained difference image data;
a function of calculating a threshold value for dividing each pixel value of the difference image into two groups based on the calculated statistical information;
a function of dividing each pixel value of the difference image into two groups based on the calculated threshold;
a function of assigning a group-specific value to each pixel included in the group as a pixel value for each of the divided groups, and obtaining a pixel value replaced image;
Based on the signal value or region information obtained from a type of predetermined medical image data different from the two types of MRI image data, and the signal value or region information obtained from the data of the pixel value replaced image, a function of determining the state of the subject part for each pixel;
It is characterized by causing the execution of the following.

According to the image processing device , image processing method, and image processing program of the present invention, difference values are calculated between corresponding pixels of two types of MRI image data taken of the same subject part, and the difference image data is generated. If there is almost no change between two image data, the difference value will be a value close to 0. On the other hand, if the change is large, the difference value will be a value far away from 0. It can be done.

However, since this difference value will vary slightly due to the positional shift between the two images, a statistical information calculation section and a threshold calculation section are used to calculate the difference value for each pixel value based on the statistical information. Based on the threshold, there is no signal change between the two images by dividing them into two groups: "within range" where the difference value is close to 0, and "outside range" where the difference value is far away from 0. Pixels can be identified accurately and objectively.

Furthermore, the pixel value replacement unit replaces the pixel value of each pixel in the two groups with a value unique to that group, and by obtaining a pixel value replacement image in which the difference image is binarized, it is possible to determine whether there is a signal change or not. can be classified more accurately and objectively. Further, when this difference image is compared with signal values of other medical images, the comparison becomes easy.

Further, according to the image processing device, image processing method, and image processing program of the present invention , the signal value or region information obtained from medical image data different from the above two types of MRI image data, and the above pixel value replacement. By determining the state of the subject part for each pixel of the subject part based on the signal value or region information obtained from image data, the state of the subject's imaged part (region of interest) can be clearly and objectively determined. For example, it is possible to clearly determine the spread of the red pulp site and tumor site.

1 is a block diagram showing an image processing device according to a first embodiment of the present invention. FIG. 1 is a flowchart showing an image processing method according to Example 1 of the present invention. In phase image ((A) (a)) and out of phase image ((A) (b)) of step S1 in FIG. 2, and a schematic diagram of the in phase image ((B) (a)) and out of phase image It is a schematic diagram ((B) (b)) of an image. 3A and 3B are a differential image (A) and a schematic diagram (B) of the differential image in step S2 in FIG. 2. FIG. 3 is a graph for determining the range of difference values in step S4 in FIG. 2. FIG. They are a pixel value replacement image (A) and a schematic diagram (B) of the pixel value replacement image in step S5 in FIG. 2. FIG. 2 is a block diagram showing an image processing device according to a second embodiment of the present invention. 7 is a flowchart showing an image processing method according to Example 2 of the present invention. The pixel value replaced image ((A)(a)) and the diffusion weighted image ((A)(b)), the schematic diagram of the pixel value replaced image ((B)(a)) and the diffusion weighted image in step S6 in FIG. It is a schematic diagram ((B) (b)) of an emphasized image. They are another pixel value replacement image (a) and a diffusion weighted image (b) in step S6 in FIG. 8.

An image processing apparatus, an image processing method, and an image processing program according to embodiments of the present invention will be described below.
<Example 1>
First, the main configuration of an image processing apparatus according to a first embodiment of the present invention will be explained using the block diagram of FIG.
This image processing device 50 includes a memory means 100 for reading two types of MRI image data obtained by photographing a region of interest of a subject using a magnetic resonance imaging device (hereinafter referred to as an MRI device), and a memory device 100 for reading two types of MRI image data obtained by photographing a region of interest of a subject using a magnetic resonance imaging device (hereinafter referred to as an MRI device). Two types of MRI image data (in this example, in-phase image data and out-of-phase image data by water/fat signal cancellation method (phase difference method, WFS method, DIXON method, etc.)) are read out, and the MRI image data is The apparatus includes a calculation processing means 200 for performing calculation processing on the data, and a monitor means 300 for displaying the MRI image data subjected to the calculation processing in the calculation processing means 200.

Further, the arithmetic processing means 200 is composed of hardware such as a CPU and hard disk memory, and software consisting of a program. a difference image calculation unit 201 that calculates difference values between corresponding pixels for two types of MRI image data to obtain difference image data; and each pixel value (hereinafter also referred to as pixel value) of the calculated difference image data. A statistical information calculation unit 202 that calculates statistical information such as the average value and standard deviation value of , and a threshold calculation unit that calculates a threshold value for dividing each pixel value of the difference image into two groups based on this statistical information. a pixel group dividing unit 204 that divides each pixel value of the difference image into two groups based on the calculated threshold; It includes a pixel value replacement unit 205 that adds a unique pixel value to obtain a pixel value replaced image.

Note that the general hardware configuration of the image processing device 50 (not necessarily shown in the drawings) includes a CPU, a memory means 100 such as the above-mentioned hard disk memory, a display memory, a display device, a mouse, The controller, keyboard, network adapter, etc. connected to the mouse are connected to the bus cable so that they can send and receive signals, and it is possible to send and receive signals to and from the MRI device and medical image database via the network. It is connected to the.

The CPU transmits programs stored in a storage device such as a hard disk memory or an optical disk memory, and data such as medical image information necessary for executing the program to the memory means 100 or the main memory and executes them.
Medical image information is acquired from a medical image database or an MRI apparatus via a network such as a LAN. On the other hand, the display memory temporarily stores display data to be displayed on the monitor means 300 such as a liquid crystal display or EL.

The mouse and keyboard are used by an operator to issue instructions to the image processing device 50, and the controller detects the state of the mouse, obtains the position of the mouse pointer on the display monitor, and displays the obtained position. Outputs information etc. to the CPU. Further, the network adapter is for connecting the image processing device 50 to a network such as a LAN, a telephone line, or the Internet.

In this embodiment, an MRI apparatus uses in-phase image data (first MRI image data) and out-of-phase image data (second MRI image data) necessary for extracting image data of a region of interest of a subject. The digital image data is stored in the memory means 100 and output to the arithmetic processing means 200 at a predetermined timing.
Here, in phase image data and out of phase image data mean in phase image data and out of phase image data in the water/fat signal cancellation method, which is an MRI fat suppression method.

Next, the image processing method according to the first embodiment of the present invention will be described using the flowchart of FIG. Note that each process of the image processing method according to this embodiment is performed using the image processing apparatus 50 shown in FIG.
As described above, in-phase image data and out-of-phase image data are used in the method of this embodiment.

First, to explain the outline, in-phase image data and out-of-phase image data obtained by photographing a region of interest of a subject's body using an MRI apparatus are input into the memory means 100 (S1). The in-phase image data and the out-of-phase image data input to the memory means 100 are subtracted for each pixel of the slice to obtain a difference image (S2).

Next, the pixel values of the entire difference image that is the imaging range are totaled, and the average value and standard deviation value of the total pixel values are calculated to calculate statistical information. Furthermore, from the calculated average value and standard deviation value, the value A-σ, which is obtained by subtracting the standard deviation value σ from the average value A, and the value A+σ, which is added, are determined, and the threshold value for the above-mentioned difference value is determined ( S3).
Based on this threshold value, a range (pixels belonging to the range) close to the average value (close to 0) and a range (pixels belonging to the outside of the range) far from the average value (deviating from around 0) are determined (S4).

After this, the pixel value of the pixel belonging to the range is replaced with a specific pixel value p, while the pixel value of the pixel belonging to the outside of the range is replaced with a specific pixel value q different from the pixel value p, and the pixel value Pixel value replacement processing is performed to obtain a replacement image (S5).
Each of the above steps S1 to S5 will be explained in more detail below using appropriate drawings. Step S1 is represented by the photograph in FIG. 3(A), step S2 is represented by the photograph in FIG. 4(A), step S3 is represented by the threshold calculation formula (1), and step S4 is represented by the photograph in FIG. , is represented by the graph in FIG. 5, and step S5 is represented by the photograph in FIG. 6(A). However, since the photographs in Figures 3(A), 4(A), and 6(A) are not necessarily clearly visible when shown as application drawings, they are schematic diagrams that only represent the essence of the content of the photographs. The diagrams are shown in FIG. 3(B), FIG. 4(B), and FIG. 6(B), respectively.

First, step S1 will be explained using FIG. 3(A) and FIG. 3(B). FIG. 3(A) is a lumbar MRI image of a subject, in which (a) shows an in-phase image, and (b) shows an out-of-phase image. 3(B)(a) is a schematic diagram of FIG. 3(A)(a), and FIG. 3(B)(b) is a schematic diagram of FIG. 3(A)(b).
As shown in FIGS. 3(B)(a) and 3(B)(b), in the region of interest of the subject 10, the area of the red pulp 1 is located in the left part of the image of the sacral region 3; The region of tumor 2 is present in the right part of the image. These two images are stored (uploaded) in the memory means 100.

Here, in-phase image data is the data at the moment when the atomic spin rotation periods of water and fat match, while out-of-phase image data is the data at the moment when the atomic spin rotation periods of water and fat are the opposite, and the rotation period is the opposite of that of in phase, and is the most mismatched data. This is image data at an instantaneous out-of-phase. In the in phase, the spins of water and fat match, which emphasizes the signal values, but in the out of phase, the spins do not match, so in pixels where water and fat coexist, their signals cancel each other out, and the signal The value is said to decrease.

Next, step S2 will be explained using FIG. 4(A) and FIG. 4(B). FIG. 4(A) shows a difference image obtained by subtracting the in phase image shown in FIG. 3(A)(a) and the out of phase image shown in FIG. 3(A)(b) for each slice image. It is something. Moreover, FIG. 4(B) is a schematic diagram of FIG. 4(A).
As is clear from FIG. 4(B), the white (or black) level in the red pulp 1 is extremely high compared to the sacral region 3 and the tumor 2 region.
That is, in the red pulp 1, since fat and water are mixed, the signal change between the in-phase image and the out-of-phase image is large, and the signal value becomes high in the difference image.
On the other hand, since the tumor 2 can be considered to be surrounded by only water, the signal change between the in-phase image and the out-of-phase image is small, and the signal value in these difference images is around 0.
As a result, the signal level in the area of the red pulp 1 becomes extremely high, making it possible to distinguish it from the tumor 2.
Note that since the outline of the subject 10 is not necessarily clear in the difference image, it is represented by a broken line on the schematic drawing of FIG. 4(B). The same applies to the corresponding drawings from FIG. 4 onwards.

Next, step S3 will be explained.
The pixel values of the entire difference image in each slice image obtained in step S2 are totaled, and the average value and standard deviation value of the total pixel values are calculated to calculate statistical information.
Here, if the average value is A and the standard deviation value is σ, then the value A - σ obtained by subtracting the standard deviation value σ from the average value A, and the added value A + σ are used as the threshold value D for the above difference value. obtain.
For example, if the average value A is 14.28 and the standard deviation value σ is 31.12, the threshold value D is obtained by the following formula (1).
D=14.28±31.12 (1)
That is, the threshold D in this case is -16.84 and 45.40.

Next, step S4 will be explained using FIG. 5. In this figure, the pixel values of each pixel of the entire difference image in each slice image are totaled, and the horizontal axis is the difference pixel value (difference signal value) and the vertical axis is the number of pixels. It is plotted on a biaxial plane. When each pixel is plotted in this way, a steep peak generally appears near the average value. Next, as shown in FIG. 5, vertical dividing lines are drawn on this graph at the difference pixel values corresponding to the threshold value D on the horizontal axis, that is, the values of -16.84 and 45.40, and if -16.8 or more, The range is determined by determining the range as being 45.4 or less as "within range", and as being less than -16.84 and greater than 45.40 as "out of range".

Next, step S5 will be explained using FIG. 6(A) and FIG. 6(B). FIG. 6(A) shows a pixel value replacement image obtained by replacing each pixel value (difference pixel value) with a group-specific value for each division based on the divisions (groups) obtained in step 4. It shows. Moreover, FIG. 6(B) is a schematic diagram of FIG. 6(A).
The value specific to the group to be replaced can be any value, but for example, in the "within range" category, each pixel value (difference pixel value) is replaced with a value of 0; '', each pixel value (difference pixel value) is replaced with a value of 100.

The pixel value replaced image thus obtained is binarized into white and black, and is clearly divided into two sections, as shown in FIG. 6(A). In other words, as shown in FIG. 6(B), only the area of red pulp 1 can be set to white (value of 100) and can be clearly distinguished from the area of tumor 2 (value of 0). It can be applied to various diagnoses.

In the above embodiment, the threshold value is calculated using the average value and the standard deviation value as the statistical information, but the threshold value may be calculated using other information as the statistical information. For example, the median value or mode may be used instead of the average value, or the interquartile range (distance value between the 25th percentile and 75th percentile) may be used instead of the standard deviation value. Further, a value calculated by weighting the statistical information by multiplying it by an arbitrary coefficient may be used as the threshold value.

<Example 2>
Next, the block diagram of FIG. 7 shows the main configuration of an image processing device 50a according to a second embodiment of the present invention. The image processing device 50a according to the second embodiment has a configuration similar to that of the image processing device 50 according to the first embodiment described above, but the arithmetic processing means 200a mainly controls the subject part state determination section 206a. They are different in that they have Therefore, in order to avoid duplication and complication of explanation, in the second embodiment, only the subject part state determination section 206a of the arithmetic processing means 200a will be explained in detail, and the rest will basically be the image processing according to the first embodiment. It is assumed that the contents explained in connection with the device 50 apply. In addition, the members in Example 2 that are common to Example 1 shall be represented by the reference numerals of the members in Example 1 with a added (a is also added to the part state discriminating unit of the subject). (Represented by the attached symbol).

Similar to the first embodiment, the arithmetic processing means 200a in the second embodiment photographs the region of interest (part) of the same subject and generates two images consisting of an in-phase image and an out-of-phase image, which are stored in the memory means 100a. A difference image calculation unit 201a calculates difference values between corresponding pixels for different types of MRI image data to obtain difference image data, and statistics such as the average value and standard deviation value of each pixel value of the calculated difference image data. A statistical information calculation unit 202a that calculates information, a threshold calculation unit 203a that calculates a threshold value for dividing each pixel value of the difference image into two groups based on this statistical information, and a calculated threshold value. Based on this, a pixel group division unit 204a divides each pixel value of the difference image into two groups, and for each divided group, a pixel value unique to that group is assigned to each pixel included in that group, and the pixel value is divided into two groups. It includes a pixel value replacement unit 205a that obtains a replacement image.

Unlike the first embodiment, the data of the pixel value replaced image is divided into the two categories "within range" and "outside of range" and replaced with a binarized image in the pixel value replacement unit 205a, A subject part state determination unit 206a is provided which determines the state of a part of the subject (region of interest) based on information of a third medical image (in this embodiment, a diffusion weighted image which is an MRI image). ing.

That is, the subject part state determination unit 206a uses the data of the pixel value replaced images shown in FIGS. 6(A) and 6(B) and the part of the subject ( Based on data of a diffusion weighted image (hereinafter referred to as DWI) of a region of interest), red pulp 1 and tumor 2 are automatically determined in this part of the subject, and processing that contributes to diagnostic support is performed.
The above-mentioned DWI is obtained as image data of the subject at the same site and at the same time as the above-mentioned in phase image data and out of phase image data, and is temporarily stored in the memory means 100a. It is something that

To begin with, DWI is an image of the degree of Brownian motion of molecules, and the greater the motion, the lower the signal. The imaging method involves applying a large gradient magnetic field (MPG) in opposite directions so as to intersect with the pulse sequence, and measuring the strength of the MPG influence (b value). Basically, DWI is also a T2-weighted image, so a lesion with a high degree of freedom that has a high signal on a T2-weighted image will also have a high signal on DWI.

In this way, the subject part state determination unit 206a performs a process based on the pixel value replacement image obtained from the pixel value replacement unit 205a and the DWI that represents the tumor 2, red pulp 1, etc. as a higher signal value than other parts. , the tumor 2 and the red pulp 1 are automatically distinguished, and content to support diagnosis is output.

Next, an image processing method according to a second embodiment of the present invention will be described using the flowchart of FIG. Note that each process of the image processing method according to this embodiment is performed using an image processing apparatus 50a shown in FIG.
The image processing method according to the second embodiment has steps similar to the image processing method according to the first embodiment described above, but in this embodiment, only the process in step S6 is added.

This step S6 will be explained using FIG. 9(A), FIG. 9(B), and FIG. 10. FIGS. 9A and 10 are MRI images of the subject's lower back, in which (a) shows a pixel value replacement image, and (b) shows a DWI (diffusion weighted image). 9(B)(a) is a schematic diagram of FIG. 9(A)(a), and FIG. 9(B)(b) is a schematic diagram of FIG. 9(A)(b).
As shown in FIG. 9(A)(b) and FIG. 9(B)(b), there is a high signal (arrow portion) in the diffusion weighted image, and FIG. 9(A)(a) and FIG. 9(B) As shown in (a), in the pixel value replaced image, an area (arrow portion) that is "within the range (pixel values are replaced with 0)" can be determined to be tumor 2.

On the other hand, as shown in FIG. 10(b), there is a high signal area (arrow area) in the diffusion weighted image, and as shown in FIG. is replaced with 100)" (arrow portion) can be determined to be red pulp 1.

Therefore, in this embodiment, by performing the process of step S6, it is possible to determine the state of the site (region of interest) of the subject 10, and in particular, clearly distinguish the site of the red pulp 1 and the site of the tumor 2. Not only can they be distinguished, but also the spread of each part can be clearly and objectively determined.

In addition, since the four images consisting of the above-mentioned in phase image, out of phase image, these two difference images, and diffusion weighted image are information (Slice location) included in the DICOM standard, if they are the same examination, , since the same parts (organs) exist in pixels at the same position (same coordinates), it is possible to easily identify pixels at the same position (same coordinates) between images without using special positioning technology. It is.

In addition, when obtaining a differential image between an in-phase image and an out-of-phase image, unnecessary areas (such as the air area outside the body of the subject 10) are masked (deleted), and then the difference between the images is It is also conceivable to perform a difference calculation between pixel values of corresponding pixels. By performing the processing according to the method of this embodiment with unnecessary areas deleted in advance, it becomes possible to specify a range with higher accuracy.

The image processing program according to the embodiment of the present invention is a program used in the image processing apparatuses 50 and 50a according to the first embodiment and the second embodiment, and is configured as follows.
That is, the computer has a function to calculate difference values between corresponding pixels for two types of MRI image data taken of the same subject part, and obtains difference image data, and a function to obtain difference image data for each of the obtained difference image data. a function of calculating statistical information based on pixel values; a function of calculating a threshold value for dividing each pixel value of the difference image into two groups based on the calculated statistical information; A function of dividing each pixel value of the difference image into two groups based on a threshold value, and for each divided group, assigning a value unique to that group as a pixel value to each pixel included in that group, a function of obtaining a pixel value replaced image; signal values or region information obtained from predetermined medical image data of a type different from the two types of MRI image data; and signal values or region information obtained from data of the pixel value replaced image. The present invention is configured to execute a function of determining the state of the subject part for each pixel of the subject part based on the information.

Note that the image processing apparatus, image processing method, and image processing program of the present invention are not limited to those of the above embodiments, and various other changes can be made.
For example, in each of the above embodiments, in-phase image data and out-of-phase image data based on the water/fat signal cancellation method are used as two types of MRI image data, but instead of this, T1-weighted image data and T2-weighted image data are used. It can be enhanced image data. In addition, there are two images before and after fat suppression in non-selective fat suppression (STIR) and frequency-selective fat suppression (CHESS), which are different fat suppression methods from the above, two diffusion-weighted images with different b values, and ADC. It is also possible to use an image.

Further, in the second embodiment described above, a diffusion weighted image which is an MRI image is used as the predetermined medical image data, but a contrast-enhanced MRI image may be used instead of this, or a CT image other than an MRI image or a PET image may be used. Images may also be used.

Further, in each of the above embodiments, an apparatus, method, and program for distinguishing or discriminating between red pulp and a tumor in a region of interest of a subject are described, but the image processing apparatus, image processing method, and image processing method of the present invention The program is not limited to distinguishing or discriminating between these two, but can be used to distinguish or distinguish various other tissues in a subject.

1 Red pulp 2 Tumor 3 Sacral region 10 Subject 50, 50a Image processing device 100, 100a Memory means 200, 200a Arithmetic processing means 201, 201a Differential image calculation section 202, 202a Statistical information calculation section 203, 203a Threshold calculation Sections 204, 204a Pixel group classification section 205, 205a Pixel value replacement section 206a Subject part state determination section 300, 300a Monitor means

Claims (5)

a difference image calculation unit that obtains difference image data by calculating difference values between corresponding pixels for two types of MRI image data taken of the same subject part;
a statistical information calculation unit that calculates statistical information based on each pixel value of the difference image data calculated by the difference image calculation unit;
a threshold calculation unit that calculates a threshold for dividing each pixel of the difference image into two groups based on the statistical information calculated by the statistical information calculation unit;
a pixel group classification unit that classifies each pixel of the difference image into two groups based on the threshold calculated by the threshold calculation unit;
a pixel value replacement unit that gives a value unique to each group as a pixel value to each pixel included in the group for each divided group, and obtains a pixel value replacement image;
Based on the signal value or region information obtained from a type of predetermined medical image data different from the two types of MRI image data, and the signal value or region information obtained from the data of the pixel value replaced image, a subject part state determination unit that determines the state of the subject part for each pixel;
An image processing device comprising: The image processing apparatus according to claim 1 , wherein the predetermined medical image data is diffusion weighted image data or region information specified by a diffusion weighted image. The image processing apparatus according to claim 1 or 2 , wherein the two types of MRI image data are in-phase and out-of-phase image data in a water/fat signal cancellation method, which is an MRI fat suppression method. . For two types of MRI image data taken of the same subject part, calculate the difference value between corresponding pixels to obtain difference image data,
Calculating statistical information based on each pixel value of the obtained difference image data,
Based on the calculated statistical information, calculate a threshold for dividing each pixel of the difference image into two groups,
Based on the calculated threshold, each pixel of the difference image is divided into two groups,
For each divided group, a value unique to that group is given to each pixel included in that group as a pixel value, and a pixel value replaced image is obtained;
Based on the signal value or region information obtained from a type of predetermined medical image data different from the two types of MRI image data, and the signal value or region information obtained from the data of the pixel value replaced image, determining the state of the subject part for each pixel;
An image processing method characterized by: to the computer,
A function to obtain difference image data by calculating a difference value between corresponding pixels for two types of MRI image data taken of the same subject part;
a function of calculating statistical information based on each pixel value of the obtained difference image data;
a function of calculating a threshold value for dividing each pixel value of the difference image into two groups based on the calculated statistical information;
a function of dividing each pixel value of the difference image into two groups based on the calculated threshold;
a function of assigning a group-specific value to each pixel included in the group as a pixel value for each of the divided groups, and obtaining a pixel value replaced image;
Based on the signal value or region information obtained from a type of predetermined medical image data different from the two types of MRI image data, and the signal value or region information obtained from the data of the pixel value replaced image, a function of determining the state of the subject part for each pixel;
An image processing program that executes.

Images