KR100291861B1 - A method of separating the white matter and the grey matter from brain image and calculating the volume of thereof - Google Patents

Abstract

The pretreatment process removes the brain region of the PD image except for the brain and removes the same region from the corresponding T2 image, and removes the cerebrospinal fluid from the T2 image after the preprocessing process and separates the cerebrospinal fluid of the same region from the corresponding PD image The process of separating the cerebrospinal cord, the process of generating histograms using PD images separated from the cerebrospinal cord, calculating the partial volume of the white matter and gray matter, and calculating the integral of the distribution function while increasing the contrast value in the histogram. The process of calculating the discriminant value which sets the first contrast value that is greater than the sum of the partial volumes of the white matter as the discriminant value, and the process of generating an image in which the white and gray matter are separated for each slice based on the discriminant value, and Each volume of white matter and gray matter is obtained by using the partial volume information of and the slice information provided by the tomography apparatus. Methods Isolation and volume calculation of the white matter and gray matter in the brain images, characterized in that comprising the step of calculating.

Description

A method of separating the white matter and the gray matter from brain image and calculating the volume of

The present invention relates to a brain image processing apparatus and processing method, and more particularly, to a method for calculating the respective volumes by separating the white matter and gray matter from the brain image provided from the magnetic resonance imaging device.

The brain is known as the most important organ of the various organs of the human body that controls physical parts of all parts of the human body and controls and controls the human mental world in order to live normally. In addition, since its function is so complex, medical research on various roles and mechanisms of the brain continues to be conducted, and various studies have been continuously conducted in other academic fields on the functions and roles of the brain.

The human brain has a very complex structure in the cortex that surrounds the brain, with white matter and gray matter part of the brain, with a small amount of cerebrospinal fluid. However, the current magnetic resonance imaging apparatus which photographs the brain having such a complicated structure and makes it into an image is obtained by photographing the brain at a certain thickness (usually 5 mm) and having a specific contrast value according to the ratio of each component present in the thickness. From these images, no quantitative method has been developed to determine the partial volume of each component of how much white and gray white matter is within a given thickness. Therefore, there was no way to diagnose the disease or detect it early by calculating the volume of white matter and gray matter and comparing it with the normal person in the case of abnormal person suffering from mental illness as well as normal person.

Thus, until now, only the necessity of relying on the primitive method of separating the white matter and the gray matter and predicting the total volume of each brain image slice by hand by a medical specialist is necessary. Since there was no way to calculate the proportion of partial volume, even medical professionals remained at unreliable levels of accuracy.

An object of the present invention is to provide a method for calculating the volume of the white matter and gray matter separated from the image of the brain provided from the magnetic resonance imaging apparatus.

It is another object of the present invention to provide a method for determining the contrast values of white and gray matter in a given brain image and determining the reference values that can be separated from each other without absolute reference.

Another object of the present invention is to remove the black background of the image present in the initial image prior to separation of the white matter and gray matter, as well as to remove the ocular and cerebellar parts and ventricles appearing in the outer layer and the fat layer and some slices surrounding the white matter and gray matter. This is to minimize the extraction of pure white and gray white and to calculate the volume error.

Another object of the present invention is to fully consider that the effect of the cerebrospinal fluid, considering that the white matter and gray matter can be expressed much brighter than the original contrast value of the white matter and gray matter according to the amount of cerebrospinal fluid as it is mixed with a small amount of cerebrospinal fluid It is in the separation of white matter and gray matter.

A feature of the present invention for the above object is a pre-processing process to remove parts except the brain from the brain image of the PD image and to remove the same region from the corresponding T2 image, and remove the cerebrospinal fluid from the T2 image is completed The cerebrospinal cord separation process of separating the cerebrospinal fluid of the same region in the PD image, the process of generating the first histogram from the PD image consisting of the white matter and the gray matter, and the partial volume of the white matter (PV1) from the first histogram. ) A step of removing the pure cerebrospinal fluid from the cerebrospinal cord images separated by the cerebrospinal cord separation process, generating a second histogram from the remaining images, calculating a partial volume of the pure cerebrospinal fluid from the second histogram, , A part occupied by white matter in the region from which the pure cerebrospinal fluid is removed Calculating a product (PV2), subtracting the sum of the partial volumes of the white matter (PV1 + PV2) from the total pixel values of the first histogram, calculating a partial volume of the gray matter, and the first histogram and the second Generating a third histogram including a histogram, and calculating an integral value of the distribution function while increasing the contrast value in the third histogram, the first of which is greater than the sum of the partial volumes of the white matter (PV1 + PV2). A process for calculating a discriminant value for setting the contrast value of the light to a discriminant value, generating an image in which the white matter and the gray matter are separated for each slice based on the discriminant value, and information and tomography of the partial volume of the white matter and the gray matter It is a process which calculates each volume of white matter and gray matter using slice information provided from an apparatus.

1 is a control block diagram showing the configuration of a brain image processing apparatus according to the present invention;

2 is a PD slice used for the extraction of white matter and gray matter,

3 is a PD image slice from which the black background of the background is removed,

4 is a slice showing four starting pixels searched from four directions for tracking fat layers,

Figure 5 is a PD image slice, the outer skin removed through the tracking and removal of the fat layer,

6 is an exemplary diagram of an image matrix showing a trimming process;

7 is a PD image slice from which noise around the brain is removed through a trimming process;

FIG. 8 is a PD image slice from which cerebellum and brain stem are removed by manual operation with respect to the image of FIG. 7;

9 is a T2 image slice of which a preprocessing process is completed using the PD image slice of FIG. 8;

10 is a contrast contrast slice for explaining a contrast separation algorithm according to the present invention;

11 shows the 8th and 9th PD and T2 image slices of the whole slice,

12 is an image slice of the cerebrospinal fluid T2 and PD images,

13 is a comparison chart of contrast values of the cerebrospinal fluid extracted from the 9th T2 image and the PD image,

14 is a histogram for a PD image,

15 is a PD image slice excluding the separation of the whole cerebrospinal fluid and the influence of the cerebrospinal fluid extracted from the PD image,

16 is a histogram showing a method of calculating the partial volume of white matter,

17 is a histogram showing determination of a discrimination value by an algorithm according to the present invention;

18 is an example of separated PD images of white matter and gray matter by the determination value determined by FIG. 17;

19 is a whole PD image slice separated into white matter and gray matter.

Hereinafter, the configuration and operation of the present invention will be described in detail. 1 is a control block diagram illustrating a control and processing relationship with devices related to image processing according to the present invention.

The image file provided from the magnetic resonance imaging apparatus 1 is transferred to the image processing unit 3 via the interface apparatus 2. At this time, the image processing unit 3 stores the transferred image information in the main memory device 4 and loads a program for separating the white matter and gray matter prepared for the present invention stored in the auxiliary memory device 5. After the pretreatment process, the white matter and the gray matter are output through the image output device 6 through the determination of the discrimination value for the separation of the white matter and the gray matter, and the volume of the white matter and the gray matter which are calculated is stored in the image database. The image database maintains information on the calculated white matter and gray matter image, calculated volume, and gender, age, and name of the person (normal person or patient) corresponding to the treated image.

The brain imaging apparatus includes a computer tomography apparatus or a magnetic resonance imaging apparatus, and the image file processed in the present invention means that it is obtained from a magnetic resonance imaging apparatus. In addition, the output device of the extracted image is a display function so that the user can recognize the processed image information, and may include a printer or a monitor.

One image slice size for the brain provided from the magnetic resonance imaging apparatus is 256 * 256 in size, and contains an image in which each pixel can be represented by a contrast value of 0 to 4095. One pixel consists of two bytes, so one slice of the brain's MRI is 256 * 256 * 2 = 128 Kbytes. In addition, the header of this file contains various types of information, from information on the composition and format of the image to information on the patient's personal information.

The present invention converts such image slice information into an 8-bit PGM (Portable Gray Map) format file and then uses it for separation of white matter and gray matter. In this process, each pixel of the image file is converted into an image having a contrast value of 0 to 255, which can be represented by 8 bits.

In order to separate the white matter and the gray matter, and to calculate the volume, the image slice to be applied to the present invention uses slices taken in cross section with respect to the cerebral portion of the whole slices produced by the magnetic resonance imaging apparatus. Two types of proton density images (hereinafter referred to as PD images) and T2 weighted images (hereinafter referred to as T2 images) are generated and made around 15 slices each for a given brain cross section.

The image slice, which is taken up to 5mm in thickness from the lower part (eye height) in a cross section with respect to the cerebral part and is produced with a compressed contrast value, has a different appearance depending on the position. A few slices at the bottom represent the cerebellum, a part of the eye, and the brain stem along with a relatively small area of the cerebrum.In the middle position, the image is largest in size and has an elliptical shape. Have Going to the top, the size of the image is getting smaller and smaller. 2 is a whole image slice to which the present invention is applied and the type is a PD image.

The T2 image and the PD image photographed for the same cross section have different characteristics in the distribution of contrast values. The T2 image is darker than the PD image, but the cerebrospinal fluid is relatively brighter than the PD image, so it is easily used to extract the cerebrospinal fluid. However, the white and gray white parts are darker than PD images, and the two components are very difficult to distinguish. On the other hand, the PD image is not easy to distinguish the cerebrospinal fluid, but since the white matter and the gray matter are brighter than the T2 image, and the white matter and the gray matter are relatively easy to distinguish, the white matter and the gray matter are separated from the PD image in the present invention.

In the present invention, the remaining portions that do not correspond to the cerebrum are removed first before separation of the white matter and the gray matter. This process first removes the black background of the image and the cortex and fat layer surrounding the brain in the PD image. The removal of the black background of the image is done as follows.

The black background of the image has a black value with a contrast value close to zero, and the brain envelope has a larger contrast value (greater than 60). Therefore, the black background of the background except the inner part of the brain can be removed by tracking and removing black pixels having a contrast value of 60 or less while being connected to the initial position (coordinate 0,0) of the entire image. 3 shows PD image slices in which the dark portion of the background is removed by this process.

Next, after the black background of the image is removed as shown in FIG. 3, the outer skin and the fat layer surrounding the brain are removed. Observing the entire image reveals the presence of a black fat layer inside the brain's envelope. This fat layer has a darker value (less than 70) is significantly darker than the outer skin (more than 100) or the inner gray matter (more than 130). Thus, the difference in intensity values is used to track the fat layer and remove it with the skin.

In the present invention, the property of 4-connectivity was used for tracking the fat layer. 4-connectivity refers to four pixels adjacent to the top, bottom, left and right connected to a given pixel. In other words, it starts from an arbitrary pixel corresponding to the fat layer and checks whether the pixel corresponding to the same fat layer among the four connectivity pixels is removed.

However, since there are cases where the fat layers distributed in a band shape are not completely connected inside the outer skin, four starting pixels for the fat layer tracking are set on the top, bottom, left, and right sides of the image as shown in FIG. 4. That is, down from the coordinates of (0,128), from the coordinates of (128,0) to the right, from the coordinates of (128,256) to the left, and upwards from the coordinates of (256,128). Searching inward can determine the pixels that correspond to the contrast values (less than 70) of the fat layer. After setting these four pixels, as described above, it is made by finding and removing pixels having the contrast value of the fat layer while being connected by 4-connectivity in each pixel.

FIG. 5 shows PD images in which the fat layer traced by 4-connectivity was removed from these four starting pixels, and the skin was removed together when the inside and the skin of the brain were completely separated by the traced fat layer.

However, as shown in FIG. 5, when the fat layer is cut because it is too thin than other parts, the fat layer is connected to the pixels constituting the envelope and the inside of the brain. In this case, the fat layer is removed but the shell is not completely separated. The problem is that some slices in the lower part of the brain are different from the middle slice in terms of overall shape or internal structure, and the fat layer does not have an elliptical band, so Since the pixels constituting the inside of the brain are attached, the outer skin is not completely removed by the above-described method. In addition, the rest of the image slices are attached to a part of the outer shell that could not be removed.

In the present invention, a trimming algorithm is used to remove such external noise, and by this process, a part or other part of the outer shell connected to the brain is completely removed, leaving only the inside of the brain including white matter and gray matter. It is a process.

This algorithm has the same effect as removing a ball from the inside of the brain that rolls out along the perimeter and removes the protruding part that the ball does not enter. More precisely, it uses a morphological filter, which scans a structured element (SE) of any size over the entire image, with outward projections of a smaller thickness than the given SE. Will be removed. In the actual implementation algorithm, the SE of 12 * 12 is used, but for convenience, the SE of 3 * 3 will be described as an example. In FIG. 6, (A) is a structured element (SE), (B) is an image matrix to which a trimming process is to be applied, (C) is an image matrix in which protrusions to the outside are cut by a trimming process, and (D) Using the inside of the brain, the image matrix is removed from the noise.

For the image as shown in FIG. (B), the SE as shown in (A) is compared with the image to be applied while moving one pixel from the upper left in order. At this time, the pixel is selected only when all pixels are equal to SE. Thus, the area marked with B in (C) and its surroundings remain while the area of thickness smaller than SE, such as the area marked A, is removed. However, although the B 'region is a terminal part of the brain, the following method is used to prevent damage due to inconsistent SE. This compares with SE and counts the number of matches so that 7 or more instead of 9 is the same as selecting the pixels of the original image. By allowing a slight error in this way, it is possible to include any irregularities existing at the boundary, thereby preventing the loss of the original image.

The images of FIG. 7 show images processed by this trimming process. However, some slices of the lower part having the eye or the cerebellum, etc., due to the configuration of the brain image, are removed by manual operation through a user interface of an appropriate image viewer since these portions cannot be removed by the trimming process. However, this manual removal process is only done for a few slices and does not require much time (about 1 minute) since the amount is not high.

The images of FIG. 8 show PD images from which cerebellum and brainstem are removed by hand after trimming. As described above, the magnetic resonance imaging apparatus generates two types of PD images and T2 images for the same cross section of the brain. Therefore, in the T2 image, the background, the outer skin, the fat layer, and the parts such as the eye and the cerebellum are removed by comparing the T2 image corresponding to the PD image of FIG. This is accomplished by removing the location of the brainstem to the same white color of 255 in the T2 image. FIG. 9 is a T2 image slice from which noises such as eyes and cerebellum appearing in the background, the outer skin, the fat layer, and some of the lower portions of the image are removed.

In the present invention, the volume calculation through the separation of the white matter and the gray matter is finally performed in the PD image of FIG. 8. The most important point in the separation of white matter and gray matter is to separate the white matter from the white matter by considering the partial volume of the white matter and gray matter by analyzing the blurred contrast value.

As described above, the magnetic resonance imaging apparatus scans a cross section of the brain at a predetermined thickness (about 5 mm) to generate an image having a specific contrast value according to the ratio of components included in the thickness. Therefore, the darkest white matter present in a certain thickness and light gray than the white matter, and the amount of light and dark spinal cord with the smallest amount of light and dark, according to the amount of the summary is summarized.

To date, no quantitative means have been developed to accurately estimate the proportion of each component originally present in a certain thickness in images of brains having a contrast value generated by these characteristics. Therefore, in order to solve the blind spot, the present invention proposes a method for analyzing the blurring contrast value of an image generated by the magnetic resonance imaging apparatus, and calculates the ratio of each component included in a predetermined thickness based on the white matter and gray matter in each image slice. We present a quantitative method for determining the discriminant for the separation of. In addition, the white matter and gray matter are separated from the white image by calculating the partial volume considering the ratio of white matter and gray matter contained within a certain thickness based on the number of pixels constituting each component by using the determined discrimination value. We want to calculate the volume of

First, the algorithm for contrast value analysis used for calculating the partial volume of white matter and gray matter in the present invention will be described briefly. 10 (A) shows that two components having different contrast values exist within a predetermined thickness. FIG. 10B is an image when the image of FIG. 10A is captured by a magnetic resonance imaging apparatus and then compressed vertically to express a single contrast value. If only one component is present at any position, the expressed contrast value has the same contrast value as the original component. However, when two components having different contrast values exist together, the contrast values are expressed differently according to the ratio of each component included. As shown in (B) of FIG. 10, a ratio of a component having a contrast value of y1 is z and a component having a contrast value of y2 is present in the image represented by the contrast value blurring due to the compression process. When the ratio is 1-z, the contrast value G in the image generated by the magnetic resonance imaging apparatus is as follows.

G = y ₁ * z + y ₂ * (1-z) .......... (0 <z <1)

From the above equation, the partial volume of a specific component can be simply inferred as If the intensity value of component A is y1 and the intensity value of component B is y2, the two components are mixed within a certain thickness to represent the intensity value of G. It can be calculated that the component is (1-z).

By this method, it is possible to find out the number of pixels constituting each component from the whole in consideration of the ratio of A or B components in the compressed image as shown in FIG. 10 (B). Determines the contrast value that separates the two parts within the overlapped area. This contrast value is called a discrimination value, and FIG. 10C is an image separated by the determined discrimination value.

If FIG. 10A is a cross section having a thickness of 5 mm in the actual brain, FIG. 10B is referred to as an image made by a magnetic resonance imaging apparatus. Here, the actual state of the brain as shown in FIG. 10A is unknown from the image as shown in FIG. 10B. However, after calculating the inclusion ratio of each component through the contrast value analysis proposed in the present invention, the determination value is determined using the determination, and using the determined determination value from the image (B) of FIG. 10 (C) It will be separated like an image. Similarly, although the original state as shown in FIG. 10A is not known from the image of FIG. 10C, the number of pixels constituting each component shown in FIG. In A), the partial volume of each component is considered and the volume of each component can be calculated accurately.

The algorithm described so far will be described in detail with respect to an image generated by a magnetic resonance imaging apparatus for a human brain handled in the present invention. Images processed by applying the algorithm described below will be shown only for the 8th and 9th images of the image of FIG. 8, and the remaining image slices are processed by the same procedure.

The image of FIG. 11 is a PD image and a T2 image for the eighth and ninth selected images of FIGS. 8 and 9. FIG. 11A shows an eighth PD image, (B) an eighth T2 image, (C) a ninth PD image, and (D) a ninth T2 image. The contrast value of the pixels constituting the PD image of FIG. 11 is a value expressed by being compressed after the human brain is scanned by the magnetic resonance imaging apparatus with a predetermined thickness. Therefore, they are represented by different contrast values depending on the ratio of the components that were in their original thickness. If only pure white matter is present, it is expressed as the darkest contrast value. In the case where both white and gray are present, the amount of gray matter is gradually expressed brightly. White matter and gray matter are present brighter with a small amount of cerebrospinal fluid, brighter with gray matter and cerebrospinal fluid, and brighter with pure cerebrospinal fluid. Therefore, in the PD image of FIG. 11, the cerebrospinal fluid that is not related to the white matter and gray matter to be calculated in the present invention should be removed.

In the present invention, the following method was used to extract cerebrospinal fluid that did not affect white matter and gray matter in PD images. The T2 image of the PD image and the T2 image to be used in the present invention is darkly expressed overall, but is brightly expressed for the cerebrospinal fluid. Therefore, T2 images can be easily used for the extraction of the whole cerebrospinal fluid. For this purpose, the observations were made on several T2 images and the results of the radiologist's opinions showed that the pixels with brighter and brighter values than the contrast value 112 were the correct part of the cerebrospinal fluid in the T2 image. Therefore, the pixel corresponding to the contrast value greater than the contrast value 112 in the T2 image is extracted to the cerebrospinal fluid region by this criterion. FIG. 12 shows T2 and PD images from which the cerebrospinal fluid is removed from each image of FIG. 11, and images of cerebrospinal fluid extracted from PD images. (A) is the 8th and 9th T2 images, (B) is the T2 image from which the cerebrospinal fluid has been removed, (C) is the same position as the cerebrospinal fluid of the T2 image in the corresponding PD image based on the T2 image of (B) The PD pixels are extracted by removing the cerebrospinal fluid from the PD image, and (D) shows the cerebrospinal fluid removed from the PD image of (C).

As shown in (C), the PD image from which the cerebrospinal fluid has been removed is an image of a part in which only white matter and gray matter exist within a certain thickness, and is used to calculate partial volume of white matter and gray matter by analyzing light and dark values of the histogram.

Contrast values of the cerebrospinal fluid removed from T2 and PD images show different distributions. FIG. 13 is a histogram showing the distribution of contrast values between the cerebrospinal fluid extracted from the T2 image and the cerebrospinal fluid extracted from the PD image for the ninth slice. In FIG. 13, the histogram of the upper portion is for a T2 image, and the histogram of the lower portion is for a PD image. Overall, the T2 image is darker than the PD image, so it has a low contrast value and is located on the left side of the histogram. The histogram for the PD image is located on the right side than T2. Here, the cerebrospinal fluid (contrast value 112 or higher) extracted from the T2 image shows that the PD is widely distributed from the contrast values A to C. These features lead to the following conclusions.

First, if there is only pure cerebrospinal fluid within a certain thickness, such cerebrospinal fluid corresponds to the contrast values of B to C in the histogram of PD images. Here, B corresponds to the contrast value of pure gray matter, and the contrast value of B can be obtained from the histogram of the PD image from which the whole cerebrospinal fluid is removed.

Second, when the cerebrospinal fluid is present with white matter as well as gray matter, the cerebrospinal fluid is darkened according to the amount of white matter contained and corresponds to the contrast value of A to B.

Therefore, the cerebrospinal fluid, which is darker than the contrast value of B in the PD image, that is, the region of A to B, becomes the cerebrospinal fluid affected by white matter or gray matter, and this part is the cerebrospinal fluid containing white matter and gray matter. This is the part to be considered in order to calculate the partial volume. Therefore, in this part, the portion corresponding to the partial volume of the cerebrospinal fluid should be removed through contrast analysis, and the partial volume of the white matter and gray matter should be calculated again.

The above analysis is once again summarized using the histogram of the PD image. The histogram of FIG. 14 assumes a histogram of a PD image (FIG. 11C) that removes background and cortical fat layers, and parts unrelated to the extraction of white matter and gray matter, such as cerebellum and brain stem, and shows white matter, gray matter and whole cerebrospinal fluid. Histogram of the containing image.

In FIG. 14, the largest area A is a histogram of an image in which the entire cerebrospinal fluid is removed from the PD image (FIG. 12C), and only white matter and gray matter are present within a predetermined thickness. Occupies most of it. Therefore, the contrast value y1 corresponds to the lower limit of the contrast value that pixels of white matter may have and y3 corresponds to the upper limit of the contrast value of pixels of the gray matter component. In the histogram of FIG. 14, the histograms of the areas marked B and C are histograms of the whole cerebrospinal fluids extracted from the PD image, and the B regions are histograms of the cerebrospinal fluids affected by the heavy and gray matters. Accordingly has a bright contrast value. And C is a region (region C in B of FIG. 13) that does not affect the white matter and gray matter, that is, where only the cerebrospinal fluid exists within a certain thickness, PD for this portion does not need to consider the volume of the white matter and gray matter It is first removed from the total cerebrospinal fluid (FIG. 12D) extracted from the image. Thus y3 is the upper limit of the contrast value that gray matter can have and the lower limit of the contrast value that the cerebrospinal fluid can have. This leads to the following conclusions.

First, partial volumes of white matter and gray matter should be considered in region A (contrast values y1 to y3), a histogram for PD images from which the entire cerebrospinal fluid has been removed.

Second, after removing the pure cerebrospinal fluid from the extracted whole cerebrospinal cord and removing the portion corresponding to the partial volume of the cerebrospinal fluid from the region B (contrast values y2 to y3), which is the histogram for the remaining image, the white matter and gray matter part remain Volume should be considered.

First, a histogram (region A of FIG. 14) is generated from an image (FIG. 12C) from which the entire cerebrospinal fluid is removed from a given PD image (A or C of FIG. 11). He mentioned that this part contains only white matter and gray matter, not the cerebrospinal fluid within a certain thickness. When the histogram distribution function is H, H (y) is the number of pixels corresponding to the intensity value y. Here, the contrast value G when the components of the white matter having the contrast value y1 are mixed with z% and the components of the gray matter having the contrast value y3 (1-z)% are as follows. G = y ₁ * z + y ₂ * (1-z) .......... (o <z <1) ...... Equation (1). In this case, H (G) is the number of pixels corresponding to the contrast value G, of which z% means the ratio corresponding to the partial volume of the white matter and is expressed as a percentage, not an absolute value. It is a value for the calculation of an expression and does not mean an absolute value based on a percentage. Accordingly, the total PV ₁ of the pixels corresponding to the partial volume of the white matter among the total pixels corresponding to each intensity value in the interval between the intensity values y1 and y3 is as follows. PV1 = H (y ₁ * z + y ₃ * (1-z)) zdy ........... Equation (2)

Next, the pure cerebrospinal fluid (region C of FIG. 14) of the extracted whole cerebrospinal fluid is removed, and a histogram (region B of FIG. 14) of the remaining images is obtained. This is the part where white matter, gray matter and cerebrospinal fluid exist within a certain thickness. In order to obtain the partial volume of the white matter or gray matter from the histogram, the portion corresponding to the partial volume of the cerebrospinal fluid must be removed first.

When the distribution function of the histogram for the region B of FIG. 14 is W, W (G) is the number of pixels corresponding to the contrast value G. The contrast value G that appears when a portion of the cerebrospinal fluid with a contrast value y3 within a certain thickness includes (1-z)% of a component having a contrast value of y2 otherwise is as follows. G = y ₃ * z + y ₂ * (1-z) ............. Equation (3)

In this case, W (G) is the number of pixels corresponding to the contrast value G, and z% is the ratio corresponding to the partial volume of the cerebrospinal fluid. From the light intensity values y3 to y2, this ratio of the total pixels corresponding to each light intensity value is removed as the cerebrospinal fluid, and the sum of the pixels corresponding to the partial volume of the removed cerebrospinal fluid is as follows. PW = W (y ₃ * z + y ₂ * (1-z)) zdy ....... Equation (4)

FIG. 15 is a PD image excluding the separation of the whole cerebrospinal fluid and the influence of the cerebrospinal fluid extracted from the PD image. Figure 15 (A) is the total cerebrospinal fluid extracted from the PD image. FIG. 15B is the cerebrospinal fluid in which the portion corresponding to the portion of the cerebrospinal fluid expressed by Equation (4) and the pure cerebrospinal fluid in 15 (A) are removed. In these areas, the partial volume of white and gray matter should be considered. 15C is a part corresponding to the pure cerebrospinal fluid and partial volume of the cerebrospinal fluid removed from the whole cerebrospinal fluid. That is, FIG. 15 (A) separated the part of 15 (B) and 15 (C). FIG. 15D is a PD image in which a pure cerebrospinal fluid and a partial volume of cerebrospinal fluid (D in FIG. 15) are removed from a given PD image (FIG. 11C), and the image of the cerebrospinal fluid is completely removed. And gray matter will be separated.

In the histogram corresponding to the region B of FIG. 14, when the distribution function of the histogram for the remaining portion except for the PW of Equation (4) is W ', It can be removed and viewed as a histogram for the remaining white and gray matter, where the partial volume of the white matter is obtained as: PV2 = W (y ₁ * z + y ₃ * (1-z)) zdy ............ Equation (5)

16 is a histogram showing a method of calculating the partial volume of white matter. Thus, the sum of PV ₂ of the formula (2) PV ₁ and equation (5) corresponds to parts by volume of white matter in the entire image PD. Meanwhile, the partial volume of the gray matter may be excluded by the partial volume of the white matter calculated from the sum of all pixels of the histogram of FIG. 16. The following equation is the number of pixels corresponding to the partial volume of gray matter.

By using the partial volume of white matter thus far, the discrimination value that can separate white matter and gray matter can be determined as follows. First, histograms are generated for images of bright cerebrospinal fluid that did not affect white matter and gray matter and cerebrospinal fluid that removed white matter and gray matter (FIG. 15D). When the histogram distribution function is T, T (G) is the number of pixels corresponding to the contrast value G. In this case, the number of pixels corresponding to each intensity value is accumulated while increasing from the intensity value y1 of white matter. The contrast value t that satisfies Equation (7) for the first time is determined as the discrimination value. Equation (7) shows this meaning, and FIG. 17 is a histogram showing this determination of determination.

So far, PD algorithm and T2 image are used to generate PD image to be used for separation of white matter and gray matter by algorithm according to the present invention, and histogram is generated from each image, and method of determining discrimination value by calculating partial volume of white matter Has been described above.

Fig. 18 shows the 8th and 9th PD images separated into white matter and gray matter using the discrimination value determined by this method. 19 is a PD image slice separated into white matter and gray matter with respect to the whole slice, the black part is white matter and the other part is the gray matter region.

By separating the white matter and the gray matter for each slice by the above method, each volume is calculated using the number of pixels constituting the separated white matter and the gray matter.

In the present invention, the information in the header of each image was used to obtain the volume of the white matter and the gray matter. In the header of each image, information on an individual's personal information such as a patient's name, age, gender, etc., information on photographing conditions, and various physical characteristics of the image are stored. Among these information, information about the size of the pixels constituting the image and information about the thickness of the brain sections taken by the magnetic resonance imaging apparatus or the spacing between slices in order to generate a compressed image, the white matter and gray matter extracted from the whole slice This is useful for calculating the volume of.

In the present invention, the following equation was used to calculate the volume of the extracted white matter and gray matter. The number of slices containing the extracted white matter (or gray matter) is referred to as N, the interval of slices is referred to as D, and the number of pixels constituting the white matter (or gray matter) extracted from the i-th slice is determined. Let's say that one pixel's width is X, and one pixel's height is Y,

Until now, the partial volume of white matter and gray matter contained within a certain thickness is calculated by analyzing the contrast values in the images generated by the magnetic resonance imaging apparatus of the brain, and based on this, a discrimination value capable of separating the white matter and gray matter from each image slice We also presented a method of determining the volume of white matter and gray matter using the number of pixels constituting the separated white matter and gray matter and information in the header of the image. Until now, no method has been developed to calculate the partial volume of white matter and gray matter within a certain thickness. Therefore, it is impossible to know each volume precisely. Also, using this information, various mental disorders, such as Alzheimer's disease and Dalhou syndrome, are used. There was no early prediction method. Therefore, by using the series of methods proposed in the present invention, first, the volume of white matter and gray matter for each age group is calculated to obtain statistics, and then the volume of white matter and gray matter is calculated for patients with possible mental disorders. Compared to the results, it can be used for early treatment and diagnosis of various mental diseases.

Claims (4)

A preprocessing process of removing portions of the PD image except for the brain and removing identical regions from the corresponding T2 images; A cerebrospinal cord separation process of removing the cerebrospinal fluid from the pre-processing completed T2 image and separating the cerebrospinal fluid of the same region from the corresponding PD image; Generating a first histogram from the PD image consisting of only white matter and gray matter by separating the cerebrospinal fluid; Calculating a partial volume PV1 of white matter from the first histogram; Removing the pure cerebrospinal fluid from the cerebrospinal cord images separated by the cerebrospinal cord separation process and generating a second histogram from the remaining images; Calculating a partial volume of the pure cerebrospinal fluid from the second histogram; Calculating a partial volume (PV2) occupied by white matter in the region where the pure cerebrospinal fluid is removed; Calculating a partial volume of gray matter by subtracting the sum of the partial volumes of the white matter (PV1 + PV2) from the total pixel values of the first histogram; Generating a third histogram including the first histogram and the second histogram; Integrating the distribution function while increasing the contrast value in the third histogram, the determination value calculation process of setting the first contrast value whose value becomes larger than the sum of the partial volumes of the white matter (PV1 + PV2) as a discrimination value and, Generating an image in which white matter and gray white matter are separated for each slice based on the determination value; A method for separating and calculating white matter and gray matter in a brain image, comprising calculating the respective volumes of the white matter and gray matter using slice information of the white matter and gray matter and slice information provided by the tomography apparatus. . The method of claim 1; The pretreatment process Removing the background of the image by using the intensity information of the image constituting the PD image; Setting up four fat layer tracking pixels at the top, bottom, left and right of the fat layer that is elliptical in the brain envelope, Based on 256 gray levels, the fat layer with a contrast value of 60 or less is traced by 4-connectivity and removed with the skin; Scanning the part of the outer skin that is not removed using the structuring element (SE), which is a morphological filter, along the periphery of the brain and removing the protrusions that do not enter the given SE; Separation and volume calculation method of white matter and gray matter in the brain image, characterized in that the process of removing the same region in the corresponding T2 image. The method of claim 1; The process of calculating the partial volume, From the histogram H of the PD image, in which the whole cerebrospinal fluid is removed and only white matter and gray matter are extracted, the contrast value y1 of pure white matter and the contrast value y3 of pure gray matter are extracted. , the contrast value G which appears when the gray matter component having a contrast value of y3 is included in the ratio of (1-z) is G = y ₁ * z + y ₃ * (1-z) ........ .. (0 <z <1) the process of calculating, PV1 = the number of pixels corresponding to the partial volume of white matter distributed from the intensity values y1 to y3 H (y ₁ * z + y ₃ * (1-z)) zdy to obtain the first partial volume (PV1) of white matter by z, and From the histogram W on the PD images of pure cerebrospinal fluid removed and mixed with white matter and gray matter, we extract the contrast y3 of the cerebrospinal fluid and the contrast y2 of the components that are not. The contrast value G that appears when a non-cerebrospinal component with a contrast value of y3 is included as a ratio of (1-z) is G = y ₃ * z + y ₂ * (1-z) ....... ...... (0 <z <1) PW = the number of pixels corresponding to the partial volume of the cerebrospinal fluid distributed from the intensity y2 to y3 W (y ₃ * z + y ₂ * (1-z)) zdy's formula calculates the total number of pixels in the cerebrospinal fluid by the ratio of z, The histogram W 'is removed from the histogram W. The white matter having a contrast value of y1 within a predetermined thickness includes the gray matter component having a contrast value of z and y3 at a ratio of (1-z). To calculate the contrast value G that appears when G = y ₁ * z + y ₃ * (1-z) ..... (0 <z <1), PV2 = the number of pixels corresponding to the partial volume of white matter distributed from intensity y2 to y3 Calculating the second partial volume PV1 of white matter by z in the formula W (y ₁ * z + y ₃ * (1-z)) zdy, And calculating the volume of the pixels constituting the white matter by adding the first partial volume (PV1) and the second partial volume of the white matter to the white matter and the gray matter in the brain image. The method of claim 1; The total volume of white matter (or gray matter) by the whole image slice is The number of slices containing the extracted white matter (or gray matter) is referred to as N, the interval of slices is referred to as D, and the number of pixels constituting the white matter (or gray matter) extracted from the i-th slice is determined. Let's say that one pixel's width is X, and one pixel's height is Y, A method for separating and calculating white matter and gray matter in a brain image, comprising calculating the volume of white matter (or gray matter) from the equation.