TWI679652B - Method, non-transitory computer-readable media and apparatus for evaluating personalized brain imaging - Google Patents

Abstract

A method for personalizing brain image evaluation includes the following steps: receiving a first T1-weighted image of a patient's brain; performing a tissue segmentation procedure on the first T1-weighted image to obtain a first gray matter image; and registering the first gray matter image To the normalized normal gray matter image template created by the superimposition procedure; according to the first gray matter image registered to the normalized normal gray matter image template, a voxel analysis formula is used to obtain the score value corresponding to each voxel; it is determined that each voxel corresponds to Whether the value of the score is greater than or less than the expected value; and based on those voxels greater than or less than the expected value, the brain region to which it belongs is pushed back, and an evaluation report is output. Therefore, the differences in brain imaging between a single patient and a healthy subject can be known by excluding relevant interference factors.

Description

Method for personalizing brain image assessment, non-transitory computer-readable media and equipment

The invention relates to a method for image evaluation, a non-transitory computer-readable medium and a device, and more particularly to a method for evaluating a personalized brain image, a non-transitory computer-readable medium and a device.

At present, the statistical analysis method commonly used in imaging is T-test, which is a many-to-many group analysis, for example: comparison between a group of patients and a group of healthy people. If this analysis method is applied to personalized evaluation It is necessary to treat a single patient as the first group and all other healthy subjects as the second group. However, in the statistical test of the T test, it is necessary to assume that the two groups have the same mean and variance ( variance). Therefore, when applied to personalized evaluation, there is a problem that does not meet this statistical assumption because there is only one person in the first group. In addition, if this analysis method is still applied to the personalized evaluation, the statistical model and the number of database samples will be insufficient because there is only one person in the first group, which will cause the problem of reduced sensitivity.

In summary, it can be known that the T test in the prior art has a problem that the statistical model and the number of database samples are insufficient due to only one person in the first group, which reduces the sensitivity. Therefore, it is necessary to propose improved technical methods to solve this problem .

The invention discloses a method, a non-transitory computer-readable medium and a device for personalizing brain image evaluation.

First, the present invention discloses a method for personalizing brain image evaluation. The steps include: receiving a first T1 weighted image of a patient's brain; and performing a tissue segmentation procedure on the first T1 weighted image to obtain a first Gray matter image; register the first gray matter image to the normalized normal gray matter image template created by the superposition procedure; obtain the first gray matter image registered to the normalized normal gray matter image template according to the voxel analysis formula to obtain each voxel it has Corresponding score value, where the voxel analysis formula is , TW is the score value corresponding to the i-th voxel, and x * is the volume size of the i-th voxel of the first gray matter image registered to the normalized gray matter image template, In order to obtain the volume of the i-th voxel of the expected image from the database based on the age and sex of the patient, s is the volume standard deviation of the i-th voxel of the estimated residual image from the database, Is the scale factor for the T test, n is the number of samples to build the database, and i is a positive integer greater than or equal to 1; determine whether the score value corresponding to each voxel is greater than or less than the expected value; based on those greater or less than the expected value The voxel pushes back the brain region to which it belongs and outputs an evaluation report.

In addition, the present invention discloses a non-transitory computer-readable medium for personalized brain imaging assessment, which is configured to store a number of instructions that, when executed by one or more processors, cause the one or more processors to The following operations are performed: receiving a first T1-weighted image of a patient's brain; performing a tissue segmentation procedure on the first T1-weighted image to obtain a first gray matter image; registering the first gray matter image to a normalization established by the superposition procedure Normal gray matter image template; the first gray matter image registered to the normalized normal gray matter image template is used to obtain the score value corresponding to each voxel it has, where the voxel analysis formula is , TW is the score value corresponding to the i-th voxel, and x * is the volume size of the i-th voxel of the first gray matter image registered to the normal image template, In order to obtain the volume of the i-th voxel of the expected image from the database based on the age and sex of the patient, s is the volume standard deviation of the i-th voxel of the estimated residual image from the database, Is the scale factor for the T test, n is the number of samples to build the database, and i is a positive integer greater than or equal to 1; determine whether the score value corresponding to each voxel is greater than or less than the expected value; based on those greater or less than the expected value The voxel pushes back the brain region to which it belongs and outputs an evaluation report.

Furthermore, the present invention discloses a device for personalizing brain image evaluation. The device includes: one or more processors, a storage unit, and at least one program, wherein the at least one program is stored in the storage unit and configured to be executed by the device. Executed by one or more processors, the at least one program generally includes the following operation instructions: receiving a first T1 weighted image of a patient's brain; performing a tissue segmentation procedure on the first T1 weighted image to obtain a first gray matter image ; Register the first gray matter image to the normalized normal gray matter image template created by the superposition procedure; obtain the first gray matter image registered to the normalized normal gray matter image template according to the voxel analysis formula to obtain the corresponding Scoring value, where the voxel analysis formula is , TW is the score value corresponding to the i-th voxel, and x * is the volume size of the i-th voxel of the first gray matter image registered to the normalized gray matter image template, In order to obtain the volume of the i-th voxel of the expected image from the database based on the age and sex of the patient, s is the volume standard deviation of the i-th voxel of the estimated residual image from the database, Is the scale factor for the T test, n is the number of samples to build the database, and i is a positive integer greater than or equal to 1; determine whether the score value corresponding to each voxel is greater than or less than the expected value; The voxel pushes back the brain region to which it belongs and outputs an evaluation report.

The system and method disclosed in the present invention are as above. The difference from the prior art is that the present invention receives the first T1 weighted image of the patient's brain; performs a tissue segmentation procedure on the first T1 weighted image to obtain the first gray matter image; The first gray matter image is registered to the normalized normal gray matter image template created by the superimposition process; the first gray matter image registered to the normalized normal gray matter image template is used to obtain the corresponding score value for each voxel it has through the voxel analysis formula. ; Determine whether the score value corresponding to each voxel is greater than or less than the expected value; and push back the brain region to which it belongs according to those voxels greater than or less than the expected value, and output an evaluation report.

Through the above-mentioned technical means, the present invention can know the difference between the brain image between a single patient and a healthy subject without excluding relevant interference factors (age factor and gender factor), which is helpful for data quantification of the brain image and detection of the naked eye. Undetectable subtle changes in brain structure.

In the following, the embodiments of the present invention will be described in detail with reference to the drawings and examples, so as to fully understand and implement the implementation process of how the present invention applies technical means to solve technical problems and achieve technical effects. It should be noted that the images described in this case are all three-dimensional images. In order to avoid the drawings being too complicated, only the two-dimensional images are used to present the schematic diagrams of each embodiment. In addition, although the terms "first", "second", etc. may be used herein to describe various images, these images should not be limited by these terms.

Please refer to "Figure 1" first. "Figure 1" is a method flowchart of an embodiment of a method for personalizing brain image assessment according to the present invention. The steps include: receiving a first T1-weighted image of a patient's brain (step 110). ); Perform a tissue segmentation procedure on the first T1 weighted image to obtain a first gray matter image (step 120); register the first gray matter image to a normalized normal gray matter image template created by the superposition procedure (step 130); The first gray matter image registered to the normalized normal gray matter image template obtains a score value corresponding to each voxel it has according to a voxel analysis formula, where the voxel analysis formula is , TW is the score value corresponding to the i-th voxel, and x * is the volume size of the i-th voxel of the first gray matter image registered to the normalized gray matter image template, In order to obtain the volume of the i-th voxel of the expected image from the database based on the age and sex of the patient, s is the volume standard deviation of the i-th voxel of the estimated residual image from the database, Is the scale factor for the T test, n is the number of samples to build the database, i is a positive integer greater than or equal to 1 (step 140); determine whether the score value corresponding to each voxel is greater than or less than the expected value; (step 150); According to those voxels that are larger or smaller than the expected value, the brain region to which they belong is pushed back, and an evaluation report is output (step 160).

The patient described in step 110 may be, but is not limited to, a patient who may be involved in the assessment of a possible abnormal state of the brain. After passing through the above steps 110 to 160, the patient can learn that he is related to healthy subjects by excluding relevant interference factors. The differences between brain images can help to digitize brain images and detect subtle changes in brain structure that cannot be detected by the naked eye. Among them, the patient is an adult.

The tissue segmentation procedure described in step 120 may be, but is not limited to, using the open source software FSL to perform tissue segmentation on the first T1 weighted image to obtain the patient's gray matter image (ie, the first gray matter image), white matter image, and cerebrospinal fluid image. However, this embodiment is not intended to limit the present invention, and can be adjusted according to actual needs. For example, the gray matter image (ie, the first gray matter image) of the patient may also be obtained through a functional magnetic resonance data processing technique of statistical parametric mapping (SPM).

Please refer to "Fig. 1" and "Fig. 2". "Fig. 2" is a flowchart of one embodiment of the superimposing procedure described in step 130 of "Fig. 1". In this embodiment, the superposition procedure described in step 130 includes the following steps: obtaining second T1-weighted images of the brains of n healthy subjects, where n is a positive integer greater than or equal to thirty (step 210); Each second T1 weighted image is subjected to a tissue segmentation procedure to obtain corresponding second gray matter images (step 220); the second gray matter images are respectively registered to the MNI (Montreal Neurological Institute) standard space template (step 230) ; And superimposing the second gray matter images registered to the MNI standard space template, and establishing a normalized normal gray matter image template by averaging the image brightness (step 240). Among them, the MNI standard space template is a standard brain template widely used in the field of international brain imaging.

Since this embodiment is used to perform personalized brain image evaluation on adults, the healthy subjects described in step 210 are healthy adults.

In step 240, the normalized normal gray matter image template is established by averaging the image brightness. Each voxel of each second gray matter image registered to the MNI standard space template has its brightness value. When registered to the MNI, After superimposing the second gray matter images of the standard space template, the luminance values of the i-th voxel (i is a positive integer greater than or equal to 1) of different second gray matter images can be added to obtain the average value, and Obtain the brightness value of the i-th voxel of the normal image template, and then establish a normalized normal gray matter image template.

Since the above n healthy subjects can all be Taiwanese, a standard space template (ie, a normalized normal gray matter image template) applicable to Taiwan can be obtained through steps 210 to 240. Therefore, the first T1-weighted images of the brains of healthy subjects in different regions can be used to establish a standard spatial template suitable for that region.

Please refer to "Figure 1" and "Figure 3", and "Figure 3" is a method flowchart of an embodiment of the method for establishing a database described in step 140 of "Figure 1". In this embodiment, since “the first gray matter image registered to the normalized normal gray matter image template is obtained according to the voxel analysis formula to obtain the score value corresponding to each voxel” described in step 140, it is necessary to borrow Required for obtaining voxel analysis formula from database And s, therefore, the database must be established before executing step 140, and the method of establishing the database includes the following steps: registering each second gray matter image to a normalized normal gray matter image template (step 310); Each second gray matter image normalized to a normal gray matter image template is smoothed (step 320); the smoothed second gray matter images are based on a general linear model and integrate age factors and gender factors to estimate the age factor. A weighting coefficient and a weighting coefficient of a gender factor, and obtaining a smoothed residual image corresponding to each second gray matter image (step 330); and calculating a standard deviation of the volume size of the i-th voxel of each residual image, To obtain the volume standard deviation of the i-th voxel of the estimated residual image, and then establish a database (step 340).

In this embodiment, the smoothing process described in step 320 may be, but is not limited to, a three-dimensional Gaussian flattening process, and may be adjusted according to actual needs. Among them, the smoothing process can reduce the error when the second gray matter image is registered to the normalized normal gray matter image template.

Since age and gender are related factors that affect brain changes, in step 330, when the smoothed second gray matter images are analyzed using a general linear model, the age factor and gender factor are integrated to estimate The weight coefficient of the age factor and the weight factor of the gender factor are obtained, and a residual image corresponding to each smoothed second gray matter image is obtained. Among them, the weight coefficients of the estimated age factor and the gender factor may be a numerical range, respectively. Next, calculate the standard deviation of the volume of the i-th voxel of each residual image to obtain the estimated standard deviation of the volume of the i-th voxel of the residual image (ie, s of the voxel analysis formula), and then establish a database (Ie step 340). Therefore, the database is a database using linear regression of voxels based on age and gender factors. In this embodiment, the factors integrated by using a general linear model for analysis may be an age factor and a gender factor, but this embodiment is not intended to limit the present invention. For example, using a general linear model to analyze the integrated factors can also integrate various factors that affect brain changes, such as education level or cognitive function performance, etc., which can be adjusted according to actual needs.

After the database is established through steps 310 to 340, the "registration of the first gray matter image registered to the normalized normal gray matter image template according to the voxel analysis formula to obtain the score value corresponding to each voxel" described in step 140 is performed. among them, This is the volume of the i-th voxel of the expected image obtained by using the age and gender voxel linear regression database based on the age and sex of the patient, and s is the estimated residual image obtained from the database The volume standard deviation of the i-th voxel.

Please continue to refer to "Figure 1". In step 150, the expected value can be, but not limited to, 0.05, and can be adjusted according to actual needs.

In step 160, the brain regions to which these voxels larger or smaller than a preset value belong may be, but not limited to, Frontal Lobe, Parietal Lobe, Occipital Lobe, and Temporal lobe), island leaves (Insula), hippocampus (Hippocampus) or amygdala (Amygdala). The content of the evaluation report may include visually marking the first gray matter image on the voxels whose score value is greater than or less than the expected value. Therefore, through the above steps 110 to 160, the difference between the brain image between a single patient and a healthy subject can be learned under the exclusion of relevant interference factors (age factor and gender factor), which is helpful for the data visualization and detection of brain images. Changes in subtle brain structures that are not detectable to the naked eye.

Next, please refer to "Figure 4". "Figure 4" is a block diagram of an embodiment of a non-transitory computer-readable medium for personalized brain image evaluation according to the present invention. The non-transitory computer-readable medium 400 for personalized brain imaging assessment includes operating instructions 402 that can be used by one or more processors 404 to perform methods including the brain imaging assessment described above. Such operating instructions may include any one or more other steps set forth herein. Implementing operating instructions 402 such as those described herein may be stored on a non-transitory computer-readable medium 400 of a personalized brain imaging assessment. The non-transitory computer-readable medium 400 for personalized brain imaging assessment may be a storage medium such as a magnetic disk or optical disk or a magnetic tape, or any other suitable non-transitory computer-readable medium known in the art.

Next, please refer to "FIG. 5". "FIG. 5" is a block diagram of an embodiment of a device for personalizing brain image evaluation according to the present invention. The device 500 for personalized brain imaging evaluation includes a processor 502, a storage unit 504, and a program 506. The program 506 is stored in the storage unit 504 and configured to be executed by the processor 502. The program 506 as a whole may include the above-mentioned Instructions for the method of brain image evaluation. The number of the processor 502, the storage unit 504, and the program 506 may be, but is not limited to, one, and may be adjusted according to actual needs.

The following description is made by way of example in conjunction with "Figure 6" to "Figure 7B".

(Stability analysis of samples from a healthy adult database for personalized brain imaging assessment)

Because the method of personalized brain image assessment is a method for evaluating the degree of change of the gray matter image of a single patient compared to a database established through the gray matter image of a normal healthy subject, it is used to establish a database The number of gray matter images of normal healthy subjects may affect the evaluation of the gray matter image stability of a single patient. Please refer to "Figure 6", "Figure 6" is the first gray matter image of the evaluation report output by the personalized brain image evaluation method for the same patient based on the database created by different sample sizes. Among them, n is the number of samples for establishing a database, and each first gray matter image is visually labeled with voxels whose evaluation value is greater than or less than the expected value. In this example, the brain gray matter images of 61-year-old female Alzheimer's patients were individually evaluated using a database created by different sample numbers. As shown in "Figure 6", when used to establish When the number of gray matter images of normal healthy subjects in the database is 30, the results can be very similar to the database established by the sample number of 152. In other words, the method of personalized brain image evaluation can use the database established by a small number of samples to obtain extremely stable statistical results.

(Sensitivity analysis of personalized brain imaging assessment methods)

In this embodiment, the sensitivity of detecting hippocampal volume atrophy is discussed. Please refer to "Figure 7A" and "Figure 7B", "Figure 7A" and "Figure 7B" respectively use the method of personalized brain image evaluation of "Figure 1" and the traditional T test statistical method for the same patient Schematic illustration of computer simulation of hippocampal atrophy. Among them, the vertical axis is the percentage of hippocampal atrophy; the horizontal axis is the number of voxels corrected for corresponding clustering-wise multiple comparison; the light-colored area is when the uncorrected P value is less than 0.05; the dark-colored area is When the cluster level corrected P value is less than 0.05. In the present embodiment, a P value of less than 0.05 is regarded as statistically significant.

The method of personalizing brain image evaluation in this embodiment uses a database established by a sample size of 152 people for evaluation. According to "Figure 7A", the method of personalizing brain image evaluation can generate 9% of the hippocampal volume. The patient was detected to have an abnormal volume decrease during atrophy; the traditional T-test statistical method can only be detected when the hippocampal body has a 12% atrophy before the patient has detected an abnormal volume decrease. Therefore, it can be seen that the method of personalized brain imaging assessment can provide better sensitivity in detecting the gray matter volume of a single patient's brain.

In summary, the difference between the present invention and the prior art is that the first T1 weighted image of the patient's brain is received; the tissue segmentation procedure is performed on the first T1 weighted image to obtain the first gray matter image; The image is registered to the normalized normal gray matter image template created by the superimposition procedure; the first gray matter image registered to the normalized normal gray matter image template is used to obtain the corresponding score value of each voxel it has through the voxel analysis formula; Whether the evaluation value corresponding to the integrated element is greater than or less than the expected value; and based on those voxels greater than or less than the expected value, the brain region to which the voxel belongs is pushed back and an evaluation report is output. The problem of the brain image can be obtained by excluding related interference factors (age factor and gender factor), which can help to visualize the brain image and detect the invisible to the naked eye. Subtle changes in brain structure.

Although the present invention is disclosed in the foregoing embodiments as above, it is not intended to limit the present invention. Any person skilled in similar arts can make some modifications and retouches without departing from the spirit and scope of the present invention. The scope of patent protection shall be determined by the scope of the patent application attached to this specification.

400‧‧‧ Non-transitory computer-readable media for personalized brain imaging assessment

402‧‧‧operation instructions

404, 502‧‧‧ processors

500‧‧‧ Personalized Brain Imaging Evaluation Equipment

504‧‧‧Storage unit

506‧‧‧program

Step 110‧‧‧ receives a first T1-weighted image of the brain of a patient

Step 120‧‧‧ Perform a tissue segmentation procedure on the first T1 weighted image to obtain a first gray matter image

Step 130‧‧‧ Register the first gray matter image to the normalized normal gray matter image template created by the superposition process

In step 140‧‧‧, the first gray matter image registered to the normalized normal gray matter image template is used to obtain a score value corresponding to each voxel according to the voxel analysis formula, where the voxel analysis formula is , TW is the score value corresponding to the i-th voxel, and x * is the volume size of the i-th voxel of the first gray matter image registered to the normalized gray matter image template, In order to obtain the volume of the i-th voxel of the expected image from the database based on the age and sex of the patient, s is the volume standard deviation of the i-th voxel of the estimated residual image from the database, Is the scale factor for the T test, n is the number of samples to build the database, and i is a positive integer greater than or equal to 1

Step 150‧‧‧ determines whether the score value corresponding to each voxel is greater than or less than the expected value

Step 160‧‧‧ Push back the corresponding brain region according to those voxels that are larger or smaller than the expected value, and output an evaluation report

Step 210‧‧‧ Obtain second T1-weighted images of the brains of n healthy subjects, where n is a positive integer greater than or equal to thirty

Step 220‧‧‧ performs a tissue segmentation procedure on each second T1 weighted image to obtain a corresponding second gray matter image

Step 230‧‧‧ respectively register the second gray matter images to the MNI standard space template

Step 240‧‧‧ superimpose the second gray matter images registered to the MNI standard space template, and establish a normalized normal gray matter image template by averaging the image brightness

Step 310‧‧‧ Register each second gray matter image to a normalized normal gray matter image template

Step 320‧‧‧ smoothing each second gray matter image registered to the normalized normal gray matter image template

Step 330‧‧‧ The smoothed second gray matter images are based on a general linear model and are integrated with the age factor and the gender factor to estimate the weight factor of the age factor and the weight factor of the gender factor, and obtain each smoothed second Residual image corresponding to gray matter image

Step 340‧‧‧ calculates the standard deviation of the volume size of the ith voxel of each residual image to obtain the estimated volume standard deviation of the ith voxel of the residual image, and then establish a database

FIG. 1 is a method flowchart of an embodiment of a method for personalizing brain image assessment according to the present invention. FIG. 2 is a flowchart of a method according to an embodiment of the overlapping procedure described in step 130 of FIG. 1. FIG. 3 is a method flowchart of an embodiment of the method for establishing a database described in step 140 of FIG. 1. FIG. 4 is a block diagram of an embodiment of a non-transitory computer-readable medium for personalized brain imaging assessment according to the present invention. FIG. 5 is a block diagram of an embodiment of a device for personalizing brain image evaluation according to the present invention. Figure 6 is the first gray matter image of the evaluation report output from the database created by different sample numbers for the same patient with a personalized brain image evaluation method. Figures 7A and 7B are schematic diagrams of computerized atrophy assessment of the hippocampus of the same patient using the personalized brain image evaluation method of Figure 1 and the traditional T test statistical method.

Claims (7)

A method for personalizing brain image evaluation, the method includes the following steps: receiving a first T1 weighted image of a patient's brain; and performing a tissue segmentation procedure on the first T1 weighted image to obtain a first A gray matter image; registering the first gray matter image to a standardized normal gray matter image template established by a superimposition procedure; obtaining the first gray matter image registered to the standardized normal gray matter image template according to an integrin analysis formula It has a score value corresponding to each voxel, wherein the voxel analysis formula is , TW is the scoring value corresponding to the i-th voxel, and x * is the volume size of the i-th voxel of the first gray matter image registered to the normalized normal gray matter image template, In order to obtain the volume of the i-th voxel of an expected image in a database based on the age and sex of the patient, s is the volume standard deviation of the i-th voxel in an estimated residual image obtained from the database, Is a scale factor for the T test, n is the number of samples to build the database, i is a positive integer greater than or equal to 1; determine whether the score value corresponding to each voxel is greater than or less than an expected value; and based on greater than Or the voxels smaller than the expected value are pushed back to the brain region to which they belong, and an evaluation report is output. The method for personalizing brain image evaluation according to item 1 of the patent application scope, wherein the superimposing procedure includes the following steps: obtaining a second T1-weighted image of the brain of one of n healthy subjects, where n is greater than or equal to thirty A positive integer; performing the tissue segmentation procedure on each of the second T1 weighted images to obtain a corresponding second gray matter image; registering the second gray matter images to the MNI (Montreal Neurological Institute) standard space template ; And superimposing the second gray matter images registered to the MNI standard space template, and establishing the normalized normal gray matter image template by averaging the image brightness. The method for personalizing brain image evaluation according to item 2 of the scope of patent application, wherein the method of establishing the database includes the following steps: registering each of the second gray matter images to the standardized normal gray matter image template; Each of the second gray matter images of the normal image template is subjected to a smoothing process; the smoothed second gray matter images are based on a general linear model and an age factor and a gender factor are integrated to estimate a weight coefficient of the age factor and the Weight coefficient of the gender factor, and obtain a residual image corresponding to each smoothed second gray matter image; and calculate the standard deviation of the volume size of the i-th voxel of each residual image to obtain the estimate The volume standard deviation of the i-th voxel in the residual image is used to establish the database. A non-transitory computer-readable medium for personalized brain imaging assessment, configured to store a number of operating instructions that, when executed by one or more processors, cause the one or more processors to perform the following operations Receiving a first T1 weighted image of a patient's brain; performing a tissue segmentation procedure on the first T1 weighted image to obtain a first gray matter image; registering the first gray matter image to a superimposed procedure A normalized normal gray matter image template established; the first gray matter image registered to the normalized normal gray matter image template is used to obtain a score value corresponding to each voxel it has according to the integrin analysis formula, wherein the voxel analysis formula for , TW is the scoring value corresponding to the i-th voxel, and x * is the volume size of the i-th voxel of the first gray matter image registered to the normalized normal gray matter image template, In order to obtain the volume of the i-th voxel of an expected image in a database based on the age and sex of the patient, s is the volume standard deviation of the i-th voxel in an estimated residual image obtained from the database, Is a scale factor for the T test, n is the number of samples to build the database, i is a positive integer greater than or equal to 1; determine whether the score value corresponding to each voxel is greater than or less than an expected value; and based on greater than or Those voxels smaller than the expected value push back the brain region to which they belong and output an evaluation report. A non-transitory computer-readable medium for personalized brain imaging assessment according to item 4 of the scope of the patent application, wherein the overlay procedure includes the following operations: obtaining a second T1-weighted image of one of the brains of n healthy subjects, n Is a positive integer greater than or equal to thirty; the tissue segmentation procedure is performed on each of the second T1 weighted images to obtain a corresponding second gray matter image; the second gray matter images are registered to an MNI standard respectively A spatial template; and superimposing the second gray matter images registered to the MNI standard spatial template, and establishing the normalized normal gray matter image template by averaging the image brightness. The non-transitory computer-readable medium for personalized brain image assessment according to item 5 of the scope of patent application, wherein the method of establishing the database includes the following operations: registering each of the second gray matter images to the standardized normal gray matter image A template; performing a smoothing process on each of the second gray matter images registered to the standardized normal gray matter image template; estimating the smoothed second gray matter images according to a general linear model and integrating an age factor and a gender factor to estimate A weighting coefficient of the age factor and a weighting coefficient of the gender factor, and obtaining a residual image corresponding to each smoothed second gray matter image; and performing the size of the i-th voxel of each residual image Standard deviation calculation to obtain the volume standard deviation of the i-th voxel of the estimated residual image, and then establish the database. A device for personalizing brain imaging assessment, the device comprising: one or more processors; a storage unit; and at least one program, wherein the at least one program is stored in the storage unit and configured to be processed by the one or more The processors execute the at least one program as a whole including the following operation instructions: receiving a first T1 weighted image of a patient's brain; and performing a tissue segmentation procedure on the first T1 weighted image to obtain a first gray matter Image; registering the first gray matter image to a normalized normal gray matter image template created by a superimposition process; registering the first gray matter image to the normalized normal gray matter image template according to an integrin analysis formula A score corresponding to each voxel of, wherein the voxel analysis formula is , TW is the scoring value corresponding to the i-th voxel, and x * is the volume size of the i-th voxel of the first gray matter image registered to the normalized normal gray matter image template, In order to obtain the volume of the i-th voxel of an expected image in a database based on the age and sex of the patient, s is the volume standard deviation of the i-th voxel in an estimated residual image obtained from the database, Is a scale factor for the T test, n is the number of samples to build the database, i is a positive integer greater than or equal to 1; determine whether the score value corresponding to each voxel is greater than or less than an expected value; and based on greater than Or the voxels smaller than the expected value are pushed back to the brain region to which they belong, and an evaluation report is output.