Lymphatic system damage quantification method and system based on multi-modal image marker

Abstract

The invention relates to a lymphatic system damage quantification method and system based on a multi-modal image marker, and the method comprises the steps: collecting multi-modal image data information, and carrying out the standardization preprocessing; determining multi-dimensional key image marker parameter information; constructing a mahalanobis distance model, and determining a mahalanobis distance between the multi-dimensional key image feature vector of the patient and the multi-dimensional key image feature vector of a corresponding group in a healthy population age hierarchical database; and evaluating the damage degree of the lymphatic-like system. Multi-dimensional key image marker information is determined through multi-modal image data information, a mahalanobis distance model is combined to determine a mahalanobis distance between a multi-dimensional key image feature vector of a patient and a multi-dimensional key image feature vector of a corresponding group in a healthy crowd age hierarchical database, and the damage degree of a lymphatic system is accurately evaluated. The problem of one-sidedness of single-index analysis is solved, and the lymphatic system damage is accurately quantified through the pathological deviation distance, so that the detection specificity of the lymphatic dysfunction is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and particularly relates to a method and system for quantifying glymphatic system damage based on multimodal imaging markers. Background Art

[0002] The glymphatic system is an important channel for clearing metabolic wastes in the brain, and its dysfunction is closely related to various neurological diseases (such as Alzheimer's disease, multiple sclerosis, etc.). The current clinical evaluation of glymphatic function through single-modal imaging markers (such as the ALPS index, choroid plexus volume, etc.) has the following problems: single markers are difficult to comprehensively reflect the complex pathological state of the glymphatic system; there is a lack of integrated analysis methods for multimodal features and it is impossible to quantify the overall damage degree; the existing evaluation methods do not consider the influence of age-related physiological changes on the reference values of the markers. For example, single-index evaluation methods (such as the ALPS index, PVS volume measurement) can only reflect local functions, simple linear combination methods (such as Z-score weighting) do not consider the mutual relationship between the markers, and age correction methods (such as linear regression) ignore the distribution characteristics of multi-dimensional data and it is difficult to effectively quantify the damage of the glymphatic system. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method and system for quantifying glymphatic system damage based on multimodal imaging markers in view of the above-mentioned deficiencies of the prior art.

[0004] The technical solution of the present invention for solving the above technical problem is as follows: A method for quantifying glymphatic system damage based on multimodal imaging markers includes the following steps:

[0005] Collect multimodal imaging data information and perform standardized preprocessing;

[0006] Determine multi-dimensional key imaging marker parameter information based on the standardized preprocessed multimodal imaging data information; wherein, the multi-dimensional key imaging marker parameter information includes choroid plexus volume fraction, ventricular cerebrospinal fluid volume fraction, enhanced perivascular space contrast volume, and mean free water in the white matter skeleton region.

[0007] Construct a Mahalanobis distance model, and determine the Mahalanobis distance between the multi-dimensional key imaging feature vector of the patient and the corresponding group of multi-dimensional key imaging feature vectors in the age-stratified database of healthy people based on the multi-dimensional key imaging marker parameter information.

[0008] Evaluate the degree of glymphatic system damage based on the Mahalanobis distance as the glymphatic damage index.

[0009] The beneficial effects of the present invention are as follows: The method for quantifying lymphatic system damage based on multi-modal imaging markers of the present invention determines multi-dimensional key imaging marker information from the multi-modal imaging data information after standard preprocessing, and then combines the Mahalanobis distance model to determine the Mahalanobis distance between the multi-dimensional key imaging feature vector of the patient and the multi-dimensional key imaging feature vector of the corresponding group in the age-stratified database of healthy people, realizing the precise assessment of the degree of lymphatic system damage. The one-sidedness problem of single-index analysis is solved through the non-linear integration of multi-dimensional imaging markers, and the lymphatic system damage is precisely quantified through the pathological deviation distance. The specificity of this method for detecting lymphatic dysfunction in stroke patients is greatly improved compared with the single ALPS index.

[0010] Based on the above technical solutions, the present invention can be further improved as follows:

[0011] Further: The collection of multi-modal imaging data information and the standard preprocessing specifically include the following steps:

[0012] Obtain multi-modal imaging data information and perform unified resampling processing using linear interpolation method according to the preset resolution;

[0013] Use ANTs software to perform N4 bias field correction processing on the multi-modal imaging data information after resampling processing.

[0014] The beneficial effects of the above further solution are: By performing resampling processing on the multi-modal imaging data information, the balance of the data information can be improved, the recognition ability of the model can be improved, and through the N4 bias field correction processing, the uniformity of the imaging data information can be improved, and the quality and reliability of the imaging data information can be improved.

[0015] Further: Determining the choroid plexus volume fraction based on the multi-modal imaging data information after standard preprocessing specifically includes the following steps:

[0016] Use FreeSurfer software to perform segmentation processing on the multi-modal imaging data information after standard preprocessing to obtain the lateral ventricle region;

[0017] Use the Gaussian mixture model to perform automatic segmentation within the lateral ventricle region to obtain the choroid plexus;

[0018] Based on the volume of the choroid plexus, calculate the choroid plexus volume fraction CPVF, and the calculation formula is as follows:

[0019]

[0020] where V ICV is the total intracranial volume, and V choroid is the volume of the choroid plexus.

[0021] The beneficial effects of the above further solution are as follows: The multi-modal image data information is segmented by FreeSurfer software to accurately obtain the lateral ventricle region, and then the choroid plexus is automatically segmented within the lateral ventricle region using the Gaussian mixture model. Finally, the choroid plexus volume fraction CPVF is calculated based on the volume of the choroid plexus, which is used as one of the multi-dimensional key image marker parameter information to evaluate the degree of glymphatic system damage.

[0022] Further: Determining the ventricular cerebrospinal fluid volume fraction based on the standardized preprocessed multi-modal image data information specifically includes the following steps:

[0023] The multi-modal image data information is segmented using the multi-atlas registration method to obtain the lateral ventricle region, the third ventricle, and the fourth ventricle;

[0024] Based on the volumes of the lateral ventricle region, the third ventricle, and the fourth ventricle, calculate the ventricular cerebrospinal fluid volume fraction. The calculation formula is:

[0025]

[0026] where V ICV is the total intracranial volume, and V CSF is the total cerebrospinal fluid volume of the ventricular region, including the cerebrospinal fluid volumes of the lateral ventricle, the third ventricle, and the fourth ventricle.

[0027] The beneficial effects of the above further solution are as follows: The multi-modal image data information is segmented using the multi-atlas registration method to obtain the lateral ventricle region, the third ventricle, and the fourth ventricle containing cerebrospinal fluid, and then the ventricular cerebrospinal fluid volume fraction is accurately calculated, which is used as one of the multi-dimensional key image marker parameter information to evaluate the degree of glymphatic system damage.

[0028] Further: Determining the enhanced perivascular space contrast volume based on the standardized preprocessed multi-modal image data information specifically includes the following steps:

[0029] Calculate the perivascular space contrast EPC based on the T1-weighted image and the T2-weighted image in the standardized preprocessed multi-modal image data;

[0030]

[0031] where T1 signal is the T1-weighted image, T2 signal is the T2-weighted image, and EPC is the perivascular space contrast;

[0032] Perform Frangi filtering enhancement processing on the perivascular space contrast EPC to obtain the enhanced perivascular space PVS volume V PVS , and the calculation formula is:

[0033]

[0034] Among them, BG is the basal ganglia region, DWM is the deep white matter region, x ∈ BG ∪ DWM represents the set of voxels belonging to the basal ganglia region and the deep white matter region, and H(·) is the Hessian matrix eigenvalue constraint.

[0035] The beneficial effects of the above further solution are as follows: The perivascular space contrast EPC is generated through T1-weighted images and T2-weighted images, thereby enhancing the visibility of the perivascular space. Then, through Frangi filtering enhancement processing, unstructured high-frequency spatial noise can be eliminated, and the resolution of the perivascular space can be greatly improved.

[0036] Further: Determining the free water mean value of the white matter skeleton region based on the multi-modal image data information after standardized preprocessing specifically includes the following steps:

[0037] Based on the multi-modal image data information after standardized preprocessing, the free water fraction FW is calculated using the dipy double tensor model;

[0038] Based on the multi-modal image data information after standardized preprocessing, diffusion statistical analysis is performed to extract the white matter skeleton region;

[0039] According to the free water fraction FW, the free water mean value of the white matter skeleton region is calculated.

[0040] The beneficial effects of the above further solution are as follows: The free water fraction is accurately calculated using the dipy double tensor model, and at the same time, the white matter skeleton region is extracted through diffusion statistical analysis. In this way, the free water mean value of the white matter skeleton region can be accurately calculated, which is used as one of the multi-dimensional key image marker parameter information to evaluate the degree of glymphatic system damage.

[0041] Further: The steps of constructing the Mahalanobis distance model and determining the Mahalanobis distance between the multi-dimensional key image feature vector of the patient and the corresponding group of multi-dimensional key image feature vectors in the healthy population age stratification database based on the multi-dimensional key image marker parameter information specifically include the following steps:

[0042] Construct a Mahalanobis distance model, group the multi-dimensional key image marker parameter information of the healthy population according to a preset age interval, and establish a healthy population age stratification database;

[0043] Calculate the multi-dimensional key image feature vectors of each group of data of the healthy population according to the healthy population age stratification database, and the multi-dimensional key image feature vectors include the mean vector μ and the covariance matrix Σ;

[0044] Determine the multi-dimensional key imaging feature vector of the patient according to the multi-dimensional key imaging marker parameter information of the patient, and calculate the Mahalanobis distance D between the multi-dimensional key imaging feature vector of the patient and the multi-dimensional key imaging feature vector of the healthy population and the patient M , and the calculation formula is:

[0045]

[0046] Wherein, X is the multi-dimensional key imaging feature vector of the patient, μ is the mean vector of the multi-dimensional key imaging marker parameter information of each group in the healthy population age-stratified database, and Σ is the covariance matrix of the multi-dimensional key imaging marker parameter information of each group in the healthy population age-stratified database.

[0047] The beneficial effect of the above further solution is: group by the multi-dimensional key imaging marker parameter information of the healthy population, so as to obtain the multi-dimensional key imaging feature vectors of the healthy population in different age groups, and then determine the Mahalanobis distance between the multi-dimensional key imaging feature vector of the patient and the multi-dimensional normal distribution parameters of the healthy population and the patient, that is, accurately quantify the lymphatic system damage through the pathological deviation distance.

[0048] Further: The specific steps of evaluating the degree of lymphatic system damage based on the Mahalanobis distance as the lymphatic damage index are as follows:

[0049] Perform ROC curve analysis based on the lymphatic damage index of the patient, and determine the lymphatic damage index threshold of the corresponding age group according to the Youden index method;

[0050] Classify the lymphatic damage index of the patient according to the lymphatic damage index threshold.

[0051] The beneficial effect of the above further solution is: by performing ROC curve analysis on the lymphatic damage index, the lymphatic damage index threshold of the corresponding age group can be obtained, so as to accurately classify the lymphatic damage index of the patient.

[0052] The present invention also provides a lymphatic system damage quantification system based on multi-modal imaging markers, including an acquisition and preprocessing module, a multi-dimensional parameter determination module, a Mahalanobis distance module, and a quantification module;

[0053] The acquisition and preprocessing module is used to acquire multi-modal imaging data information and perform standardized preprocessing;

[0054] The multi-dimensional parameter determination module determines multi-dimensional key imaging marker parameter information based on the multi-modal imaging data information after standardized preprocessing; wherein, the multi-dimensional key imaging marker parameter information includes choroid plexus volume fraction, ventricular cerebrospinal fluid volume fraction, enhanced perivascular space contrast volume, and white matter skeleton region free water mean value;

[0055] The Mahalanobis distance module is used to construct a Mahalanobis distance model and determine the Mahalanobis distance between the multi-dimensional key image feature vector of the patient and the multi-dimensional key image feature vector of the corresponding group in the age-stratified database of the healthy population based on the multi-dimensional key image marker parameter information;

[0056] The quantization module is used to evaluate the degree of lymphatic system damage based on the Mahalanobis distance as the lymphatic damage index and complete the quantization of the lymphatic system damage.

[0057] The present invention also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the lymphatic system damage quantization method based on multi-modal image markers is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic flowchart of the lymphatic system damage quantization method based on multi-modal image markers according to an embodiment of the present invention;

[0059] Figure 2 It is a schematic structural diagram of the lymphatic system damage quantification system based on multi-modal image markers according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0060] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0061] As Figure 1 shown, a lymphatic system damage quantization method based on multi-modal image markers includes the following steps:

[0062] S1: Collect multi-modal image data information and perform standardized preprocessing;

[0063] S2: Determine multi-dimensional key image marker parameter information based on the multi-modal image data information after standardized preprocessing; wherein, the multi-dimensional key image marker parameter information includes choroid plexus volume fraction, ventricular cerebrospinal fluid volume fraction, enhanced perivascular space contrast volume, and mean free water in the white matter skeleton region;

[0064] S3: Construct a Mahalanobis distance model and determine the Mahalanobis distance between the multi-dimensional key image feature vector of the patient and the multi-dimensional key image feature vector of the corresponding group in the age-stratified database of the healthy population based on the multi-dimensional key image marker parameter information;

[0065] S4: Evaluate the degree of lymphatic system damage based on the Mahalanobis distance as the lymphatic damage index and complete the quantization of the lymphatic system damage.

[0066] The method for quantifying lymphatic system damage based on multi-modal imaging markers of the present invention determines multi-dimensional key imaging marker information from the multi-modal imaging data information after standardized preprocessing, and then combines the Mahalanobis distance model to determine the Mahalanobis distance between the multi-dimensional key imaging feature vectors of patients of different age groups and the multi-dimensional key imaging feature vectors of the corresponding groups in the healthy population age stratification database, so as to achieve an accurate assessment of the degree of lymphatic system damage. By non-linearly integrating multi-dimensional imaging markers, the one-sidedness problem of single-index analysis is solved, and the lymphatic system damage is accurately quantified by the pathological deviation distance. The specificity of this method for detecting lymphatic dysfunction in stroke patients is greatly improved compared with the single ALPS index.

[0067] In one or more embodiments of the present invention, the acquisition of multi-modal imaging data information and the standardized preprocessing specifically include the following steps:

[0068] S11: Obtain multi-modal imaging data information, and perform unified resampling processing according to a preset resolution by using the linear interpolation method;

[0069] Here, in the embodiments of the present invention, the multi-modal imaging data includes T1-weighted images, T2-weighted images, and diffusion tensor imaging (DTI) data.

[0070] Specifically, by acquiring 3D T1-weighted images, a 3T magnetic field strength is used, and the parameters include the time interval TR = 2000 ms between two adjacent radio frequency pulses in the pulse sequence, the echo time TE = 2.5 ms, the flip angle is 9°, and the resolution is 1 mm. By acquiring 3D T2-weighted images, the parameters include the time interval TR = 3000 ms between two adjacent radio frequency pulses in the pulse sequence, the echo time TE = 300 ms, and the resolution is 1 mm. By acquiring diffusion tensor imaging (DTI) data, 64 diffusion directions are used, the b value = 1000 s / mm2, and the resolution is 2 mm. After the acquisition is completed, all images are resampled to a resolution of 1 mm3 by using the linear interpolation method.

[0071] S12: Use ANTs (Advanced Normalization Tools) software to perform N4 bias field correction processing on the multi-modal imaging data information after resampling processing.

[0072] By performing resampling processing on the multi-modal imaging data information, the balance of the data information can be improved, the recognition ability of the model can be improved, and by performing N4 bias field correction processing, the uniformity of the imaging data information can be improved, and the quality and reliability of the imaging data information can be improved.

[0073] In one or more embodiments of the present invention, determining the choroid plexus volume fraction based on the multi-modal image data information after standardized preprocessing specifically includes the following steps:

[0074] S21a: Use FreeSurfer software to segment the multi-modal image data information after standardized preprocessing to obtain the lateral ventricle region;

[0075] S22a: Use the Gaussian mixture model (GMM) to automatically segment within the lateral ventricle region to obtain the choroid plexus;

[0076] S23a: Calculate the choroid plexus volume fraction CPVF (Choroid Plexus Volume Fraction) based on the volume of the choroid plexus. The calculation formula is as follows:

[0077]

[0078] where V ICV is the total intracranial volume, and V choroid is the volume of the choroid plexus.

[0079] The multi-modal image data information is segmented by FreeSurfer software to accurately obtain the lateral ventricle region, and then the Gaussian mixture model is used to automatically segment within the lateral ventricle region to obtain the choroid plexus. Finally, the choroid plexus volume fraction CPVF is calculated based on the volume of the choroid plexus. As one of the multi-dimensional key image marker parameter information, it is used to evaluate the degree of glymphatic system damage.

[0080] In one or more embodiments of the present invention, determining the ventricle cerebrospinal fluid volume fraction based on the multi-modal image data information after standardized preprocessing specifically includes the following steps:

[0081] S21b: Use the multi-atlas registration method in ANTs software to segment the multi-modal image data information to obtain the lateral ventricle region, the third ventricle, and the fourth ventricle;

[0082] S22b: Calculate the ventricle cerebrospinal fluid volume fraction VCSFVF (Ventricle Cerebrospinal Fluid Volume Fraction) based on the volumes of the lateral ventricle region, the third ventricle, and the fourth ventricle. The calculation formula is:

[0083]

[0084] where V ICV is the total intracranial volume, and V CSF is the total cerebrospinal fluid volume of the ventricle region, including the cerebrospinal fluid volumes of the lateral ventricle, the third ventricle, and the fourth ventricle.

[0085] The multi-modal image data information is segmented by the multi-atlas registration method to obtain the lateral ventricle region containing cerebrospinal fluid, the third ventricle and the fourth ventricle, and then the cerebroventricular cerebrospinal fluid volume fraction is accurately calculated, which is used as one of the multi-dimensional key image marker parameter information to evaluate the degree of glymphatic system damage.

[0086] In one or more embodiments of the present invention, determining the enhanced perivascular space contrast volume based on the standardized preprocessed multi-modal image data information specifically includes the following steps:

[0087] S21c: Calculate the perivascular space contrast EPC (Enhanced PVS Contrast) based on the T1-weighted image and the T2-weighted image in the standardized preprocessed multi-modal image data;

[0088]

[0089] Among them, T1 signal is the T1-weighted image, T2 signal is the T2-weighted image, and EPC is the perivascular space contrast;

[0090] S22c: Perform Frangi filtering (multi-scale vessel enhancement filtering based on the Hessian matrix) enhancement processing on the perivascular space contrast EPC to obtain the enhanced perivascular space PVS (Perivascular Space) volume V PVS , and the calculation formula is:

[0091]

[0092] Among them, BG is the basal ganglia region, DWM is the deep white matter region, x ∈ BG ∪ DWM represents the set of voxels belonging to the basal ganglia region and the deep white matter region, and H(·) is the Hessian matrix eigenvalue constraint.

[0093] The perivascular space contrast EPC is generated through the T1-weighted image and the T2-weighted image, thereby enhancing the visibility of the perivascular space. Then, through Frangi filtering enhancement processing, unstructured high-frequency spatial noise can be eliminated, and the resolution of the perivascular space can be greatly improved.

[0094] In one or more embodiments of the present invention, determining the free water mean value of the white matter skeleton region based on the standardized preprocessed multi-modal image data information specifically includes the following steps:

[0095] S21d: Calculate the free water fraction FW using the dipy double tensor model based on the standardized preprocessed multi-modal image data information;

[0096] Here, the calculation process of the free water fraction is prior art and will not be elaborated in detail in the embodiments of the present invention.

[0097] S22d: Perform diffusion statistical analysis based on the multi-modal image data information after standardized preprocessing, and extract the white matter skeleton region;

[0098] In the embodiments of the present invention, the FSL6.0 software is used to directly perform diffusion statistical analysis of TBSS (Tract-Based Spatial Statistics) to extract the white matter skeleton region.

[0099] S23d: Calculate the mean free water of the white matter skeleton region according to the free water fraction FW.

[0100] The mean free water of the white matter skeleton region reflects the dynamic exchange ability of the interstitial fluid in the brain tissue and is closely related to the glymphatic system clearance function. The dipy double tensor model is used to accurately calculate the free water fraction, and at the same time, the white matter skeleton region is extracted through diffusion statistical analysis, so that the mean free water of the white matter skeleton region can be accurately calculated, which is used as one of the multi-dimensional key image biomarker parameter information to evaluate the degree of damage to the glymphatic system.

[0101] In one or more embodiments of the present invention, the construction of the Mahalanobis distance model and the determination of the Mahalanobis distance between the multi-dimensional key image feature vector of the patient and the corresponding group of multi-dimensional key image feature vectors in the healthy population age stratification database based on the multi-dimensional key image biomarker parameter information specifically include the following steps:

[0102] S31: Construct a Mahalanobis distance model, group the multi-dimensional key image biomarker parameter information of the healthy population according to a preset age interval, and establish a healthy population age stratification database;

[0103] In the embodiments of the present invention, 1000 individuals without a history of neurological diseases are selected as the healthy population sample, with an age range of 20 - 75 years old, grouped at 5-year intervals, such as 20 - 25 years old, 25 - 30 years old, etc., with at least 100 people in each group, and a healthy population age stratification database is established.

[0104] S32: Calculate the multi-dimensional key image feature vectors of each group of data of the healthy population according to the healthy population age stratification database. The multi-dimensional key image feature vectors include the mean vector μ and the covariance matrix Σ; the specific calculation process is prior art and will not be elaborated in detail here.

[0105] Here, according to the above steps, the mean vector μ and the covariance matrix Σ (including CPVF, VCSFVF, PVS, MSFW) of the multi-modal (four-dimensional) image data information are calculated for each age group.

[0106] It should be noted that, in order to improve the accuracy of the calculation results, in the embodiments of the present invention, after calculating the mean vector μ and covariance matrix Σ of the multi-modal (four-dimensional) key image feature vectors, the Kolmogorov-Smirnov test is also used to test the multi-modal (four-dimensional) key image feature vectors to ensure that the data is normally distributed within each age group, and the detected outliers are processed by the Grubbs test method.

[0107] S33: Determine the multi-dimensional key image feature vectors of the patient according to the multi-dimensional key image marker parameter information of the patient, and calculate the Mahalanobis distance D between the multi-dimensional key image feature vectors of the patient and those of the healthy population and the patient. M , and the calculation formula is:

[0108]

[0109] where X is the multi-dimensional key image feature vector of the patient, μ is the mean vector of the multi-dimensional key image marker parameter information of each group in the healthy population age-stratified database, and Σ is the covariance matrix of the multi-dimensional key image marker parameter information of each group in the healthy population age-stratified database.

[0110] Here, for the patient, according to the multi-modal image data information of the patient, the multi-dimensional key image feature vector of the patient can be obtained by using the method of the previous steps, which will not be elaborated here.

[0111] Group by the multi-dimensional key image marker parameter information of the healthy population, so as to obtain the multi-dimensional key image marker parameter information of the healthy population in different age groups, and then determine the Mahalanobis distance between the multi-dimensional key image feature vector of the patient and the multi-dimensional normal distribution parameters of the healthy population and the patient according to the multi-dimensional normal distribution parameters, that is, accurately quantify the lymphatic system damage through the pathological deviation distance.

[0112] In one or more embodiments of the present invention, the specific steps of evaluating the degree of lymphatic system damage based on the Mahalanobis distance as the lymphatic damage index are as follows:

[0113] S41: Perform ROC curve analysis according to the lymphatic damage index of the patient, and determine the lymphatic damage index threshold of the corresponding age group according to the Youden index method.

[0114] S42: Classify the lymphatic damage index of the patient according to the lymphatic damage index threshold.

[0115] Here, according to the thresholds of the lymphatic damage index for different age groups, the degree of lymphatic damage in patients can be classified into no damage, mild damage, moderate damage, severe damage, etc. The thresholds of the lymphatic damage index for different age groups are different.

[0116] By performing the ROC curve analysis method on the lymphatic damage index, the threshold of the lymphatic damage index for the corresponding age group can be obtained, so as to accurately classify the lymphatic damage index of patients.

[0117] Preferably, after the assessment of the degree of damage to the lymphatic system of patients is completed, the present invention further includes the following steps:

[0118] S5: Verify the assessment results through a preset independent test dataset (such as the multi-modal image data information of 200 patients), and calculate the detection specificity.

[0119] Specifically, in the embodiments of the present invention, the multi-modal image data information of 200 patients with different degrees of damage is selected as an independent test set for verification, and the detection specificity is calculated to determine the accuracy of the assessment results of the present invention.

[0120] In the embodiments of the present invention, according to the clinical trial results, the detection specificity of the method for quantifying lymphatic system damage based on multi-modal imaging markers of the present invention for detecting dysfunction caused by lymphatic system damage in stroke patients reaches 92%, which is 17% higher than that of the single ALPS index, greatly improving the accuracy of the assessment of lymphatic system damage.

[0121] As Figure 2 shown, the present invention also provides a system for quantifying lymphatic system damage based on multi-modal imaging markers, including an acquisition and preprocessing module, a multi-dimensional parameter determination module, a Mahalanobis distance module, and a quantification module;

[0122] The acquisition and preprocessing module is used to acquire multi-modal image data information and perform standardized preprocessing;

[0123] The multi-dimensional parameter determination module determines multi-dimensional key imaging marker parameter information based on the standardized preprocessed multi-modal image data information; wherein, the multi-dimensional key imaging marker parameter information includes the choroid plexus volume fraction, the ventricular cerebrospinal fluid volume fraction, the enhanced perivascular space contrast volume, and the mean free water in the white matter skeleton region;

[0124] The Mahalanobis distance module is used to construct a Mahalanobis distance model and determine between the multi-dimensional key imaging marker feature vectors of healthy people and patients based on the multi-dimensional key imaging marker parameter information;

[0125] The quantization module is used to determine a classification comparison table for lymphatic system damage by taking the Mahalanobis distance as the lymphatic damage index, and complete the quantization of lymphatic system damage.

[0126] The present invention also provides a computer-readable storage medium storing a computer program, characterized in that when the computer program is executed by a processor, the method for quantifying lymphatic system damage based on multimodal imaging markers as described above is implemented.

[0127] The method and system for quantifying lymphatic system damage based on multimodal imaging markers of the present invention integrate multi-dimensional markers such as ventricle cerebrospinal fluid volume fraction (VCSFVF), perivascular space volume (PVS), choroid plexus volume fraction (CPVF), and mean skeletonized free water (MSFW). Through the integration of multi-dimensional markers, the input (VCSFVF), transportation (PVS), and output (MSFW) links of the lymphatic system can be evaluated simultaneously. A multi-dimensional physiological reference space is constructed based on the Mahalanobis distance, and the marker correlation and age distribution are automatically calculated to eliminate the collinearity interference, so as to calculate the pathological deviation distance, thereby realizing the precise quantization of lymphatic system damage and outputting a single quantization index, which is applicable to cross-group comparison of patients of different ages.

[0128] The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for quantifying lymphatic system damage based on multimodal imaging markers, characterized in that, It includes the following steps: Collect multi-modal image data information and perform standardized preprocessing; Determine multi-dimensional key image marker parameter information based on the multi-modal image data information after standardized preprocessing; wherein, the multi-dimensional key image marker parameter information includes choroid plexus volume fraction, ventricular cerebrospinal fluid volume fraction, enhanced perivascular space contrast volume, and mean free water in the white matter skeleton region; Construct a Mahalanobis distance model and determine the Mahalanobis distance between the multi-dimensional key image feature vector of the patient and the multi-dimensional key image feature vector of the corresponding group in the healthy population age-stratified database based on the multi-dimensional key image marker parameter information; Based on the Mahalanobis distance as the glymphatic damage index, evaluate the degree of glymphatic system damage and complete the quantification of glymphatic system damage.

2. The method for quantifying lymphatic system damage based on multimodal imaging markers according to claim 1, characterized in that, The step of collecting multi-modal image data information and performing standardized preprocessing specifically includes the following steps: Obtain multi-modal image data information and perform unified resampling processing according to a preset resolution using linear interpolation; Use ANTs software to perform N4 bias field correction processing on the multi-modal image data information after resampling processing.

3. The method for quantifying lymphatic system damage based on multimodal imaging markers according to claim 1, wherein The step of determining the choroid plexus volume fraction based on the multi-modal image data information after standardized preprocessing specifically includes the following steps: Use FreeSurfer software to segment the multi-modal image data information after standardized preprocessing to obtain the lateral ventricle region; Use a Gaussian mixture model to automatically segment within the lateral ventricle region to obtain the choroid plexus; Calculate the choroid plexus volume fraction CPVF based on the volume of the choroid plexus. The calculation formula is as follows: Among them, V ICV is the total intracranial volume, and V choroid is the choroid plexus volume.

4. The method for quantifying lymphatic system damage based on multimodal imaging markers according to claim 1, wherein The step of determining the ventricular cerebrospinal fluid volume fraction based on the multi-modal image data information after standardized preprocessing specifically includes the following steps: Use a multi-atlas registration method to segment the multi-modal image data information to obtain the lateral ventricle region, the third ventricle, and the fourth ventricle; Calculate the ventricular cerebrospinal fluid volume fraction VCSFVF based on the volumes of the lateral ventricle region, the third ventricle, and the fourth ventricle. The calculation formula is: Among them, V ICV is the total intracranial volume, and V CSF is the total cerebrospinal fluid volume in the ventricular region, including the cerebrospinal fluid volumes of the lateral ventricles, the third ventricle, and the fourth ventricle.

5. The method for quantifying lymphatic system damage based on multimodal imaging markers according to claim 1, wherein The step of determining the enhanced perivascular space contrast volume based on the multi-modal image data information after standardized preprocessing specifically includes the following steps: Calculate the perivascular space contrast EPC based on the T1-weighted image and the T2-weighted image in the multi-modal image data after standardized preprocessing; Among them, T1 signal is a T1-weighted image, T2 signal is a T2-weighted image, and EPC is the perivascular space contrast; Perform Frangi filtering enhancement processing on the perivascular space contrast EPC to obtain the enhanced perivascular space PVS volume V PVS , and the calculation formula is: Wherein, BG is the basal ganglia region, DWM is the deep white matter region, x∈BG∪DWM represents the set of voxels belonging to the basal ganglia region and the deep white matter region, and H(·) is the Hessian matrix eigenvalue constraint.

6. The method for quantifying lymphatic system damage based on multimodal imaging markers according to claim 1, wherein The step of determining the mean free water in the white matter skeleton region based on the multi-modal image data information after standardized preprocessing specifically includes the following steps: Based on the multi-modal image data information after standardized preprocessing, use the dipy double tensor model to calculate the free water fraction FW; Perform diffusion statistical analysis based on the multi-modal image data information after standardized preprocessing and extract the white matter skeleton region; According to the free water fraction FW, calculate the mean free water in the white matter skeleton region.

7. The method for quantifying lymphatic system damage based on multimodal imaging markers according to claim 1, characterized in that: Building the Mahalanobis distance model and determining the Mahalanobis distance between the multi-dimensional key imaging feature vector of the patient and the multi-dimensional key imaging feature vector of the corresponding group in the age-stratified database of the healthy population based on the multi-dimensional key imaging marker parameter information specifically includes the following steps: Build the Mahalanobis distance model, group the multi-dimensional key imaging marker parameter information of the healthy population according to a preset age interval, and establish an age-stratified database of the healthy population; Calculate the multi-dimensional key imaging feature vector of each group of data of the healthy population according to the age-stratified database of the healthy population, and the multi-dimensional key imaging feature vector includes the mean vector μ and the covariance matrix Σ; Determine the multi-dimensional key image feature vector of the patient according to the multi-dimensional key image marker parameter information of the patient, and calculate the Mahalanobis distance D between the multi-dimensional key image feature vector of the patient and that of the healthy population and the patient M , and the calculation formula is: Among them, X is the multi-dimensional key imaging feature vector of the patient, μ is the mean vector of the multi-dimensional key imaging marker parameter information of each group in the age-stratified database of the healthy population, and Σ is the covariance matrix of the multi-dimensional key imaging marker parameter information of each group in the age-stratified database of the healthy population.

8. The method for quantifying lymphatic system damage based on multimodal imaging markers according to claim 7, wherein: Evaluating the degree of lymphatic system damage based on the Mahalanobis distance as the lymphatic damage index specifically includes the following steps: Perform the ROC curve analysis method according to the lymphatic damage index of the patient, and determine the lymphatic damage index threshold of the corresponding age group according to the Youden index method; Classify the lymphatic damage index of the patient according to the lymphatic damage index threshold.

9. A lymphatic system damage quantification system based on multimodal imaging markers, characterized in that Including an acquisition preprocessing module, a multi-dimensional parameter determination module, a Mahalanobis distance module, and a quantification module; The acquisition preprocessing module is used to acquire multi-modal imaging data information and perform standardized preprocessing; The multi-dimensional parameter determination module determines multi-dimensional key imaging marker parameter information based on the standardized preprocessed multi-modal imaging data information; wherein, the multi-dimensional key imaging marker parameter information includes the choroid plexus volume fraction, the ventricular cerebrospinal fluid volume fraction, the enhanced perivascular space contrast volume, and the mean free water in the white matter skeleton region; The Mahalanobis distance module is used to build the Mahalanobis distance model and determine the Mahalanobis distance between the multi-dimensional key imaging feature vector of the patient and the multi-dimensional key imaging feature vector of the corresponding group in the age-stratified database of the healthy population based on the multi-dimensional key imaging marker parameter information; The quantification module is used to evaluate the degree of lymphatic system damage based on the Mahalanobis distance as the lymphatic damage index and complete the quantification of the lymphatic system damage.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the lymphatic system damage quantification method based on multi-modal imaging markers according to any one of claims 1-8.

Images