PD and MSA identification system based on lymphatic-like system function and free water imaging

Abstract

The invention discloses a PD and MSA identification system based on a lymphatic-like system function and free water imaging, and the system is characterized in that the system comprises an image data collection module which is used for obtaining high-quality and multi-contrast original MRI data; the preprocessing module is used for preprocessing the data acquired by the image data acquisition module; the Glymptic function evaluation module is used for calculating a DTI-ALPS index so as to evaluate a brain lymphatic system function; the free water imaging and analysis module is used for quantitatively analyzing whole-brain free water distribution and dynamic change thereof; and the statistical modeling and diagnosis module is used for integrating the DTI-ALPS index and the free water distribution index to construct an identification model. According to the invention, differentiated pathophysiological mechanisms are disclosed for the first time, a solid scientific basis is provided for differential diagnosis, unprecedented high-precision and high-reliability differential diagnosis is realized, a brand-new, quantifiable and repeatable imaging biomarker combination is provided, and the application prospect is wide. The system has powerful disease progress monitoring and potential curative effect evaluation functions.

Description

Technical Field

[0001] This invention relates to the field of medical information, and more particularly to a PD and MSA identification system based on lymphoid system function and free water imaging. Background Technology

[0002] Parkinson's disease (PD) and multiple system atrophy (MSA) both belong to the synucleinopathies, a group of neurodegenerative diseases characterized by the abnormal accumulation of α-synuclein in the brain. Although their clinical manifestations overlap significantly, such as bradykinesia, rigidity, and autonomic dysfunction, their clinical courses, treatment responses, and underlying pathophysiological mechanisms differ fundamentally. This difference makes accurate differential diagnosis between PD and MSA in the early stages of the disease extremely challenging yet crucial. Currently, there is an urgent clinical need to develop sensitive and specific novel biomarkers to differentiate between these two diseases, guide individualized treatment strategies, improve the accuracy of prognostic assessments, and ultimately support the development of targeted therapies.

[0003] In recent years, the role of the glymphatic system in neurodegenerative diseases has received increasing attention as a “cleaning” system of the brain. This glymphatic system relies on the directed return of cerebrospinal fluid along the perivascular spaces (PVS) and plays a central role in clearing metabolic waste products such as α-synuclein through aquaporin-4 (AQP4) on the terminal feet of astrocytes. However, it is noteworthy that astrocytes, the cornerstone of glymphatic function, exhibit drastically different pathological changes in PD and MSA: in PD, astrocytes tend to present a neurotoxic phenotype, while in MSA, they primarily exhibit reactive gliosis. These differential astrocyte responses suggest that glymphatic system dysfunction may manifest in different ways and through different mechanisms in these two synucleinopathies. However, current evidence is insufficient to fully capture the heterogeneous glymphatic system alterations across synucleinopathies, thus necessitating multimodal research approaches to elucidate disease-specific pathophysiological divergences.

[0004] In imaging techniques, the diffusion tensor imaging perivascular spatial analysis (DTI-ALPS) index has been used to assess lymphoid system function. This index quantifies lymphoid activity by calculating the difference in water diffusion along the periventricular white matter fiber bundles in the direction of the medullary vein (PVS) versus the vertical direction. Studies have shown that the DTI-ALPS index is significantly reduced in patients with both PD and MSA. On the other hand, the free water (FW) fraction derived from the neuronal synaptic dispersion and density imaging (NODDI) model can indirectly reflect the static free water distribution in the PVS and cerebrospinal fluid system. FW imaging has become a promising biomarker for diagnosis and disease progression monitoring in PD and MSA, with elevated FW values ​​found in multiple brain regions.

[0005] Although the aforementioned studies have revealed the importance of lymphoid dysfunction and fat accumulation (FW) in synucleinopathies, current technology has significant limitations: First, most existing studies examine lymphoid function (DTI-ALPS) or FW changes in isolation, failing to organically combine the two to explore their intrinsic relationship; second, there is a lack of direct evidence to prove that the lymphoid system regulates FW distribution and its potential contribution to synucleinopathies; most importantly, no study has systematically elucidated whether there are fundamental differences in the pathological coupling relationship between lymphoid dysfunction and synucleinopathies (PD) and synucleinopathic atrophy (MSA), and this difference may be the core mechanism leading to their different clinical manifestations.

[0006] Therefore, there is an urgent need in this field for a novel comprehensive imaging method that can integrate lymphoid system function assessment and free water pathological distribution analysis to reveal the pathological mechanisms of PD and MSA differences, and ultimately transform them into a clinical tool that can achieve early and accurate differential diagnosis. Summary of the Invention

[0007] The purpose of this invention is to provide a PD and MSA differentiation system based on glymphatic system function and free water imaging. By quantifying the DTI-ALPS index, which reflects the function of the brain's glymphatic system, and an index reflecting the distribution of free water (FW) in the brain, the invention extracts disease-specific patterns in the spatial distribution and correlation between the two, and constructs a fusion model to achieve high-precision automatic differential diagnosis, thereby solving the problems mentioned in the background art.

[0008] To achieve the above-mentioned objectives, this invention provides a PD and MSA differentiation system based on lymphatic system function and free water imaging. The system for differentiation includes an image data acquisition module, a preprocessing module, a Glymphatic functional assessment module, a free water imaging and analysis module, and a statistical modeling and diagnostic module, wherein:

[0009] The image data acquisition module is used to acquire high-quality, multi-contrast raw MRI data;

[0010] The preprocessing module is used to preprocess the data acquired by the image data acquisition module;

[0011] The Glymphatic functional assessment module is used to calculate the DTI-ALPS index to assess the function of the brain's lymphoid system.

[0012] The free water imaging and analysis module is used to quantitatively analyze the distribution and dynamic changes of free water in the whole brain;

[0013] The statistical modeling and diagnostic module is used to integrate the DTI-ALPS index and free water distribution index to build a diagnostic classification model.

[0014] Furthermore, the image data acquisition module uses a high-field-strength MRI scanner to simultaneously acquire the following sequences:

[0015] High-resolution T1-weighted structural images were used for anatomical localization of brain regions, tissue segmentation, and subsequent image registration. The magnetization preparation of the fast gradient echo (MPRAGE) sequence was performed using isotropic resolution (e.g., 1.0 mm³).

[0016] DTI (Diffusion-weighted image) is used to calculate the DTI-ALPS index and FW map. It is acquired using a dual-shell protocol and includes at least a b=0 s / mm² (unweighted) image and a b=1000 s / mm² image along 30 or more diffusion-encoded directions. It also uses alternating acquisition of front and back phase-encoded directions to eliminate magnetic susceptibility artifacts.

[0017] QSW (Quantitative Magnetic Sensitivity Mapping) is used to clearly display the structure of the medullary vein. It is acquired using a multi-echo 3D gradient echo sequence and retains complex data for offline magnetic susceptibility reconstruction.

[0018] Furthermore, the preprocessing includes DTI preprocessing and spatial normalization, wherein:

[0019] DTI preprocessing involves performing head motion and eddy current corrections on the raw DTI data and removing non-brain tissue.

[0020] Spatial standardization normalizes each subject's T1 image and DTI-derived image to a standard brain space using linear and nonlinear registration algorithms, enabling group-level voxel-level comparative analysis.

[0021] Furthermore, the Glymphatic functional assessment module calculates the DTI-ALPS index using the following steps:

[0022] Step S001: On an axial slice at the level of the lateral ventricle body in standard MNI space, define the spherical regions covering projection fibers and association fibers in the white matter regions of both hemispheres, respectively.

[0023] Step S002: By registering the individual's T1 image and B0 image to the MNI template respectively and calculating its inverse deformation field, the above standard ROI template is accurately mapped inversely to the native DTI space of each subject individual.

[0024] Step S003: Overlay the mapped ROI onto the QSM image for visual verification to ensure that the ROI covers at least one linear low-signal medullary vein, so as to ensure the anatomical accuracy of the ROI.

[0025] Step S004: Within the ROI of the projected fibers and combined fibers of each individual, calculate the average diffusion coefficient along the x, y, and z directions respectively;

[0026] Step S005: Calculate the DTI-ALPS index using the average diffusion coefficients of the three directions from step S004, and take the average value of the ROI results for both hemispheres.

[0027] Step S006: The calculated DTI-ALPS index is harmonized using the ComBat algorithm to eliminate the influence of multicenter, multi-scan, and demographic confounding factors.

[0028] Furthermore, the formula for calculating the DTI-ALPS index is as follows:

[0029] DTI-ALPS index = mean(Dxx_proj, Dxx_assoc) / mean(Dyy_proj, Dzz_assoc),

[0030] Wherein, Dxx_proj is the x-axis diffusion coefficient extracted from the projected fiber region, Dxx_assoc is the x-axis diffusion coefficient extracted from the combined fiber region, Dyy_proj is the y-axis diffusion coefficient extracted from the projected fiber region, and Dzz_assoc is the z-axis diffusion coefficient extracted from the combined fiber region.

[0031] Furthermore, the calculation process of the free water imaging and analysis module includes the following steps:

[0032] Step S101: Based on multi-shell diffusion data, using the Neuron Direction Dispersion and Density Imaging (NODDI) model, and through the Convex Optimization Accelerated Microstructure Imaging (AMICO) algorithm, calculate the free water fraction (FW fraction) of each voxel to generate a whole-brain FW map.

[0033] Step S102: Use FSL-FAST to segment the T1 image into gray matter (GM), white matter (WM) and cerebrospinal fluid (CSF) probability maps, and use a linear regression model to correct the influence of CSF partial volume effect on FW value to generate the corrected FW map.

[0034] Step S103: Perform whole-brain voxel-level analysis and comparison on the corrected FW map;

[0035] Step S104: Define the significant clusters in the above analysis and comparison as disease-specific ROIs (such as PD-specific ROI1, MSA-specific ROI2), and extract the average FW value of the disease-specific clusters for subsequent correlation and diagnostic analysis.

[0036] Step S105: The ComBat algorithm is used to coordinate the average FW value of the extracted ROI in order to eliminate the influence of confounding factors.

[0037] Furthermore, the whole-brain voxel-level analysis comparison was divided into inter-group comparisons, among which:

[0038] The intergroup comparisons were performed by standardizing the corrected FW plots to the MNI space and smoothing them, and then using an unpaired t-test to perform voxel-level comparisons of PD vs HC, MSA vs HC, and PD vs MSA. Multiple comparison correction was performed using the false detection rate (FDR) (e.g., P < 0.01, cluster size > 200 voxels).

[0039] Furthermore, the statistical modeling and diagnostic module uses the DTI-ALPS index, the average FW value of ROI1, and the average FW value of ROI2 as independent variables, and the diagnostic group (e.g., PD / MSA / HC) as the dependent variable. It constructs a logistic regression model as the diagnostic classification model, incorporating demographic covariates (age, gender, years of education) to correct for their influence, and outputs the predicted probability of the diagnostic category. This diagnostic classification model uses clinical diagnosis as the gold standard, applies SPSS software, and uses a logistic regression classification model to output individual probability values, expressed by the following formula:

[0040] Logit(P) = ln[ P(MSA) / (1 - P(MSA)) ] = 28.556 - 123.331×ROI1 +75.792×ROI2 - 14.363×ALPS - 0.033×age + 0.133×gender,

[0041] Where: P(MSA) represents the probability that the individual is diagnosed with MSA; 1 - P(MSA) represents the probability that the individual is diagnosed with PD; ALPS is the DTI-ALPS index; ROI1 is the average FW value of ROI1; ROI2 is the average FW value of ROI2; age is age; and gender is sex.

[0042] Compared with existing technologies, this system and method have the following advantages:

[0043] 1. This invention systematically elucidates and verifies for the first time the fundamental differences between PD and MSA in the pathological coupling relationship of the "lymphoid system-free water". In PD, a strong spatial coupling relationship was found: glymphatic dysfunction (decreased DTI-ALPS index) and FW accumulation are highly co-localized in the cortical-subcortical region, and FW plays a mediating role between glymphatic dysfunction and clinical symptoms. This proves that one of the core pathologies of PD is fluid accumulation caused by the failure of clearance function, providing direct imaging evidence for the use of "enhanced clearance" treatment strategies. In MSA, a mechanism decoupling relationship was found: although glymphatic function also declines, its characteristic cerebellar FW accumulation is not significantly associated with DTI-ALPS changes. This indicates that FW accumulation in MSA mainly originates from primary mechanisms independent of lymphoid clearance, such as oligodendrocyte lesions, neuroinflammation, or myelin degradation. This discovery transcends the traditional level of differential diagnosis based solely on morphological differences, elevating the diagnostic basis to the level of pathological mechanisms, making the differential diagnosis results more interpretable and reliable.

[0044] 2. By fusing multimodal imaging biomarkers, this invention achieves extremely high differentiation accuracy between PD and MSA, significantly outperforming current clinical MRI diagnostic standards. The fusion model constructed in this invention (combining the DTI-ALPS index, PD-FW, and MSA-FW) exhibits an area under the curve (AUC) of 0.982 when distinguishing between PD and MSA, achieving near-perfect differentiation. This performance addresses the clinical challenge of high specificity but extremely low sensitivity (especially in the early stages) of MRI biomarkers in the current Movement Disorder Society International (MDS) MSA diagnostic criteria, providing a powerful tool for the diagnosis of early and complex cases, potentially reducing misdiagnosis and avoiding ineffective treatment.

[0045] 3. This invention provides a novel, quantifiable, and reproducible combination of imaging biomarkers. Based entirely on objective imaging computational indicators (DTI-ALPS index, FW value), this invention eliminates the reliance on subjective experience (such as visual assessment of the "cross sign" and "hummingbird sign") in traditional MRI diagnosis, significantly reducing inter-diagnostic variability and improving the consistency and reproducibility of results. All data are acquired from routine MRI sequences, requiring no contrast agent injection or lumbar puncture, making it safe, non-invasive, highly acceptable to patients, and easy to promote and popularize in clinical practice. From ROI mapping and QSM validation to feature extraction and model calculation, this invention emphasizes automated or semi-automated processing, minimizing manual operation steps and subjective bias, ensuring the stability and scalability of the technical solution.

[0046] 4. This invention is not only applicable to cross-sectional differential diagnosis, but its longitudinal analysis capability also gives it a deeper value in disease management. It can quantitatively monitor the evolution trajectory of glymphatic function (DTI-ALPS index) and FW accumulation over time, providing a sensitive tool for assessing the natural progression rate of disease. Attached Figure Description

[0047] Figure 1 This is a structural diagram of a PD and MSA discrimination system based on lymphoid system function and free water imaging.

[0048] Figure 2 A graph comparing DTI-ALPS index and FW in healthy controls and patients with synucleinosis. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] This invention provides a system for the differential diagnosis of Parkinson's disease (PD) and multiple system atrophy (MSA) based on multimodal magnetic resonance imaging (MRI). This system comprehensively assesses the DTI-ALPS index, which reflects the function of the brain's glymphatic system, and quantitative indicators reflecting the distribution of free water (FW) in the brain. It extracts disease-specific features of both in terms of spatial patterns, dynamic changes, and coupling relationships, and constructs a high-precision machine learning classification model, ultimately achieving automated and objective differentiation between PD and MSA. The system is used to automate the process of this method.

[0051] like Figure 1The diagram shown is the system architecture of this system. This system includes an image data acquisition module, a preprocessing module, a Glymphatic functional evaluation module, a free water imaging and analysis module, and a statistical modeling and diagnosis module.

[0052] The image data acquisition module is responsible for acquiring high-quality, multi-contrast raw MRI data and preparing it for subsequent analysis. Using a high-field (≥3T) MRI scanner, the following sequences are acquired simultaneously:

[0053] High-resolution T1-weighted structural images: used for anatomical localization of brain regions, tissue segmentation, and subsequent image registration. Fast gradient echo (MPRAGE) sequences are prepared using magnetization at isotropic resolution (e.g., 1.0 mm³).

[0054] Diffusion-weighted imaging (DTI): Used to calculate the DTI-ALPS index and free water (FW) map. Dual-shell acquisition is employed, including at least b=0 s / mm² (acquiring 3 volumes) and b=1000 s / mm² (acquiring along 30 or more directions). Alternating acquisition directions with forward and backward phase encoding are used to eliminate magnetic susceptibility artifacts.

[0055] Quantitative magnetic susceptibility imaging (QSM): used to clearly visualize the medullary vein structure, providing anatomical validation for the ROI localization in DTI-ALPS. A multi-echo 3D gradient echo sequence was employed.

[0056] The preprocessing module is used to preprocess and normalize images, including:

[0057] DTI preprocessing: Head motion and eddy current corrections are performed on the raw DTI data, and non-brain tissue is removed.

[0058] Spatial standardization: The T1 images and DTI-derived images (such as FA images and FW images) of each subject are standardized to a standard brain space (such as MNI152 space) through linear and nonlinear registration algorithms to achieve voxel-level comparative analysis at the population level.

[0059] The Glymphatic Functional Assessment module is used to automatically and objectively calculate the DTI-ALPS index to assess the function of the brain's lymphoid system. This module employs an automated and repeatable calculation process to objectively quantify lymphoid system function. Its core lies in standardized ROI definition and mapping, as well as QSM validation, specifically including the following steps:

[0060] Step S001, Standard ROI Template Definition. On an axial section at the level of the lateral ventricle body in standard MNI space, a spherical ROI template is predefined in the areas where projection fibers and association fibers run in both hemispheres. This position ensures that the perivascular space (PVS) direction (mainly x-axis) within the ROI is perpendicular to the main directions (y-axis and z-axis) of the adjacent fiber bundles.

[0061] Step S002, Individual Spatial ROI Mapping. By registering the individual's T1 and B0 images to the MNI template and calculating their inverse deformation fields, the aforementioned standard ROI template is accurately mapped inversely to the native DTI space of each subject individual.

[0062] Step S003, QSM-guided ROI verification. The mapped individual ROI is overlaid with the QSM image, and a visual quality check is performed to ensure that the ROI covers at least one linear low-signal medullary vein, thereby ensuring the anatomical accuracy of the ROI and avoiding deviations introduced by individual differences in vascular structure.

[0063] Step S004, Diffusion Coefficient Extraction and Calculation. Within the validated individual ROI, calculate the average diffusion coefficient in the following directions:

[0064] Projection fiber region: diffusion coefficient in the x-axis direction (Dxx_proj) and diffusion coefficient in the y-axis direction (Dyy_proj).

[0065] Combined fiber region: diffusion coefficient in the x-axis direction (Dxx_assoc) and diffusion coefficient in the z-axis direction (Dzz_assoc).

[0066] DTI-ALPS Index Calculation: The index is calculated using the formula DTI-ALPS index = mean(Dxx_proj, Dxx_assoc) / mean(Dyy_proj, Dzz_assoc), and the average of the ROI results from both hemispheres is taken as the final index for the individual.

[0067] Step S005, Data Harmonization. To eliminate the influence of multicenter, multi-scan, and demographic confounding factors (age, sex), the ComBat algorithm is used to harmonize the final calculated DTI-ALPS index.

[0068] The free water imaging and analysis module is used to quantitatively analyze the distribution and dynamic changes of free water throughout the brain. It includes the following steps:

[0069] Step S101, FW map calculation. Based on DTI data, using the Neuronal Dispersion and Density Imaging (NODDI) model, the free water fraction (FW fraction) of each voxel is calculated using the Convex Optimization Accelerated Microstructure Imaging (AMICO) algorithm to generate a whole-brain FW map.

[0070] Step S102, partial volume effect correction. The T1 image is segmented into gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) probability maps using FSL-FAST. The partial volume effect of CSF on the FW value of adjacent brain tissue is corrected by a linear regression model to generate a high-precision corrected FW map.

[0071] Step S103, whole-brain voxel-level statistical analysis. Between-group comparisons were used: after standardizing the corrected FW maps of all individuals to the MNI space and smoothing them, two-sample t-tests were used for whole-brain voxel-level comparisons (PD vs HC), (MSA vs HC), and (PD vs MSA). False discovery rate (FDR) was used for multiple comparison correction (e.g., setting a significance level of P < 0.01 and cluster size > 200 voxels). Statistically significant brain regions were identified as disease-specific FW accumulation areas.

[0072] Step S104, ROI Definition and Feature Extraction. The significant clusters obtained from the above statistical comparisons are defined as PD-specific ROIs (e.g., ROI_PD involving the cortex, putamen, and thalamus) and MSA-specific ROIs (e.g., cerebellar ROI_MSA). The average FW value for each subject within these ROIs is extracted as a key input feature for subsequent modeling.

[0073] Step S105, Data Harmonization. The ComBat algorithm is also used to harmonize the extracted average FW value of the ROI to eliminate the influence of confounding factors.

[0074] The statistical modeling and diagnostic module is used to integrate the aforementioned indicators and construct the final diagnostic classification model.

[0075] The process of building the diagnostic classification model is as follows:

[0076] The coordinated DTI-ALPS index, the average FW value of ROI_PD, and the average FW value of ROI_MSA are used as core features;

[0077] Using clinical diagnosis as the gold standard, SPSS software was used. Demographic information such as DTI-ALPS, fetal weight, and gender and age were input. A logistic regression classification model was employed to output individual probability values. Age and gender could be included as covariates in the model. The formula is as follows:

[0078] Logit(P) = ln[ P(MSA) / (1 - P(MSA)) ] = 28.556 - 123.331×ROI1 +75.792×ROI2 - 14.363×ALPS - 0.033×age + 0.133×gender,

[0079] Where: P(MSA): represents the probability that the individual is diagnosed with MSA. 1 - P(MSA): represents the probability that the individual is diagnosed with PD.

[0080] The model outputs the probability that an individual belongs to PD, MSA, or HC. By plotting receiver operating characteristic (ROC) curves and calculating indicators such as the area under the curve (AUC), sensitivity, and specificity, the model's diagnostic efficacy (especially for PD vs. MSA) is comprehensively evaluated. Figure 2 The figure shows a comparison of DTI-ALPS index and FW between healthy controls and patients with synucleinosis. Figure 2 A represents the DTI-ALPS index value for each group (HC, PD, MSA). Figure 2 B-2D is a whole-brain voxel level analysis diagram (P < 0.01 after FDR correction). As can be seen from the figure, Figure 2 Compared with the HC group, FW was higher in the PD group in B; Figure 2 Compared with the HC group, FW was higher in the MSA group in C. Figure 2 Differences in FW distribution between the PD group and the MSA group in D. Figure 2 E is a schematic diagram of disease-specific regions of interest, showing ROI1 (PD-specific: parietal / frontal / temporal / midline cluster) and ROI2 (MSA-specific: cerebellar cluster). Figure 2 F is a bar chart of FW for each group within ROI1. Figure 2 G is a bar chart of the FW values ​​for each group within ROI2. Figure 2 H-2J shows the ROC curves for assessing the diagnostic accuracy of combined ROI and lymphoid function (DTI-ALPS index), as follows: Figure 2 H is HC vs. PD. Figure 2 I is HC vs. MSA. Figure 2 J represents MSA vs. PD. ROC analysis showed that, in distinguishing between Parkinson's disease (PD) and healthy controls (HC), the diagnostic efficacy of the combined ROI indicator (ROI1 + ROI2) was superior to that of a single ROI indicator (AUC = 0.901 vs. AUC = 0.900 for ROI1). Figure 2H). For differentiating between multiple system atrophy (MSA) and healthy controls, the combination of ROI index and DTI-ALPS index achieved the highest diagnostic accuracy (AUC = 0.960), which was significantly better than ROI index alone; while the diagnostic efficacy of DTI-ALPS index alone was moderate (AUC = 0.836). Figure 2 I). In differentiating between Parkinson's disease and multiple system atrophy, the combined use of ROI1 + ROI2 and the DTI-ALPS index showed the strongest discriminative power (AUC = 0.982), which was significantly better than using the DTI-ALPS index alone (AUC = 0.612). Figure 2 J).

[0081] ANOVA was used for comparisons between groups, and pairwise comparisons were performed using a post-hoc Tukey multiple comparison test. P-values ​​were generated using this test. Error bars represent standard errors. Significance markers: * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.

[0082] Compared with the prior art, the technical solution described in this invention has the following significant advantages and inventiveness:

[0083] This invention is the first to reveal the differentiated pathophysiological mechanisms, providing a solid scientific basis for differential diagnosis. Existing technologies mostly remain at the level of isolated observation and simple comparison of a single indicator (such as DTI-ALPS or FW).

[0084] This invention achieves unprecedented high precision and reliability in differential diagnosis. By fusing multimodal imaging biomarkers, it achieves extremely high accuracy in differentiating between PD and MSA, significantly outperforming current clinical MRI diagnostic standards. The fusion model constructed in this invention (combining the DTI-ALPS index, PD-FW, and MSA-FW) exhibits an area under the curve (AUC) of 0.982 when distinguishing between PD and MSA, demonstrating near-perfect differentiation. This performance addresses the clinical challenge of high specificity but extremely low sensitivity (especially in the early stages) of MRI biomarkers in the current Movement Disorder Society International (MDS) MSA diagnostic criteria, providing a powerful tool for diagnosing early and complex cases, potentially reducing misdiagnosis and avoiding ineffective treatment.

[0085] This invention provides a novel, quantifiable, and reproducible combination of imaging biomarkers. Based entirely on objective imaging computational indicators (DTI-ALPS index, FW value), this invention eliminates the reliance on subjective experience (such as visual assessment of the "cross sign" and "hummingbird sign") in traditional MRI diagnosis, significantly reducing inter-diagnostic variability and improving the consistency and reproducibility of results. All data are acquired from routine MRI sequences, requiring no contrast agent injection or lumbar puncture, making it safe, non-invasive, highly acceptable to patients, and easy to promote and popularize in clinical practice. From ROI mapping and QSM validation to feature extraction and model calculation, this invention emphasizes automated or semi-automated processing, minimizing manual steps and subjective bias, ensuring the stability and scalability of the technical solution.

[0086] This invention possesses powerful capabilities for monitoring disease progression and assessing potential treatment efficacy. Beyond its applicability to cross-sectional differential diagnosis, its longitudinal analysis capabilities further enhance its value in disease management.

[0087] Dynamic monitoring: It can quantitatively monitor the evolution trajectory of glymphatic function (DTI-ALPS index) and FW accumulation over time, providing a sensitive tool for assessing the natural progression rate of the disease.

[0088] Efficacy assessment: If future therapies targeting the glymphatic system (such as AQP4 agonists) or neuroinflammation become available, this technical approach can serve as an ideal surrogate endpoint for objectively and quantitatively assessing in clinical trials whether the drug effectively improves the underlying pathophysiological processes.

[0089] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A PD and MSA differentiation system based on lymphoid system function and free water imaging, characterized in that, The system is used to identify modules including an image data acquisition module, a preprocessing module, a Glymphatic functional evaluation module, a free water imaging and analysis module, and a statistical modeling and diagnostic module, wherein: The image data acquisition module is used to acquire high-quality, multi-contrast raw MRI data; The preprocessing module is used to preprocess the data acquired by the image data acquisition module; The Glymphatic functional assessment module is used to calculate the DTI-ALPS index to assess the function of the brain's lymphoid system. The free water imaging and analysis module is used to quantitatively analyze the distribution and dynamic changes of free water in the whole brain; The statistical modeling and diagnostic module is used to integrate the DTI-ALPS index and free water distribution index to build a diagnostic classification model.

2. The PD and MSA discrimination system based on lymphoid system function and free water imaging according to claim 1, characterized in that, The image data acquisition module uses a high-field MRI scanner to simultaneously acquire the following sequences: High-resolution T1-weighted structural images are used for anatomical localization of brain regions, tissue segmentation, and subsequent image registration. DTI is used to calculate the DTI-ALPS index and FW plot. It is acquired using a dual-shell protocol and uses alternating acquisition of the front and rear phase encoding directions to eliminate magnetic susceptibility artifacts. QSW is used to clearly visualize the structure of the medullary vein. It is acquired using a multi-echo 3D gradient echo sequence and retains complex data for offline magnetic susceptibility reconstruction.

3. The PD and MSA discrimination system based on lymphoid system function and free water imaging according to claim 1, characterized in that, The preprocessing includes DTI preprocessing and spatial normalization, wherein: DTI preprocessing involves performing head motion and eddy current corrections on the raw DTI data and removing non-brain tissue. Spatial standardization normalizes each subject's T1 images and DTI-derived images to a standard brain space using linear and nonlinear registration algorithms, enabling group-level voxel-level comparative analysis.

4. The PD and MSA identification system based on lymphoid system function and free water imaging according to claim 1, characterized in that, The Glymphatic functional assessment module calculates the DTI-ALPS index using the following steps: Step S001: On an axial slice at the level of the lateral ventricle body in the standard MNI space, define spherical regions covering projection fibers and association fibers in the white matter regions of both hemispheres, respectively. Step S002: By registering the individual's T1 image and B0 image to the MNI template respectively and calculating its inverse deformation field, the above standard ROI template is accurately mapped inversely to the native DTI space of each subject individual. Step S003: Overlay the mapped ROI onto the QSM image for visual verification to ensure that the ROI covers at least one linear low-signal medullary vein, so as to ensure the anatomical accuracy of the ROI. Step S004: Within the ROI of the projected fibers and combined fibers of each individual, calculate the average diffusion coefficient along the x, y, and z directions respectively; Step S005: Calculate the DTI-ALPS index using the average diffusion coefficients of the three directions from step S004, and take the average value of the ROI results for both hemispheres. Step S006: The calculated DTI-ALPS index is processed using the ComBat algorithm to eliminate the influence of multicenter, multi-scan, and demographic confounding factors.

5. The PD and MSA identification system based on lymphoid system function and free water imaging according to claim 4, characterized in that, The formula for calculating the DTI-ALPS index is: DTI-ALPS index = mean(Dxx_proj, Dxx_assoc) / mean(Dyy_proj, Dzz_assoc), Wherein, Dxx_proj is the x-axis diffusion coefficient extracted from the projected fiber region, Dxx_assoc is the x-axis diffusion coefficient extracted from the combined fiber region, Dyy_proj is the y-axis diffusion coefficient extracted from the projected fiber region, and Dzz_assoc is the z-axis diffusion coefficient extracted from the combined fiber region.

6. The PD and MSA identification system based on lymphoid system function and free water imaging according to claim 1, characterized in that, The calculation process of the free water imaging and analysis module includes the following steps: Step S101: Based on multi-shell diffusion data, using the neurite direction dispersion and density imaging model, and through convex optimization to accelerate the microstructure imaging algorithm, calculate the free water fraction of each voxel and generate a whole-brain FW map. Step S102: Use FSL-FAST to segment the T1 image into gray matter, white matter and cerebrospinal fluid probability maps, and use a linear regression model to correct the influence of CSF partial volume effect on FW value to generate the corrected FW map. Step S103: Perform whole-brain voxel-level analysis on the corrected FW map to compare the differences between PD and MSA; Step S104: Define the significant clusters in the above analysis and comparison as disease-specific ROIs, and extract the average FW value of the disease-specific clusters for subsequent correlation and diagnostic analysis. Step S105: The ComBat algorithm is used to coordinate the average FW value of the extracted ROI in order to eliminate the influence of confounding factors.

7. The PD and MSA discrimination system based on lymphoid system function and free water imaging according to claim 6, characterized in that, Whole-brain voxel-level analysis comparisons were performed using inter-group comparisons. These inter-group comparisons involved standardizing the corrected FW plots to the MNI space and smoothing them, followed by unpaired t-tests for voxel-level comparisons of PD vs HC, MSA vs HC, and PD vs MSA. Multiple comparison corrections were performed using FDR.

8. The PD and MSA identification system based on lymphoid system function and free water imaging according to claim 1, characterized in that, The statistical modeling and diagnosis module uses the DTI-ALPS index, the average FW value of ROI1, and the average FW value of ROI2 as independent variables, and the diagnostic group as the dependent variable to construct a logistic regression model as a diagnostic classification model. Demographic covariates are included in the diagnostic classification model to correct for their influence, and the predicted probability of the diagnostic category is output. The formula is expressed as follows: Logit(P) = ln[ P(MSA) / (1 - P(MSA)) ] = 28.556 - 123.331×ROI1 + 75.792×ROI2 - 14.363×ALPS - 0.033×age + 0.133×gender, Where: P(MSA) represents the probability that the individual is diagnosed with MSA; 1 - P(MSA) represents the probability that the individual is diagnosed with PD; ALPS is the DTI-ALPS index; ROI1 is the average FW value of ROI1; ROI2 is the average FW value of ROI2; age is age; and gender is sex.

Images