Multi-system atrophy progress quantification method based on volume change of focus area

Abstract

The invention relates to a multi-system atrophy progress quantification method based on volume change of a lesion area. Comprising the following steps of: capturing details of a focus in an image by adopting a deep segmentation network based on multi-scale feature fusion and including a multi-scale UNet, and simultaneously considering overall context information; time sequence medical image data is used, a dynamic positioning algorithm is designed, position changes of a focus at different time points are tracked, and displacement characteristics including excursion of brainstem and basal node positions in progress are captured; a three-dimensional space-time volume change model is introduced, and a space and time conjoint analysis framework is established in combination with the three-dimensional geometrical characteristics including the volume and the surface area of the lesion area and the dynamic change trend of the time dimension; designing a lesion structure stability function, evaluating the stability or mutation of the lesion form in follow-up visit, and quantifying the continuity and abrupt property of atrophy; and calculating a priority region and a diffusion mode of atrophy in the focus based on spatial gradient analysis of morphological change.

Description

Technical Field

[0001] The present invention relates to a method for quantifying the progression of multiple system atrophy, in particular to a method for quantifying the progression of multiple system atrophy based on changes in the volume of a lesion area. Background Art

[0002] Although the current methods for quantifying the progression of multiple system atrophy (MSA) have made some progress in medical image analysis, follow-up data processing, and disease mechanism research, there are still many deficiencies and drawbacks, which seriously limit its wide application in clinical practice and the improvement of diagnosis and treatment efficiency. First, most of the existing methods are based on static image analysis and lack comprehensive capture and modeling of dynamic changes in the time dimension. As a progressive neurological disease, the changes in MSA lesions are not only reflected in the volume atrophy at a single time point, but also in the dynamic changes and expansion patterns over time. However, most traditional methods mainly focus on volume comparison analysis at static time points, which cannot accurately reflect the development trend of lesions in the time dimension, resulting in the difficulty of timely detection of weak changes in the early stage of the disease, delaying the diagnosis and intervention. Secondly, the lack of research on coordinated changes between regions is another significant problem. MSA lesions are usually distributed in multiple regions such as the brainstem, cerebellum, and basal ganglia. The changes in lesions in these regions are often not isolated, but have certain synergies and interactions. However, traditional methods usually analyze these regions separately, ignoring the dynamic coupling effects and propagation mechanisms between regions. For example, focal changes in the brainstem may affect the functions of the cerebellum and basal ganglia through neural pathways, but existing studies lack quantitative assessment of coordinated changes in multiple regions, resulting in an incomplete understanding of the disease progression pattern, which further affects the formulation of precision treatment.

[0003] In addition, the current methods for quantifying the progression of multiple system atrophy generally lack a multi-level indicator system, making it difficult to take into account local changes, overall changes, and systemic changes at the same time. The fine quantification of local changes (such as the atrophy rate of the core area of ​​the brainstem) can provide accurate early diagnostic information, but if only local changes are focused on, the overall progression trend of the lesion may be ignored. The simple analysis of overall changes may mask the specific changes of local lesions, especially in the early stages, when small changes in the overall volume may not attract enough clinical attention. In addition, if the coordinated changes between regions are not effectively integrated into the overall assessment, it will be difficult to reveal the global progression characteristics and disease dynamics of multiple system atrophy. Therefore, the lack of a unified, multi-level quantitative system is a major shortcoming of the current methods. In terms of temporal dynamic analysis, existing methods have obvious technical limitations in modeling nonlinear changes. The lesion changes of MSA may manifest as linear atrophy (e.g., slow changes in the early stage of the disease) and nonlinear accelerated atrophy (e.g., rapid deterioration in the middle stage of disease progression), and traditional methods are difficult to capture these nonlinear change characteristics. In addition, the weight control of time decay was not fully considered in the follow-up data analysis, which may lead to the underestimation of changes in the late stages of the disease or the failure to pay enough attention to subtle changes in the early stages.

[0004] Another limiting factor is the lack of accuracy in spatial partition analysis. Although some methods attempt to partition the lesion area for analysis, the granularity of the partition is often too coarse, such as simply dividing the lesion into a core area and a marginal area, ignoring the more detailed change patterns within the lesion. In addition, most existing methods lack the ability to calculate spatial gradients, making it difficult to identify the priority atrophy areas within the lesion and the diffusion path of the lesion, which is particularly important for precise intervention and prediction of disease progression.

[0005] In terms of technical application, current quantitative methods generally rely on traditional image processing technology, and lack the integration and application of advanced technologies such as artificial intelligence and deep learning. Although traditional methods can provide high accuracy in single image segmentation tasks, they are unable to cope with lesions with complex morphological changes at multiple time points. In addition, existing methods are weak in automatic data processing and dynamic adjustment of model parameters, resulting in low efficiency in the analysis of large-scale follow-up data. At the same time, due to the lack of a unified scoring system and quantitative rules, the evaluation results of different methods are difficult to compare in clinical practice, further limiting their promotion and application. Summary of the invention

[0006] The purpose of the present invention is to provide a method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area, thereby solving some of the drawbacks and shortcomings pointed out in the background technology.

[0007] The present invention solves the above-mentioned technical problems by adopting the following technical solution, which includes the following steps:

[0008] S1. Identification and dynamic positioning of lesion areas:

[0009] S1.1, using a deep segmentation network based on multi-scale feature fusion including multi-scale UNet to capture the details of the lesion in the image while taking into account the overall contextual information;

[0010] S1.2. Use time series medical imaging data to design a dynamic positioning algorithm to track the position changes of the lesion at different time points and capture the displacement characteristics of the brainstem and basal ganglia during progression;

[0011] S2. 3D spatiotemporal modeling of volume changes:

[0012] S2.1. Introduce a three-dimensional spatiotemporal volume change model, combine the three-dimensional geometric characteristics of the lesion area including volume and surface area and the dynamic change trend of the time dimension, and establish a joint spatial and temporal analysis framework;

[0013] S2.2. Design a lesion structure stability function to assess the stability or mutation of lesion morphology during follow-up and quantify the continuity and suddenness of atrophy;

[0014] S2.3. Calculate the priority areas and diffusion patterns of atrophy within the lesion based on spatial gradient analysis of morphological changes;

[0015] S3. Multi-level dynamic definition of quantitative indicators:

[0016] S3.1. Define hierarchical quantitative indicators based on lesion volume changes:

[0017] Local change index: The volume reduction rate of the lesion core area including the brainstem center is used to define the microscopic change characteristics;

[0018] Overall change index: The relative change rate of the overall lesion volume (RVC) is used to evaluate the macroscopic characteristics of disease progression;

[0019] Inter-regional coupling change index: The multi-regional coordinated changes of multiple system atrophy were evaluated by the degree of spatial coupling between lesions;

[0020] Create a dynamic progression scoring system to form a comprehensive quantitative score based on hierarchical quantitative indicators for clinical staging assessment;

[0021] S4. Exploration of progression patterns and key atrophy drivers:

[0022] S4.1. Using time series cluster analysis, extract the typical progression patterns of multiple system atrophy, including linear atrophy and accelerated atrophy, and conduct group comparative analysis of patients with different patterns;

[0023] S4.2. Design a key driver factor extraction algorithm to quantify the contribution of specific factors including atrophy starting point and expansion rate to overall progression based on imaging features, volume change indicators and patient clinical data;

[0024] S4.3. Introduce the lesion-affecting network, simulate the interaction between lesions, and calculate the propagation effect of multiple system atrophy on the functional network.

[0025] Furthermore, the three-dimensional spatiotemporal modeling method of volume change includes:

[0026] Acquire 3D medical imaging data at multiple time points, use registration technology, and spatially align the 3D images of continuous follow-up by matching feature points or image content. Then segment the image data at each time point based on a deep learning dynamic segmentation algorithm to obtain a 3D geometric model of the lesion area. The model includes the volume V of the lesion t Surface area A t , and the morphological characteristics of the lesion boundary B t ; For the volume change of the lesion, a spatiotemporal analysis framework is constructed to model the change trend of the lesion volume over time; and the dynamic trend of the volume change is described by the following spatiotemporal function:

[0027]

[0028] in:

[0029] It indicates the instantaneous rate of change of the lesion volume in the time dimension; it reflects the increase or decrease of the lesion volume at each moment during the follow-up process; It is the second-order Laplace operator of the lesion volume in three-dimensional space, which is used to describe the uniformity of the lesion volume diffusion; α is the control parameter, which is used to adjust the influence of spatial diffusion on volume change; β is the control parameter, which controls the speed of time decay; κ is the time decay coefficient, which describes the decay speed of the lesion over time.

[0030] Furthermore, the three-dimensional spatiotemporal modeling method of volume change includes:

[0031] After obtaining the three-dimensional geometric model of the lesion, the changes in the lesion morphology are tracked, including the rate of volume change and the degree of deformation of the surface area; and the stability of the lesion morphology is quantified by introducing the lesion structure stability function S(t), which calculates the rate of change of the surface area and the degree of deformation of the boundary contour:

[0032]

[0033] in:

[0034] It is the rate of change of surface area in the time dimension, indicating the increase or decrease of the lesion surface at each moment during the follow-up process; the change of surface area is directly related to the pattern of lesion shrinkage or expansion; w 1 (x, y, z) is the spatial weight function of the lesion area; different spatial positions are weighted according to the importance of each area within the lesion, and the contribution of each area is reflected in the calculation; |B t+1 B t | is the change amplitude of the lesion boundary in the time dimension, and the L2 norm is used to calculate the change of the boundary morphology; represents the degree of distortion of the lesion boundary contour; dΩ is the integral area, which represents the spatial range of the lesion area. The integral is used through all spatial areas to quantify the overall morphological change of the lesion.

[0035] Furthermore, the three-dimensional spatiotemporal modeling method of volume change includes:

[0036] The spatial gradient analysis method is used to perform fine-grained partitioning of the lesion area; the lesion area is divided into multiple sub-areas of the core area, edge area and diffusion area, and the change trends of the volume and surface area are calculated respectively; by calculating the spatial gradient G(x, y, z) of the internal morphological changes of the lesion, the priority areas with significant atrophy and the propagation mode of the changes are identified:

[0037]

[0038] in:

[0039] It is the gradient of the lesion volume in the spatial dimension, indicating the distribution characteristics of the lesion volume change in space; It is the second-order rate of change of volume in space and time dimensions, capturing the accelerated changes of lesions; calculating the change in the speed of expansion or shrinkage of lesions; is the spatial gradient of the rate of surface area change, accounting for subtle differences in lesion surface changes, and η and ξ are weight parameters that control the importance of volume change and surface area change in gradient analysis.

[0040] Furthermore, the multi-level dynamic definition method of the quantitative indicators includes:

[0041] For the core area of ​​the lesion including the brainstem center, cerebellum and basal ganglia, the three-dimensional geometric model of the area was extracted using a three-dimensional segmentation algorithm, and the rate of change of volume over time was calculated; the changes in the core area were volume atrophy and distortion of boundary morphology, and the dynamic change characteristics were evaluated by combining spatial and temporal gradient refinement; the definition of the local change characteristic function was as follows:

[0042]

[0043] in:

[0044] V(x,y,z) represents the lesion volume, which is the volume distribution at the lesion position (x,y,z) in three-dimensional space; is the time change rate of the volume, which is used to capture the dynamic changes of the core area of ​​the lesion in the time dimension; is the spatial gradient of the lesion boundary, which is used to quantify the severity of boundary changes; α is an adjustment parameter, which is used to balance the contribution of volume change rate and boundary change; Ω c is the spatial extent of the lesion core area, and the integral is passed through the entire core area to achieve feature quantification.

[0045] Furthermore, the multi-level dynamic definition method of the quantitative indicators includes:

[0046] The overall change index focuses on the dynamic change trend of the volume of the lesion in the entire spatial range; by calculating the relative change rate of the lesion, it reflects the change amplitude of the lesion volume relative to the initial state, and at the same time combines the dynamic changes of the follow-up time to quantify the linear and nonlinear change patterns; the overall change index formula is defined as follows:

[0047]

[0048] Formula explanation:

[0049] |Ω| is the volume of the entire spatial range of the lesion; V t and V t-1 are the total volumes of the lesions at time t and t-1, respectively; is the relative rate of change, used to quantify the volume increase or decrease over time; Δt is the time interval, reflecting the difference between consecutive time points; β is the time attenuation coefficient, used to control the effect of follow-up time on the rate of change; e -β · Δt It is a time decay term. As the follow-up time increases, its weight gradually decreases, reflecting the nonlinear trend of lesion changes.

[0050] Furthermore, the multi-level dynamic definition method of the quantitative indicators includes:

[0051] The synergy between lesion areas is quantified by the inter-regional coupling change index, and the inter-regional dynamic coupling function is defined as follows:

[0052]

[0053] in:

[0054] V i and V j are the volumes of lesion areas i and j respectively; and is the time change rate of volume, indicating the dynamic change characteristics of each region over time; ρ ijis the coupling coefficient between regions i and j, which is used to measure the intensity of the coordinated changes between the two regions; and is the spatial gradient of the regional volume, which is used to describe the characteristics of spatial variation; γ is the adjustment parameter, which controls the weight of temporal dynamics and spatial gradient; is the difference in spatial gradients between the two regions, and is used to quantify the spatial heterogeneity of the coordinated changes between the two regions.

[0055] Furthermore, the multi-level dynamic definition method of the quantitative indicators includes:

[0056] The dynamic progression scoring system generates a dynamic score to quantify the overall progression of the lesion by integrating local change indicators, overall change indicators, and inter-regional coupling change indicators. The scoring system is dynamically adjusted according to follow-up data to form a time curve of lesion progression and calculates the stage prediction of the disease in combination with the staging rules. The comprehensive scoring formula is defined as follows:

[0057]

[0058] in:

[0059] L(t) is the local change index score, which is used to capture the dynamic changes of the core area; RVC(t) is the overall change index score, which reflects the overall progression of the lesion; C ij (t) is the score for inter-regional coupling changes, quantifying the synergy of multiple regions; w 1 ,w 2 ,w 3 is a weight coefficient, which is jointly determined by clinical experts and data models and is used to balance the contribution of different indicators to the score. S(t) is a comprehensive score, which dynamically reflects the overall changes of the lesion during the follow-up period and provides a basis for disease staging and progression prediction.

[0060] The method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area of ​​the present invention has the following beneficial effects:

[0061] Through multi-level analysis of local change indicators, overall change indicators and inter-regional coupling change indicators, the dynamic change characteristics of the volume of the lesion area are comprehensively quantified. From the microscopic local to the macroscopic overall, as well as the synergistic effects between regions, the evolutionary laws of the lesions are fully revealed, which significantly improves the depth of understanding of the dynamic changes of multiple system atrophy lesions.

[0062] The introduction of time series analysis methods and dynamic progression scoring systems enables this method to adjust the scores in real time over time, generate a progression time curve, and predict the future change trend of lesions, providing a basis for clinical decision-making. It is especially suitable for patients with multiple system atrophy whose disease progression speed varies and whose manifestations are complex. Spatial gradient analysis and fine-grained partitioning methods are used to divide the lesion area into core area, edge area and diffusion area, and the change characteristics of these sub-areas are quantified respectively, which can identify the priority areas and diffusion paths of lesion atrophy and guide doctors to select key areas for precise intervention. Through the introduction of inter-regional coupling change indicators, the intensity and pattern of multi-regional coordinated changes are systematically quantified for the first time, revealing the dynamic interaction and propagation effects between lesions. It is particularly suitable for diseases involving multiple brain regions such as multiple system atrophy, providing a new perspective for the study of pathological mechanisms. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 The present invention is a flow chart of the method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area.

[0064] Figure 2 The figure is a flow chart of the three-dimensional spatiotemporal modeling method of volume change of the present invention.

[0065] Figure 3 This is a flow chart of the multi-level dynamic definition method of quantitative indicators of the present invention.

[0066] Figure 4 PET-CT imaging data are shown, with arrows indicating significantly reduced metabolism in the bilateral cerebellar hemispheres (A), which show areas of metabolic abnormalities compared to the cerebral hemispheres (B).

[0067] Figure 5 MRI images showing sagittal T1 (A) and FLAIR (B) sequences show marked cerebellar atrophy (arrows) with concomitant dilatation of the fourth ventricle, indicating that the structural lesions are consistent with the areas of functional abnormalities.

[0068] Figure 6 Axial T2 (A) and DWI (B) sequences show no obvious abnormalities in the basal ganglia (arrows). Although there are no significant lesions, the synergistic effects between the basal ganglia, cerebellum, and brainstem affect the overall disease progression. DETAILED DESCRIPTION

[0069] The specific implementation modes of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0070] Combined with Figure 1The process, a method for quantifying the progression of multiple system atrophy based on the volume changes of the lesion area, first involves the identification and dynamic positioning of the lesion area. In this part, the core technology is a deep segmentation network with multi-scale feature fusion, especially the multi-scale UNet (an extension of the U-Net architecture), which can effectively capture the detail information in the image at different scales and retain sufficient contextual information in the entire image. The UNet architecture itself is widely used in medical image processing due to its superior image segmentation capabilities, especially when dealing with tasks that require accurate segmentation of complex structures (such as lesion areas). Multi-scale UNet fuses convolutional features of different scales at different levels of feature extraction, allowing the model to focus on both the local details and the global context of the lesion area, thereby improving segmentation accuracy and avoiding segmentation errors caused by insufficient resolution or scale changes.

[0071] Subsequently, the time series medical imaging data can be analyzed through the designed dynamic positioning algorithm. The core function of this dynamic positioning algorithm is to track the changes in the lesion area at different time points, especially to track the position and morphological changes of key areas (such as the brainstem and basal ganglia). Unlike traditional static image processing methods, time series medical imaging data provides dynamic information about the changes of lesions over time, which is crucial for atrophic diseases (such as multiple system atrophy). At each follow-up, the dynamic positioning algorithm accurately extracts the displacement characteristics of the lesion location by comparing the imaging data at consecutive time points, paying special attention to the displacement of core areas such as the brainstem and basal ganglia. As the lesions change over time, atrophy in areas such as the brainstem and basal ganglia will show obvious spatial offset. The accurate capture of this offset is an important basis for evaluating the progression of the lesions.

[0072] By combining the three-dimensional geometric features of the lesion area (such as volume and surface area) and the dynamic change trends in the time dimension, an analytical framework that can comprehensively describe the evolution of the lesion was established. In the spatial dimension, the volume of the lesion represents its overall atrophy, while the surface area reflects the morphological characteristics of its boundary. By simultaneously analyzing the changes in volume and surface area, the progression pattern of the lesion can be more accurately revealed. The dynamic change trend in the time dimension captures the evolution trajectory of the lesion at different time points through sequential imaging data, providing a higher time resolution for quantitative analysis. By combining the spatial and temporal dimensions, the framework can dynamically monitor changes in the volume and morphology of the lesion and identify continuous changes and sudden accelerated atrophy of the lesion.

[0073] In order to quantify the stability or mutation of lesion morphology during follow-up, a lesion structural stability function was designed, which evaluates the fluctuation characteristics of lesion morphology through multi-parameter interaction. Specifically, the lesion structural stability function takes the volume change rate and surface area change rate of the lesion as the core parameters, and combines the deformation degree of the boundary morphology to comprehensively analyze the change trend of the lesion in the time dimension. This function can effectively distinguish whether the lesion presents a stable atrophy pattern or has sudden abnormal changes, providing a scientific basis for the continuous quantification of lesion morphological changes. For example, when the volume change rate of the lesion fluctuates little between multiple follow-up points and shows a linear trend, it can be determined that the lesion has a high stability; if the volume change rate suddenly increases or the surface area deformation is significant, it can be determined that the lesion has sudden changes. This quantification of stability and suddenness can help clinicians identify the key nodes of lesion progression and provide opportunities for early intervention.

[0074] Spatial gradient analysis based on morphological changes is an important part of this method. By finely partitioning the lesion area (such as core area, edge area and diffusion area), the gradient distribution of morphological changes inside the lesion can be further calculated. Spatial gradient analysis is based on the volume and surface area changes in different areas of the lesion to quantify the heterogeneity of its spatial distribution and the dynamic change pattern. By calculating the morphological change gradient within the lesion area, the priority areas with the most significant atrophy can be identified, and the pattern of atrophy spreading from the core to the edge or other areas can be analyzed. Spatial gradient analysis can also reveal the difference in changes between internal areas of the lesion and provide precise positioning for clinical intervention. For example, in the core area of ​​the lesion, a higher morphological change gradient may mean the main active area of ​​the lesion, while in the edge area, an increase in the gradient may indicate the expansion trend of the lesion. Ultimately, this analysis can not only generate a dynamic map of the lesion diffusion pattern, but also guide clinicians to select intervention target areas through precise regional priority sorting, providing comprehensive support for the precise treatment and follow-up management of the lesion.

[0075] Through the definition of multi-level dynamic quantitative indicators, the changes of lesions are described in a refined and multi-dimensional manner, so as to comprehensively evaluate the progression of lesions. First, in the definition of local change indicators, the focus is on the core area of ​​the lesion (such as the brainstem center), which is often the main lesion site for the progression of multiple system atrophy. The three-dimensional geometric model of the core area is obtained through a high-precision segmentation algorithm, and the rate of volume reduction is calculated in the time dimension. At the same time, morphological parameters such as morphological changes of the boundary and internal density changes are extracted. The core of the local change indicator is to capture the microscopic change characteristics inside the lesion, such as the rate of volume reduction in the brainstem area and the dynamic changes of surface area deformation. These characteristics can accurately reflect the local characteristics of lesion progression and provide a scientific basis for early diagnosis and intervention. Secondly, in the definition of the overall change indicator, the focus is on the volume change trend of the lesion within the overall spatial range. By calculating the relative rate of change (Relative Volume Change, RVC) of the lesion, the change in the total volume of the lesion relative to the baseline can be quantified. The RVC indicator combined with the dynamic changes of the follow-up time can reveal the overall linear or nonlinear progression pattern of the lesion. For example, as the disease worsens, RVC may show a nonlinear accelerated change trend, while it may tend to be flat during the stable period. The advantage of the overall change index is that it provides a global quantitative result of the lesion from a macroscopic level, and is a key parameter for judging the overall severity and progression rate of the disease. Third, the inter-regional coupling change index is extended to multiple regions of the lesion (such as the brainstem, cerebellum, and basal ganglia), and by quantifying their synergistic change characteristics, it reveals the systematic progression pattern of multiple system atrophy. Specifically, the inter-regional coupling change index constructs a dynamic coupling model between lesion regions by analyzing the dynamic correlation and spatial distribution characteristics of volume changes in different regions. This model can capture the mutual influence between different lesion regions, such as whether the atrophy of the brainstem region has a synergistic effect with the changes in the basal ganglia. By introducing spatial correlation and temporal dynamic characteristics, the degree of synergistic change between regions can be quantified, providing a quantitative basis for the study of the mechanism of systemic diseases. Finally, the above-mentioned hierarchical quantitative indicators are combined to create a dynamic progression scoring system to achieve comprehensive quantification of lesion changes. The dynamic progression scoring system integrates local change indicators, overall change indicators, and inter-regional coupling change indicators according to weights to generate a comprehensive score to describe the progression characteristics of lesions over time. The scoring system has a dynamic adjustment function, which can update the score according to the data at different follow-up time points to form a time curve of lesion progression.

[0076] Using time series cluster analysis, typical progression patterns of multiple system atrophy were extracted from the continuous follow-up imaging data, including linear atrophy patterns and accelerated atrophy patterns. In the linear atrophy pattern, the reduction in the volume of the lesion area showed a relatively stable and uniform trend of change, which is suitable for patients with a slower progression of the disease. In the accelerated atrophy pattern, the rate of change in the volume of the lesion area increased significantly at a specific time point or stage of the disease, showing a nonlinear acceleration trend. This analysis can help identify the characteristic differences in disease progression in patients with different patterns by grouping patients according to progression patterns, and provide data support for the clinical formulation of personalized treatment strategies.

[0077] Furthermore, in order to deeply understand the key factors affecting multiple system atrophy, a key driving factor extraction algorithm was designed. Based on the imaging features of the lesions, volume change indicators and clinical data of the patients, the algorithm quantifies the contribution of different factors to the overall progression of the disease through data mining technology. The starting point of atrophy is one of the important analysis targets of this method. By tracking the earliest location of atrophy in the time series images, the starting point of the spatial evolution of the lesion can be revealed and its key impact on the expansion of the disease can be analyzed. In addition, the expansion rate is another key driving factor, which describes the temporal dynamic characteristics of the volume and morphological changes of the lesion, and can reflect the diffusion speed of the lesion and its spatial propagation law. By quantifying the role of these specific factors, a more accurate basis can be provided for clinical prediction of disease progression and evaluation of treatment effects.

[0078] Finally, a lesion influence network was introduced to simulate the interaction relationship and synergistic progression mechanism between lesions, further expanding the understanding of the propagation effect of multiple system atrophy. The lesion influence network quantifies the interactions and propagation paths between lesion regions by constructing functional connections between lesion regions. Specifically, the network analyzes the structural associations and functional interactions between lesion regions, such as how atrophy of the brainstem affects the extension of lesions in the cerebellum and basal ganglia. By calculating the propagation effect of the lesion influence network, it is possible to reveal the diffusion characteristics of the disease on the functional network, quantify the synergistic effects between regions, and provide a new perspective for the study of the pathological mechanism of multiple system atrophy.

[0079] Embodiment 1:

[0080] Combined with Figure 2 Process, Patient A was diagnosed with early multiple system atrophy and received regular brain MRI examinations. The follow-up period was 12 months, and images were collected every 3 months. A total of 3D image data at 5 time points were collected (T = {0, 3, 6, 9, 12} months). The target areas included the brainstem, cerebellum, and basal ganglia. The purpose was to quantify the volume change characteristics of lesions in these areas and predict the future lesion spread trend through spatiotemporal modeling.

[0081] Brain imaging data of patient A at 5 time points were obtained through MRI examination, and the resolution of each image was 1mm. 3 The images at different time points are spatially aligned using registration technology to ensure spatial consistency of the lesion area. The registration algorithm is based on image content matching (such as key feature point alignment and global gradient optimization) to eliminate deviations caused by patient movement and imaging equipment.

[0082] Using deep learning dynamic segmentation algorithms (such as 3D-UNet), lesions in the brainstem, cerebellum, and basal ganglia are automatically segmented, and the three-dimensional geometric features of each lesion, including volume V_t, surface area A_t, and boundary B_t, are extracted. 0 = 0, the initial volume of the brainstem is measured Surface Area The lesion volume at each time point was then recorded as:

[0083] V t＝3 =12.1cm 3

[0084] V t＝6 =11.8cm 3

[0085] V t＝9 =11.3cm 3

[0086] V t＝12 =10.7cm 3

[0087] A spatiotemporal analysis framework was constructed to describe the dynamic trend of volume change using the following spatiotemporal functions:

[0088]

[0089] in:

[0090] Indicates the instantaneous rate of change of lesion volume.

[0091] is the second-order Laplace operator of the lesion volume, and the measured diffusion uniformity value is set

[0092] α=0.5 (control parameter, value range 0.1≤α≤1).

[0093] β=1.2 (time decay parameter, value range 1.0≤β≤2.0).

[0094] κ=0.1 (time attenuation coefficient, value range 0.05≤κ≤0.2).

[0095] Substitute the calculation results:

[0096] 1. In time interval t 0 =0 to t 1 =12 Total change calculation:

[0097] Instantaneous rate of change:

[0098] Laplace term contribution:

[0099] Time decay term: β·e -κT =1.2·e -0.1·12 =0.36.

[0100] 2. Substitute each term into:

[0101]

[0102] The results showed that the lesion volume decreased by a total of 1.56 cm in 12 months. 3 , showing an obvious shrinking trend.

[0103] The results showed that the lesions in the brainstem region of the patient continued to shrink within 12 months, and the shrinkage rate showed a nonlinear slowing trend over time. Through spatiotemporal modeling, the overall change trend of the lesions was accurately quantified, and the dynamic attenuation characteristics in the time dimension were captured.

[0104] This embodiment further analyzes the changing characteristics of the lesion morphology, including the volume change rate, the deformation degree of the surface area, and the stability quantification of the lesion morphology. This process describes and quantifies these dynamic changes by introducing the lesion structure stability function S(t).

[0105] Initial surface area A of the cerebellar lesion in patient A during imaging follow-up at time T = {0, 3, 6, 9, 12} months t＝0 =54.3cm 2 , and the subsequent measurement results are:

[0106] A t＝3 =53.5cm 2

[0107] A t＝6 =52.4cm 2

[0108] A t＝9 =50.8cm 2

[0109] A t＝12 =49.2cm 2

[0110] Boundary morphology of cerebellar lesions Bt The distortion gradually occurred during the follow-up period, and the L2 norm of the boundary change at each time point was measured as:

[0111] |B t＝3 B t＝0 |=0.8

[0112] |B t＝6 B t＝3 |=1.1

[0113] |B t＝9 B t＝6 |=1.6

[0114] |B t＝12 B t＝9 |=1.8

[0115] The lesion structural stability function S(t) is used to comprehensively quantify the surface area change and boundary distortion, and the formula is as follows:

[0116]

[0117] in:

[0118] is the time rate of change of surface area, which can be expressed by calculate.

[0119] w 1 (x, y, z) is the spatial weight function. In cerebellar lesions, the weight of the core area of ​​the lesion (such as the anterior lobe of the cerebellum) is set to 1.5, and the weight of the edge area is 1.0. δ = 0.7 is the control coefficient (value range: 0.5 ≤ δ ≤ 1.0), which is used to balance the contribution of surface area change and boundary morphological distortion. dΩ represents the spatial range of the lesion area.

[0120] Calculation steps and results:

[0121] 1. Calculation of surface area change rate:

[0122] Calculated within the time interval Δt = 3 months:

[0123]

[0124] 2. Substitute into the formula to calculate S(t):

[0125] S(t=3)=∫ Ω (-0.27·1.5+0.7·0.8)dΩ

[0126] The proportion of cerebellar lesions in the integral area is set to 10 units of space, and the calculation is:

[0127] S(t=3)=∫Ω (-0.405+0.56)dΩ=∫ Ω 0.155 10 = 1.55

[0128] S(t=6)=∫ Ω (-0.37·1.5+0.7·1.1)dΩ

[0129] S(t=6)=∫ Ω (-0.555+0.77)dΩ=∫ Ω 0.215 10 = 2.15

[0130] S(t=9)=∫ Ω (-0.53·1.5+0.7·1.6)dΩ

[0131] S(t=9)=∫ Ω (-0.795+1.12)dΩ=∫ Ω 0.325 10 = 3.25

[0132] S(t=12)=∫ Ω (-0.53·1.5+0.7·1.8)dΩ

[0133] S(t=12)=∫ Ω (-0.795+1.26)dΩ=∫ Ω 0.465 10 = 4.65

[0134] The calculation results show that the lesion structural stability function S(t) gradually increases over time, indicating that the surface area change rate and boundary distortion degree of the lesion increased during the follow-up process. In particular, at t=9 and t=12, the boundary change and surface area change rates reached peak values, indicating that the dynamic changes of the lesion at these time points were most significant, which may be the critical stage of disease aggravation. By quantifying S(t), doctors can accurately locate the key time points and areas of lesion progression, thereby providing a scientific basis for formulating intervention strategies.

[0135] This embodiment focuses on analyzing the spatial gradient characteristics of cerebellar lesions, uses the spatial gradient analysis method to perform fine-grained partitioning of the lesion area, further quantifies the volume and surface area change trends of the core area, edge area, and diffusion area, and identifies the priority areas with significant atrophy and the change propagation pattern through the spatial gradient G(x, y, z, t) of morphological changes.

[0136] In the previous analysis, the total volume of the cerebellar lesions V was calibrated by follow-up imaging data. t and surface area A t Trends over time. Cerebellar lesions are now divided into three subregions:

[0137] Core area (40% of the total volume): the lesion changes are most concentrated;

[0138] Edge zone (35% of the total volume): a transitional area adjacent to the core zone;

[0139] Diffuse area (accounting for 25% of the total volume): the part of the lesion that expands to the surrounding area.

[0140] The initial volume and surface area of ​​each subregion are calculated by the segmentation algorithm:

[0141] Core area:

[0142] Fringe Area:

[0143] Diffusion area:

[0144] During the follow-up period, the volume and surface area changes of these sub-areas were gradually measured and recorded. The core area atrophied fastest, and the diffusion area changed the slowest. In order to analyze the morphological changes of the lesions, the spatial gradient G(x,y,z,t) was calculated using the following formula:

[0145]

[0146] Formula explanation:

[0147] The gradient of volume in the spatial dimension is used to describe the distribution characteristics of the lesion volume at different locations;

[0148] The second-order rate of change of volume in space and time quantifies the intensity of accelerated change of the lesion;

[0149] The spatial gradient of the rate of surface area change is used to reveal subtle differences in lesion surface changes;

[0150] η=0.8,ξ=0.5:weight parameters, which respectively control the influence of volume change and surface area change on the gradient (value range 0.5≤η≤1.0,0.3≤ξ≤0.7).

[0151] At the initial moment, the core area volume distribution gradient

[0152] Second-order rate of change of volume in time dimension:

[0153]

[0154] The spatial gradient of the rate of change of surface area:

[0155]

[0156] Substituting into the formula:

[0157] G core (t=3)=-0.15+0.8·(-0.01)+0.5·(-0.01)=-0.159cm 3 / cm

[0158] Fringe Area:

[0159] G edge (t=3)=-0.12+0.8·(-0.008)+0.5·(-0.009)=-0.1254cm 3 / cm

[0160] Diffusion area:

[0161] G spread (t=3)=-0.09+0.8·(-0.005)+0.5·(-0.007)=-0.0946cm 3 / cm

[0162] The results show that the gradient value G in the core area core (t=3)=-0.159cm 3 The gradient value of the edge area is the second highest, and the diffusion area is the smallest, indicating that the lesion gradually spreads from the core area to the edge and diffusion area. This gradient change also reveals the spread pattern of the lesion, which is consistent with the clinical characteristics of multiple system atrophy lesions extending from the center to the periphery.

[0163] Embodiment 2:

[0164] Combined with Figure 3 The process is carried out based on the example of a patient A, focusing on analyzing the core areas of the lesions in the brainstem center, cerebellum and basal ganglia. The 3D geometric models of these areas are extracted through a high-precision 3D segmentation algorithm, and the rate of change of the volume over time and the degree of distortion of the boundary morphology are quantified. The following are the specific implementation steps of the solution and the calculation results:

[0165] During the follow-up of patient A at time T = {0, 3, 6, 9, 12} months, the 3D geometric model of the lesion core area (such as the brainstem center) was extracted through MRI images. t＝0 Measured 12.5cm 3 , initial boundary gradient At subsequent time points, the measured volume and boundary changes were as follows:

[0166] V t＝3 =12.1cm 3 ,

[0167] V t＝6 =11.8cm 3 ,

[0168] V t＝9 =11.3cm 3 ,

[0169] V t＝12 =10.7cm 3 ,

[0170] In order to quantify the local change characteristics of the core area of ​​the lesion, the local change characteristic function L(x,y,z,t) is introduced:

[0171]

[0172] Formula explanation:

[0173] V(x,y,z): represents the lesion volume, which is the volume distribution of the lesion location (x,y,z) in three-dimensional space; The temporal rate of change of volume reflects the dynamic shrinkage characteristics of the core area of ​​the lesion; The spatial gradient of the lesion boundary is used to quantify the degree of drastic change of the boundary; α = 0.7 is an adjustment parameter (value range: 0.5 ≤ α ≤ 1.0) used to balance the contribution of volume change rate and boundary morphological change; Ω c : The spatial extent of the lesion core area, through which the integral is quantified.

[0174] Calculate the time rate of change of volume:

[0175] The calculation within the time interval Δt = 3 months is as follows:

[0176] t=3:

[0177] t=6:

[0178] t=9:

[0179] t=12:

[0180] Compute the squared gradient term of the boundary shape:

[0181] At each time point the following is calculated:

[0182] t=3:(0.8) 2 =0.64cm -2

[0183] t=6:(1.1) 2 =1.21cm -2

[0184] t=9:(1.5) 2 =2.25cm -2

[0185] t=12:(1.9) 2 =3.61cm -2

[0186] Substitute into the formula to calculate L(t):

[0187] Set the spatial range of the lesion core area Ω c For 10 unit spaces, calculate L(t):

[0188]

[0189]

[0190] The calculation results show that the local change feature L(t) of the core area of ​​the lesion increases significantly over time, from 3.18 at t = 3 to 23.27 at t = 12. This trend reflects the dynamic change process of the lesion volume change rate and boundary morphological distortion. Especially in the later follow-up period of t = 9 and t = 12, the contribution of boundary changes increases significantly, indicating that the morphological changes in the core area of ​​the lesion tend to be drastic.

[0191] This embodiment then evaluates the dynamic change trend of the volume of the lesion in the entire spatial range through the overall change index. The overall change index focuses on quantifying the change in the lesion volume relative to the initial state, and combines the dynamic changes in the follow-up time to evaluate the linear and nonlinear change patterns of the lesion. The overall change index RVC(t) is defined by the following formula:

[0192]

[0193] The total volume of brainstem lesions in patient A is V t The data measured during the follow-up period T = {0, 3, 6, 9, 12} months are as follows:

[0194] V t＝0 =12.5cm 3

[0195] V t＝3 =12.1cm 3

[0196] V t＝6 =11.8cm 3

[0197] V t＝9 =11.3cm 3

[0198] V t＝12 =10.7cm 3

[0199] The overall spatial range of the lesion |Ω| is set to 50 unit spaces, the time interval Δt=3 months, and the time attenuation coefficient β=0.1 (the value range is 0.05≤β≤0.2).

[0200] Calculate the relative rate of change

[0201] t=3:

[0202] t=6:

[0203] t=9:

[0204] t=12:

[0205] Calculate the time decay term e -β·Δt :

[0206] t=3:e -0.1·3 =e -0.3 ≈0.7408

[0207] t=6:e -0.1·6 =e -0.6 ≈0.5488

[0208] t=9:e -0.1·9 =e -0.9 ≈0.4066

[0209] t=12:e -0.1·12 =e -1.2 ≈0.3012

[0210] Substitute into the formula to calculate RVC(t):

[0211] t=3:

[0212]

[0213] t=6:

[0214]

[0215] t=9:

[0216]

[0217] t=12:

[0218]

[0219] The results showed that the relative change rate RVC(t) of the brainstem lesion of patient A showed a gradual downward trend, from -0.0237 at t=3 to -0.0159 at t=12. This result indicates that the overall volume dynamic change of the lesion is more significant in the early follow-up, while the change tends to be gentle in the later period. -β·Δt The introduction of gradually reduces the weight of lesion changes over time, thereby paying more attention to the dynamic changes in the early stage of follow-up.

[0220] In order to further analyze the progression characteristics of multiple system atrophy in patient A, the synergy between the three major lesion areas of brainstem, cerebellum and basal ganglia was quantified by the inter-regional coupling change index. Combined with the follow-up data of the patient, the dynamic coupling degree between the lesion areas was modeled to evaluate their synergistic change characteristics.

[0221] The MRI data of patient A recorded the volume changes of the brainstem (region i), cerebellum (region j), and basal ganglia (region k) over time T = {0, 3, 6, 9, 12} months:

[0222] Brainstem V i (t): 12.5,12.1,11.8,11.3,10.7cm 3

[0223] Cerebellum V j (t): 42.8, 41.9, 40.7, 39.5, 38.0cm 3

[0224] Basal ganglia V k (t): 10.2, 10.0, 9.7, 9.3, 8.9cm 3

[0225] The dynamic coupling function of the coupling change indicator is defined as follows:

[0226]

[0227] Formula explanation:

[0228] and Respectively represent the volume change rate of the brainstem and cerebellum over time, reflecting the dynamic changes of the two regions over time; ρ ij =0.8 (value range 0.5≤ρ ij ≤1.0): coupling coefficient between brainstem and cerebellum, used to measure the strength of the coordinated changes in the two regions; and The volumetric spatial gradient of the brainstem and cerebellum is used to describe the spatial variation characteristics; γ = 0.6 (value range 0.3 ≤ γ ≤ 0.7): adjustment parameters to control the weight of temporal dynamics and spatial gradients; Differences in spatial gradients between the brainstem and cerebellum were used to quantify the heterogeneity of covariation.

[0229] Calculate the time rate of change of volume:

[0230] Brainstem

[0231] t=3:

[0232] t=6:

[0233] t=9:

[0234] t=12:

[0235] Cerebellum

[0236] t=3:

[0237] t=6:

[0238] t=9:

[0239] t=12:

[0240] Compute spatial gradient differences

[0241] The spatial gradients of the brainstem and cerebellum were measured as follows:

[0242]

[0243] At each time point:

[0244] t=3:|0.6·1.2| 2 =0.5184cm -6

[0245] t=6:|0.8·1.5| 2 =1.44cm -6

[0246] t=9:|1.0·1.8| 2 =3.24cm -6

[0247] t=12:|1.2·2.0| 2 =5.76cm -6

[0248] Substitute into the formula to calculate C ij (t):

[0249] t=3:

[0250] C ij (3) = ∫ 0 3 (0.8·(-0.13)·(-0.30)+0.6·0.5184)dt=0.8·0.039+0.6·0.5184

[0251] =0.0312+0.310=0.3412

[0252] t=6:

[0253] C ij (6) = ∫ 3 6 (0.8·(-0.10)·(-0.40)+0.6·1.44)dt=0.8·0.04+0.6·1.44

[0254] =0.032+0.864=0.896

[0255] t=9:

[0256] C ij (9) = ∫ 6 9 (0.8·(-0.17)·(-0.40)+0.6·3.24)dt=0.8·0.068+0.6·3.24

[0257] =0.0544+1.944=1.9984

[0258] t=12:

[0259] C ij (12) = ∫ 9 12 (0.8·(-0.20)·(-0.50)+0.6·5.76)dt=0.8·0.1+0.6·5.76

[0260] =0.08+3.456=3.536

[0261] The calculation results show that the coupling intensity between the brainstem and cerebellum increases gradually with time, from C ij (3) = 0.3412 rises to C at t = 12 ij(12) = 3.536. This indicates that as the disease progresses, the coordinated atrophy between the brainstem and cerebellum becomes more significant, especially in the later stages (t = 9 to t = 12), when the synchronization of gradient differences and volume changes is greatly improved.

[0262] In order to continue to analyze the progression of multiple system atrophy (MSA) in patient A, the dynamic progression scoring system was used to integrate the local change index L(t), the overall change index RVC(t), and the inter-regional coupling change index · to generate a dynamic score S(t) to quantify the overall progression of the lesion and predict the disease stage. The local change index, overall change index, and inter-regional coupling change index of the patient have been calculated in the previous analysis as follows:

[0263] Local variation index L(t): L(3)=3.18, L(6)=7.47, L(9)=14.05, L(12)=23.27.

[0264] Overall change indicators:

[0265] RVC(t): RVC(3)=-0.0237, RVC(6)=-0.0137, RVC(9)=-0.0171, RVC(12)=-0.0159.

[0266] Inter-regional coupling change index C ij (t) (synergistic score of brainstem and cerebellum): C ij (3) = 0.3412, C ij (6) = 0.896, C ij (9) = 1.9984, C ij (12)=3.536.

[0267] The comprehensive formula of the dynamic progress scoring system is:

[0268]

[0269] in:

[0270] w 1 =0.5(value range 0.3≤w 1 ≤0.6), the weight distribution is used to emphasize the contribution of local change indicators;

[0271] w 2 =0.3 (value range 0.2≤w_2≤0.4, weight distribution is used to balance the impact of the overall change index;

[0272] w 3 =0.2(value range 0.1≤w 3 ≤0.3), and the weight distribution is used to reflect the effect of inter-regional coupling changes.

[0273] Time point t=3:

[0274] S(3)=0.5·3.18+0.3·(-0.0237)+0.2·0.3412

[0275] S(3)=1.59+(-0.00711)+0.06824=1.65113

[0276] 2. Time point t=6:

[0277] S(6)=0.5·7.47+0.3·(-0.0137)+0.2·0.896

[0278] S(6)=3.735+(-0.00411)+0.1792=3.91009

[0279] 3. Time point t=9:

[0280] S(9)=0.5·14.05+0.3·(-0.0171)+0.2·1.9984

[0281] S(9)=7.025+(-0.00513)+0.39968=7.41955

[0282] 4. Time point t=12:

[0283] S(12)=0.5·23.27+0.3·(-0.0159)+0.2·3.536

[0284] S(12)=11.635+(-0.00477)+0.7072=12.33743

[0285] The calculation results show that the dynamic progression score S(t) of patient A gradually increases over time, specifically from 1.651 at t=3 to 12.337 at t=12, showing the overall progression trend of the lesion. This scoring system integrates information at three levels: local changes, overall changes, and coordinated changes between regions, and comprehensively reflects the progression characteristics of the lesion. In the early stages (t=3 and t=6), the score is mainly driven by the local change index L(t), indicating that changes in the core areas of the brainstem and cerebellum dominate the early progression. In the later stages (t=9 and t=12), the inter-regional coupling changes C ij The contribution of (t) increased significantly, suggesting an enhanced synergy between the brainstem and cerebellum, reflecting the systemic progression of multiple system atrophy.

[0286] According to the time curve of the score, the stage of the disease can be further predicted in combination with the clinical staging rules. The clinical staging threshold is set as:

[0287] Mild: S(t)<5

[0288] Moderate: 5≤S(t)<10

[0289] Severe: S(t)≥10

[0290] At t=3 and t=6, the patient scores were 1.651 and 3.910, respectively, which belonged to the mild stage; at t=9, the score rose to 7.419, indicating that the disease entered the moderate stage; at t=12, the score was 12.337, indicating that the disease had progressed to the severe stage. This dynamic scoring system provides doctors with a clear staging basis and can be used to determine the key progression nodes of the disease, providing scientific support for the formulation of treatment strategies and follow-up plans. This example verifies that the dynamic progression scoring system can effectively integrate multi-level indicators to achieve a comprehensive quantitative assessment of the progression of multiple system atrophy, demonstrating its practicality and feasibility in clinical applications.

[0291] Embodiment 3:

[0292] This embodiment combines PET-CT and MRI multimodal imaging data and uses the method of the present invention to calculate the local change index L(t), the overall change index RVC(t) and the inter-regional coupling change index C ij The dynamic score S(t) is generated by calculating the volume change characteristics of the lesion, which provides support for disease staging prediction and personalized treatment.

[0293] Figure 4 PET-CT imaging data are shown, with arrows indicating significantly reduced metabolism in the bilateral cerebellar hemispheres (A), which show areas of metabolic abnormalities compared to the cerebral hemispheres (B). Areas of reduced metabolism suggest that there may be progressive lesions in the cerebellum, which is a characteristic manifestation of multiple system atrophy.

[0294] Formula-based analysis:

[0295] 1. Quantify the local changes in the cerebellar lesion area using the formula L(t):

[0296]

[0297] The volume change rate describes the dynamic changes of the metabolic abnormality area over time;

[0298] The spatial gradient of boundary morphology changes reflects the irregular changes in boundaries;

[0299] Ω c : Cerebellar metabolic abnormality area segmented by PET-CT, initial volume V t＝0 =12.0cm 3 , V after 6 months of follow-upt＝6 =11.2cm 3 .

[0300] calculate:

[0301]

[0302] The measured value is 0.8.

[0303] Substituting into the formula and setting α = 0.7, we get:

[0304]

[0305] (Set the area integration range to 12 unit space)

[0306] Results: The local change index L (t=6) showed that the cerebellar metabolic area was shrinking significantly and the boundary morphology was changing more rapidly.

[0307] Figure 5 MRI images showing sagittal T1 (A) and FLAIR (B) sequences show marked cerebellar atrophy (arrows) with concomitant dilatation of the fourth ventricle, indicating that the structural lesions are consistent with the areas of functional abnormalities.

[0308] Formula-based analysis:

[0309] 1. Use the formula RVC(t) to quantify the volume change of the entire cerebellum:

[0310]

[0311] V t and V t-1 : The values ​​of cerebellum volume at different time points of follow-up;

[0312] β = 0.1, the time decay coefficient controls the impact of time span on the overall change.

[0313] Set the follow-up time V t＝0 =45.0cm 3 , V t＝6 =42.5cm 3 , V t＝12 =40.0cm 3 ,calculate:

[0314] When t=6 When t=12

[0315] e -β·Δt :t=6 is e -0.1·6 =0.5488, e at t = 12-0.1·12 =0.3012.

[0316] Substituting into the formula:

[0317]

[0318] The overall change indicators show that the rate of cerebellar atrophy presents a nonlinear trend, with a relatively slow change in the early stage and a significant acceleration in the later stage.

[0319] Figure 6 Axial T2 (A) and DWI (B) sequences show no obvious abnormalities in the basal ganglia (arrows). Although there are no significant lesions, the synergistic effects between the basal ganglia and the cerebellum and brainstem may affect the overall disease progression.

[0320] Formula-based analysis:

[0321] Use Formula C ij (t) Analysis of coordinated changes in the basal ganglia and cerebellum:

[0322]

[0323] Set the coupling coefficient ρ between the basal ganglia and the cerebellum ij =0.8, spatial gradient difference

[0324] Basal ganglia volume change rate Cerebellar rate

[0325] calculate:

[0326] C ij (t=6)=∫ 0 6 (0.8·(-0.10)·(-0.20)+0.6·0.5)dt=0.48

[0327] Inter-regional coupling change index C ij (t=6) showed that there was significant synergy between the basal ganglia and the cerebellum, which may be involved in the dynamic changes of cerebellar lesions.

[0328] Calculation of comprehensive score S(t):

[0329] Use the formula:

[0330]

[0331] Set the weight w 1 =0.5, w 2 =0.3, w 3 =0.2, calculate the comprehensive score:

[0332] S(t=6)=0.5·5.112+0.3·(-0.0305)+0.2·0.48=2.5560.00915+0.096=2.64285

[0333] S(t=12)=0.5·12.5+0.3·(-0.0196)+0.2·0.96=6.250.00588+0.192=6.43612

[0334] Based on the quantitative method of the present invention, combined with the analysis results of the three images, the dynamic score showed that the patient progressed from the early mild stage S (t = 6) = 2.64 to the mid-term severe stage S (t = 12) = 6.43. These results effectively quantified the dynamic changes, overall trends and regional synergy of cerebellar lesions, providing a scientific basis for disease staging and treatment optimization.

Claims

1. A method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area, characterized by The following steps are involved: S1. Identification and dynamic positioning of lesion areas: S1.1, using a deep segmentation network based on multi-scale feature fusion including multi-scale UNet to capture the details of the lesion in the image while taking into account the overall contextual information; S1.

2. Use time series medical imaging data to design a dynamic positioning algorithm to track the position changes of the lesion at different time points and capture the displacement characteristics of the brainstem and basal ganglia during progression; S2. 3D spatiotemporal modeling of volume changes: S2.

1. Introduce a three-dimensional spatiotemporal volume change model, combine the three-dimensional geometric characteristics of the lesion area including volume and surface area and the dynamic change trend of the time dimension, and establish a joint spatial and temporal analysis framework; S2.

2. Design a lesion structure stability function to assess the stability or mutation of lesion morphology during follow-up and quantify the continuity and suddenness of atrophy; S2.

3. Calculate the priority areas and diffusion patterns of atrophy within the lesion based on spatial gradient analysis of morphological changes; S3. Multi-level dynamic definition of quantitative indicators: S3.

1. Define hierarchical quantitative indicators based on lesion volume changes: Local change index: The volume reduction rate of the lesion core area including the brainstem center is used to define the microscopic change characteristics; Overall change index: The relative change rate of the overall lesion volume (RVC) is used to evaluate the macroscopic characteristics of disease progression; Inter-regional coupling change index: The multi-regional coordinated changes of multiple system atrophy were evaluated by the degree of spatial coupling between lesions; Create a dynamic progression scoring system to form a comprehensive quantitative score based on hierarchical quantitative indicators for clinical staging assessment; S4. Exploration of progression patterns and key atrophy drivers: S4.

1. Using time series cluster analysis, extract the typical progression patterns of multiple system atrophy, including linear atrophy and accelerated atrophy, and conduct group comparative analysis of patients with different patterns; S4.

2. Design a key driver factor extraction algorithm to quantify the contribution of specific factors including atrophy starting point and expansion rate to overall progression based on imaging features, volume change indicators and patient clinical data; S4.

3. Introduce the lesion-affecting network, simulate the interaction between lesions, and calculate the propagation effect of multiple system atrophy on the functional network.

2. The method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area according to claim 1, characterized in that The three-dimensional spatiotemporal modeling method of volume change includes: Acquire 3D medical imaging data at multiple time points, use registration technology, and spatially align the 3D images of continuous follow-up by matching feature points or image content. Then segment the image data at each time point based on a deep learning dynamic segmentation algorithm to obtain a 3D geometric model of the lesion area. The model includes the volume V of the lesion t Surface area A t , and the morphological characteristics of the lesion boundary B t ; For the volume change of the lesion, a spatiotemporal analysis framework is constructed to model the change trend of the lesion volume over time; and the dynamic trend of the volume change is described by the following spatiotemporal function: in: It indicates the instantaneous rate of change of the lesion volume in the time dimension; it reflects the increase or decrease of the lesion volume at each moment during the follow-up process; It is the second-order Laplace operator of the lesion volume in three-dimensional space, which is used to describe the uniformity of the lesion volume diffusion; α is the control parameter, which is used to adjust the influence of spatial diffusion on volume change; β is the control parameter, which controls the speed of time decay; κ is the time decay coefficient, which describes the decay speed of the lesion over time.

3. The method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area according to claim 2, characterized in that The three-dimensional spatiotemporal modeling method of volume change includes: After obtaining the three-dimensional geometric model of the lesion, the changes in the lesion morphology are tracked, including the rate of volume change and the degree of deformation of the surface area; and the stability of the lesion morphology is quantified by introducing the lesion structure stability function S(t), which calculates the rate of change of the surface area and the degree of deformation of the boundary contour: in: is the rate of change of surface area in the time dimension, indicating the increase or decrease of the lesion surface at each moment during the follow-up process; the change of surface area is directly related to the pattern of lesion shrinkage or expansion; w1(x, y, z) is the spatial weight function of the lesion area; different spatial positions are weighted according to the importance of each area in the lesion, and the contribution of each area is reflected in the calculation; |B t+1 B t | is the change amplitude of the lesion boundary in the time dimension, and the L2 norm is used to calculate the change of the boundary morphology; represents the degree of distortion of the lesion boundary contour; dΩ is the integral area, which represents the spatial range of the lesion area. The integral is used through all spatial areas to quantify the overall morphological change of the lesion.

4. The method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area according to claim 3, characterized in that The three-dimensional spatiotemporal modeling method of volume change includes: using a spatial gradient analysis method to perform fine-grained zoning of the lesion area; dividing the lesion area into multiple sub-areas of a core area, an edge area, and a diffusion area, and calculating the change trends of the volume and surface area respectively; by calculating the spatial gradient of the internal morphological changes of the lesion, identifying the priority areas with significant atrophy and the propagation pattern of the changes.

5. The method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area according to claim 1, characterized in that The multi-level dynamic definition method of the quantitative indicators includes: For the core area of ​​the lesion including the brainstem center, cerebellum and basal ganglia, the three-dimensional geometric model of the area was extracted using a three-dimensional segmentation algorithm, and the rate of change of volume over time was calculated; the changes in the core area were volume atrophy and distortion of boundary morphology, and the dynamic change characteristics were evaluated by combining spatial and temporal gradient refinement; the definition of the local change characteristic function was as follows: in: V(x,y,z) represents the lesion volume, which is the volume distribution at the lesion position (x,y,z) in three-dimensional space; is the time change rate of the volume, which is used to capture the dynamic changes of the core area of ​​the lesion in the time dimension; is the spatial gradient of the lesion boundary, which is used to quantify the severity of boundary changes; α is an adjustment parameter, which is used to balance the contribution of volume change rate and boundary change; Ω c is the spatial extent of the lesion core area, and the integral is passed through the entire core area to achieve feature quantification.

6. The method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area according to claim 5, characterized in that The multi-level dynamic definition method of the quantitative index includes: the overall change index focuses on the dynamic change trend of the volume of the lesion in the entire spatial range; by calculating the relative change rate of the lesion, it reflects the change amplitude of the lesion volume relative to the initial state, and at the same time combines the dynamic changes of the follow-up time to quantify the linear and nonlinear change patterns.

7. The method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area according to claim 1, characterized in that The multi-level dynamic definition method of the quantitative indicators includes: The synergy between lesion areas is quantified by the inter-regional coupling change index, and the inter-regional dynamic coupling function is defined as follows: in: V i and V j are the volumes of lesion areas i and j respectively; and is the time change rate of volume, indicating the dynamic change characteristics of each region over time; ρ ij is the coupling coefficient between regions i and j, which is used to measure the intensity of the coordinated changes between the two regions; and is the spatial gradient of the regional volume, which is used to describe the characteristics of spatial variation; γ is the adjustment parameter, which controls the weight of temporal dynamics and spatial gradient; is the difference in spatial gradients between the two regions, and is used to quantify the spatial heterogeneity of the coordinated changes between the two regions.

8. The method for quantifying the progression of multiple system atrophy based on the volume change of the lesion area according to claim 1, characterized in that The multi-level dynamic definition method of the quantitative indicators includes: the dynamic progression scoring system generates a dynamic score to quantify the overall progression of the lesion by integrating local change indicators, overall change indicators and inter-regional coupling change indicators; the scoring system is dynamically adjusted according to follow-up data to form a time curve of lesion progression, and the staging prediction of the disease is calculated in combination with the staging rules.

Images