Microdevices and methods for measuring synaptic morphology changes in neurons after aluminum exposure

Abstract

The present application relates to neuroscience and biomedical image processing field, particularly to a device and method for measuring synaptic morphology changes of neurons after aluminum exposure, the device comprises a confocal microscopic imaging module, a preprocessing module, an edge detection module, an edge point classification module, a geodesic network construction module and a morphological parameter quantification module, the present application introduces differential geometry theory, through three key innovative technologies of curvature driven edge recognition, manifold hierarchical decomposition edge point classification and geodesic network morphological parameter quantification, high-precision recognition of neuron synaptic edge and accurate quantification of morphological parameters are realized, the present application can detect synaptic morphology micro changes caused by aluminum exposure as low as 5 μg / L, and the measurement accuracy of synaptic neck width reaches nanometer level.

Description

Technical Field

[0001] This invention relates to the fields of neuroscience and biomedical image processing, and in particular to a micro-measurement device and method for morphological changes of neuronal synapses after aluminum exposure, for high-precision detection and analysis of morphological changes in the microstructure of neuronal synapses. Background Technology

[0002] Neuronal synapses are key structures for information transmission between neurons, and their morphological changes are closely related to nervous system function. Studies have shown that heavy metals such as aluminum in the environment may have toxic effects on neuronal synapses, leading to changes in synaptic structure and function, thereby affecting nerve signal transmission and potentially linked to the development of neurodegenerative diseases such as Alzheimer's disease.

[0003] In current technologies, the detection of morphological changes in neuronal synapses mainly relies on traditional microscopic imaging techniques and manual measurement methods. These methods have several limitations: firstly, traditional edge detection algorithms, such as the Canny algorithm, are insufficiently accurate when processing biological samples with complex three-dimensional structures, such as neuronal synapses, making it difficult to accurately identify synaptic edges; secondly, manual measurement methods are highly subjective, inefficient, and have poor repeatability, making them unsuitable for large-scale sample analysis. Furthermore, most existing morphological parameter quantification methods are based on simple geometric measurements and cannot comprehensively reflect the complex features and subtle changes in synaptic structures.

[0004] Therefore, developing a micro-measurement device and method capable of high-precision, automated detection and analysis of morphological changes in neuronal synapses after aluminum exposure is of significant scientific importance and application value. Summary of the Invention

[0005] The purpose of this invention is to provide a micro-measurement device and method for morphological changes of neuronal synapses after aluminum exposure. By introducing an innovative method that combines differential geometry theory with image processing technology, it achieves high-precision identification of neuronal synaptic edges and accurate quantification of morphological parameters.

[0006] This invention proposes a micro-measurement device for neuronal synaptic morphological changes after aluminum exposure, comprising:

[0007] The confocal microscopy imaging module is used for:

[0008] Acquire three-dimensional image data of neuronal synapses;

[0009] The three-dimensional image data is transmitted to the preprocessing module;

[0010] The preprocessing module, which is communicatively connected to the confocal microscopy imaging module, is used for:

[0011] Receive the three-dimensional image data sent by the confocal microscopy imaging module;

[0012] The three-dimensional image data is subjected to noise suppression and contrast enhancement to generate enhanced three-dimensional image data;

[0013] The enhanced 3D image data is transmitted to the edge detection module;

[0014] The edge detection module, which is communicatively connected to the preprocessing module, is used for:

[0015] Receive the enhanced 3D image data sent by the preprocessing module;

[0016] The enhanced 3D image data is converted into a continuous surface representation;

[0017] Based on the continuous surface representation, the principal curvature and principal direction of each point are calculated to generate a curvature map.

[0018] Based on the curvature map, identify the peak region of curvature change rate and extract edge candidate points;

[0019] Apply edge consistency constraints to the candidate edge points to generate an edge point set;

[0020] The edge point set is transmitted to the edge point classification module;

[0021] The edge point classification module, which is communicatively connected to the edge detection module, is used for:

[0022] Receive the set of edge points sent by the edge detection module;

[0023] Construct a multi-scale representation of the edge point set;

[0024] Perform curvature flow evolution analysis to track the curvature changes of each edge point at different scales;

[0025] Based on the curvature change, the edge point set is classified into a positive edge point set and a negative edge point set;

[0026] The positive edge point set and the negative edge point set are transmitted to the geodesic network construction module;

[0027] The geodesic network construction module, which is communicatively connected to the edge point classification module, is used for:

[0028] Receive the positive edge point set and the negative edge point set sent by the edge point classification module;

[0029] Establish the connection relationship between the points in the positive edge point set and the negative edge point set;

[0030] Calculate the geodesic distance between edge points;

[0031] Based on the aforementioned geodesic distances, a geodesic network is constructed;

[0032] The geodesic network is transmitted to the morphological parameter quantization module;

[0033] The morphological parameter quantization module, which is communicatively connected to the geodesic network construction module, is used for:

[0034] Receive the geodesic network sent by the geodesic network construction module;

[0035] Based on the geodesic network, calculate the synaptic neck width, synaptic spine volume, and synaptic length parameters;

[0036] Based on the geodesic network, the parameters of synaptic regularity, synaptic uniformity, and synaptic heterogeneity are calculated.

[0037] Output the morphological parameters of the neuron synapse.

[0038] Preferably, the confocal microscopy imaging module includes:

[0039] A laser scanning device for generating laser beams with wavelengths of 405-635 nanometers;

[0040] A three-dimensional precision translation stage is used for precise positioning and scanning in the XYZ directions;

[0041] An imaging system used to receive reflected light signals and convert them into electrical signals;

[0042] And a data acquisition unit, used to receive the electrical signals and generate the three-dimensional image data.

[0043] Preferably, the edge detection module is further used for:

[0044] An adaptive curvature change threshold is determined, wherein the curvature change threshold is determined based on the overall distribution statistics of the curvature map;

[0045] Perform nonmaximum suppression to preserve points of maximum curvature change.

[0046] It also assesses edge continuity based on the principal curvature direction and supplements discontinuous edges through curvature interpolation.

[0047] Preferably, the edge point classification module is further used for:

[0048] Define a scale sequence from fine to coarse, where the scaling factor between adjacent scales is 1.5;

[0049] Perform surface smoothing operations at each scale to generate a multi-scale surface sequence;

[0050] Record the points where the curvature sign changes and the rate of change at each point;

[0051] And it identifies edge points where curvature changes from positive to negative as positive edge points, and edge points where curvature changes from negative to positive as negative edge points.

[0052] Preferably, the geodesic network construction module is further used for:

[0053] Construct Delaunay triangulations for the edge points;

[0054] Initial geodesic paths are generated based on the shortest path algorithm;

[0055] The initial geodesic path was smoothed using tension spline interpolation;

[0056] And to identify key nodes in the geodesic network, including intersections and endpoints.

[0057] Preferably, the morphological parameter quantization module is further used for:

[0058] The minimum geodesic distance between the positive and negative edge points is calculated as the synaptic neck width.

[0059] Calculate the volume integral of the region enclosed by the geodesics as the volume of the synaptic spine;

[0060] The variance of the principal direction of the geodesic is analyzed as the regularity of synaptic alignment;

[0061] And the degree of deviation from the standard morphology is calculated as synaptic heteromorphism.

[0062] As a preferred option, it also includes:

[0063] The data storage module is communicatively connected to the morphological parameter quantization module and is used to save the neuronal synaptic morphological parameter results and original image data.

[0064] The system also includes a results visualization module, which is communicatively connected to the morphological parameter quantification module and the data storage module, and is used to present the neuronal synaptic morphological parameter results in a graphical manner.

[0065] Preferably, the edge detection module is also used for:

[0066] The edge detection window size is set to a value between 3×3×3 and 7×7×7, wherein the edge detection window size is adaptively adjusted according to the image resolution;

[0067] And set the edge continuity threshold to 20% of the average curvature, wherein the edge continuity threshold can be adjusted according to the characteristics of the sample.

[0068] Preferably, the result visualization module is used for:

[0069] Generate a three-dimensional structural model of a neuronal synapse;

[0070] Use different colors to mark the positive and negative edge point sets;

[0071] Display the topology of the geodesic network;

[0072] And the statistical analysis results of synaptic morphological parameters are presented in chart form.

[0073] A micro-measurement method for neuronal synaptic morphological changes after aluminum exposure using the aforementioned device, characterized by comprising the following steps:

[0074] Acquire three-dimensional image data of neuronal synapses;

[0075] The three-dimensional image data is preprocessed to generate enhanced three-dimensional image data;

[0076] The enhanced 3D image data is converted into a continuous surface representation;

[0077] Based on the continuous surface representation, the curvature map is calculated and edge candidate points are extracted;

[0078] Apply edge consistency constraints to the candidate edge points to generate an edge point set;

[0079] Construct a multi-scale representation of the edge point set and perform curvature flow evolution analysis;

[0080] Based on curvature changes, the edge point set is classified into positive edge point set and negative edge point set;

[0081] Establish the connection relationship between the points in the positive edge point set and the negative edge point set;

[0082] Calculate the geodesic distances between edge points and construct a geodesic network;

[0083] Based on the geodesic network, the morphological parameters of neuronal synapses are calculated and output.

[0084] This invention overcomes the limitations of traditional techniques in processing the microstructure of neuronal synapses by applying edge detection technology based on curvature analysis, edge point classification method based on manifold hierarchical decomposition, and morphological parameter quantification method for geodesic networks, and has the following beneficial effects:

[0085] 1. Significantly improved edge detection accuracy: Compared with traditional methods, the edge detection accuracy of this invention is improved by about 45%, which can accurately identify the edge structure of neuronal synapses;

[0086] 2. Significantly improved detection sensitivity: This invention can detect minute changes in synaptic morphology caused by exposure to aluminum as low as 5 μg / L, with a sensitivity more than 5 times higher than existing technologies;

[0087] 3. Measurement accuracy reaches the nanometer level: the measurement error of synaptic neck width is controlled within ±5nm, and the volume measurement error is controlled within 5%;

[0088] 4. Significantly improved processing efficiency: The automated processing workflow reduces the analysis time from 4-6 hours using traditional methods to 15-30 minutes, improving efficiency by 8-12 times;

[0089] 5. High adaptability: The system automatically adjusts processing parameters according to sample characteristics, adapting to different types of samples and significantly reducing the professional technical threshold;

[0090] 6. High value for early diagnosis: Synaptic morphological changes can be detected 6-18 months before the onset of clinical symptoms of neurodegenerative diseases, providing a time window for early intervention. Attached Figure Description

[0091] Figure 1 This is a block diagram of the overall structure of the micro-measuring device for neuronal synaptic morphological changes after aluminum exposure according to the present invention.

[0092] Figure 2 This is a schematic diagram of the confocal microscopy imaging module of the present invention;

[0093] Figure 3 This is a flowchart of the edge detection module of the present invention;

[0094] Figure 4 This is a flowchart of the edge point classification module of the present invention;

[0095] Figure 5 This is a flowchart of the geodesic network construction module of the present invention;

[0096] Figure 6 This is a flowchart of the morphological parameter quantization module of the present invention.

[0097] Figure 7 This is a flowchart of the micro-measurement method of the present invention. Detailed Implementation

[0098] Please refer to Figure 1-7 The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The scope of protection of the present invention is not limited to the specific embodiments described below.

[0099] Reference Figure 1The micro-measurement device for morphological changes of neuronal synapses after aluminum exposure provided by the present invention includes: a confocal microscopy imaging module 1, a preprocessing module 2, an edge detection module 3, an edge point classification module 4, a geodesic network construction module 5, a morphological parameter quantification module 6, a data storage module 7, and a result visualization module 8.

[0100] Reference Figure 2 The confocal microscopy imaging module 1 is used to acquire three-dimensional image data of neuronal synapses and transmit the three-dimensional image data to the preprocessing module 2.

[0101] In a preferred embodiment of the present invention, the confocal microscopy imaging module 1 includes a laser scanning device 11, a three-dimensional precision translation stage 12, an imaging system 13, and a data acquisition unit 14. The laser scanning device 11 generates a laser beam with a wavelength of 405-635 nm. Preferably, different wavelengths of laser light can be selected according to the characteristics of the sample; for example, a 488 nm wavelength laser can be used to excite GFP fluorescent protein in fluorescently labeled neuronal samples. The three-dimensional precision translation stage 12 is used for precise positioning and scanning in the XYZ directions, where the XY range is 50 mm × 50 mm, the Z-axis travel is 20 mm, and the positioning accuracy is preferably 0.1 μm. The imaging system 13 receives reflected light signals and converts them into electrical signals, and includes a microscope objective, a beam splitter, and a photomultiplier tube. The data acquisition unit 14 receives electrical signals and generates three-dimensional image data, and includes an analog-to-digital converter and a data buffer module.

[0102] In practical applications, the operator first places the neuron sample in a culture dish and then on the three-dimensional precision translation stage 12. Scanning parameters are set via the system control interface, such as a Z-axis scanning range typically set to 10-50 μm, a scanning step size of 0.2-0.5 μm, and a resolution of 1024×1024 pixels for each XY plane layer. The laser beam emitted by the laser scanning device 11 is reflected by a beam splitter and focused onto the sample by a microscope objective. The fluorescence or reflected light emitted by the sample returns along the same optical path, passes through the beam splitter, is received by a photomultiplier tube, and is converted into an electrical signal. Finally, the data acquisition unit 14 generates a three-dimensional image dataset.

[0103] Different objectives can be selected for different types of neuron samples. Generally, a 20× or 40× objective can be used to observe the neuron cell body, while a 60× or 100× oil immersion lens is preferred for observing the fine structure of synapses to obtain sufficient resolution.

[0104] The preprocessing module 2 is communicatively connected to the confocal microscopy imaging module 1. It is used to receive the three-dimensional image data sent by the confocal microscopy imaging module 1, perform noise suppression and contrast enhancement on the three-dimensional image data, generate enhanced three-dimensional image data, and transmit the enhanced three-dimensional image data to the edge detection module 3.

[0105] In a specific embodiment of the present invention, the main processing steps performed by the preprocessing module 2 include: First, applying an anisotropic diffusion filtering algorithm to denoise the original three-dimensional image data. This algorithm can effectively suppress noise while preserving edge features. Preferably, the diffusion coefficient can be set to 0.1-0.3, and the number of iterations is 10-20. Second, enhancing image contrast through an adaptive histogram equalization algorithm, wherein the histogram equalization window size is preferably set to 16×16×16 pixels, and the cropping limit threshold is set to 0.01-0.05. Finally, performing background suppression operation, setting the background area to zero by setting a grayscale threshold (usually 5%-10% of the total grayscale range), thereby highlighting the neuronal synaptic structure.

[0106] After the above preprocessing steps, noise in the original 3D image is effectively suppressed, and the contrast of neuronal synaptic structures is significantly improved, providing a good data foundation for subsequent edge detection. Practice shows that appropriate preprocessing can improve the signal-to-noise ratio by 2-3 times, significantly contributing to the improvement of edge detection accuracy.

[0107] The edge detection module 3 is communicatively connected to the preprocessing module 2. It receives the enhanced 3D image data sent by the preprocessing module 2, converts the enhanced 3D image data into a continuous surface representation, calculates the principal curvature and principal direction of each point based on the continuous surface representation, generates a curvature map, identifies the peak region of curvature change rate based on the curvature map, extracts edge candidate points, applies edge consistency constraints to the edge candidate points, generates an edge point set, and transmits the edge point set to the edge point classification module 4.

[0108] Reference Figure 3 The processing flow of edge detection module 3 includes the following steps:

[0109] First, the enhanced 3D image data is converted into a continuous surface representation. Specifically, this is achieved by establishing a continuous functional relationship between discrete voxel data points using a cubic spline interpolation algorithm, thus forming a parametric surface representation. ,in The coordinates are in the parameter domain.

[0110] in: For parametric surface representation functions, represent coordinates in the parametric domain. Points on the curved surface at that location; , and These are the three coordinate axes of the parameter domain, corresponding to the three dimensions of the physical space.

[0111] For typical neuronal synapse data, the resolution of the parameter domain is usually set to 2-4 times the resolution of the original image to ensure sufficient interpolation accuracy.

[0112] Secondly, based on the continuous surface representation, the principal curvature and principal direction of each point are calculated to generate a curvature map. In this invention, shape operators from differential geometry are used to calculate the principal curvature of each point on the surface. and and the corresponding main direction and :

[0113] ,

[0114] ,

[0115] in: The mean curvature represents the average degree of curvature of the surface at that point. Let be the Gaussian curvature, representing the degree of intrinsic curvature of the surface at that point; and Principal curvature represents the degree of curvature of the surface at that point along the two principal directions; and For corresponding and The principal direction unit vectors are mutually orthogonal.

[0116] The calculation of principal curvature involves solving for the first and second fundamental forms of the surface. In the case of discrete data, it is approximated by fitting a local quadratic surface. Specifically, for each point on the surface, a quadratic surface is fitted by taking its 3×3×3 to 7×7×7 neighborhood (adaptively adjusted according to image resolution):

[0117] ,

[0118] Where: z is the height of the surface in the local coordinate system; x and y are the planar coordinates in the local coordinate system; a, b, c, d, e, and f are the fitting parameters of the quadratic surface, which are obtained by the least squares method.

[0119] The parameters a, b, c, d, e, and f are solved using the least squares method, and then the principal curvatures of the surface are calculated based on these parameters. Specifically, the principal curvatures can be calculated using the following formula:

[0120] ,

[0121] ,

[0122] in: The maximum principal curvature; is the minimum principal curvature; a, b, and c are the fitting parameters for the quadratic surface.

[0123] Third, based on the curvature map, the rate of curvature change is calculated, peak regions of the rate of curvature change are identified, and candidate edge points are extracted. (Rate of curvature change) The calculation is as follows:

[0124] ,

[0125] in: The rate of change of curvature represents the degree of drastic change in the curvature of the surface; Let be the gradient vector of the mean curvature, representing the mean curvature. The rate of change in space; Let be the gradient vector of Gaussian curvature, representing the Gaussian curvature. The rate of change in space; These are weighting coefficients used to balance the average curvature gradient vector. With the gradient vector of Gaussian curvature The contribution typically ranges from 0.1 to 0.5.

[0126] Mean curvature gradient vector and Gaussian curvature gradient vector The calculations are as follows:

[0127] ,

[0128] ,

[0129] in: For the mean curvature, Gaussian curvature; , , They are the mean curvatures For parameters , , The partial derivatives; , , Gaussian curvature For parameters , , The partial derivatives of .

[0130] Edge candidate point extraction employs an adaptive thresholding method, with the threshold value set at 100%. Determined based on the overall distribution statistics of the rate of change of curvature:

[0131] ,

[0132] in: The rate of change of curvature threshold is used to determine whether a point is a candidate edge point; Rate of change of curvature The mean value represents the average level of curvature variation across the entire surface; Rate of change of curvature The standard deviation represents the degree of dispersion of the curvature variation; is a coefficient used to control the strictness of the threshold, typically set to 2.0-3.0. When When the value is large, there will be fewer but more reliable edge candidate points; when When the value is small, it will contain more edge candidate points but may introduce noise.

[0133] The rate of change of curvature is greater than the threshold. The points were identified as edge candidate points:

[0134] ,

[0135] in: For the set of candidate edge points; Let be a point on the curved surface; For point The rate of change of curvature at that point.

[0136] Fourth, apply non-maximum suppression (NMS) to edge candidate points, preserving points with local maximum curvature changes. In this step, each edge candidate point is examined along the curvature gradient direction; if the rate of curvature change at that point is not a local maximum, it is excluded. The NMS radius is typically set to 1.5 pixels. NMS can be expressed as:

[0137] ,

[0138] in: The set of edge points after nonmaximum suppression; Candidate points for the edge; For Centered on, with radius ; This is the non-maximum suppression radius, typically set to 1.5 pixel units; and Points and The rate of change of curvature at that point.

[0139] Finally, edge consistency constraints are applied to the candidate edge points to generate the final set of edge points. The edge consistency constraint is based on the principal curvature direction and determines the integrity of the edge by evaluating the continuity of the curvature direction between adjacent edge points. The edge continuity threshold is set to 20% of the average curvature, and any discontinuous edges that may exist are supplemented through curvature interpolation. The edge consistency constraint can be expressed as:

[0140] ,

[0141] in: For the final set of edge points; and For edge points; For point and The Euclidean distance between them; The connection radius is typically set to 2 to 3 pixels. For point and The angle between the principal curvature directions; The directional consistency threshold is typically set between 30° and 45°.

[0142] The edge detection module 3 also adaptively determines the curvature change threshold, which is based on the overall distribution statistics of the curvature map. Curvature distributions may differ in different types of neuron samples; therefore, using an adaptive threshold method can improve the adaptability and accuracy of edge detection.

[0143] Practice has shown that, when dealing with biological structures with complex geometric characteristics, such as neuronal synapses, edge detection methods based on differential geometry offer higher detection accuracy and better noise resistance compared to traditional gradient-based methods like the Canny algorithm. Experimental results show that, in typical neuronal synapse samples, the edge detection accuracy of this invention can reach 90%–95%, while traditional methods typically only achieve 55%–65%.

[0144] The edge point classification module 4 is communicatively connected to the edge detection module 3. It is used to receive the edge point set sent by the edge detection module 3, construct a multi-scale representation of the edge point set, perform curvature flow evolution analysis, track the curvature change of each edge point at different scales, classify the edge point set into positive edge point set and negative edge point set based on the curvature change, and transmit the positive edge point set and negative edge point set to the geodesic network construction module 5.

[0145] Reference Figure 4 The processing flow of edge point classification module 4 includes the following steps:

[0146] First, a multi-scale representation of the edge point set is constructed. In this invention, a scale sequence from fine to coarse is defined, where the scaling factor between adjacent scales is 1.5. A typical five-level scale sequence is {1.0, 1.5, 2.25, 3.375, 5.0625}. At each scale, the surface is smoothed using Gaussian convolution.

[0147] ,

[0148] in: For scale The surface below is represented by a scale of . The smoothed surface; Represented as the original surface; The standard deviation is A three-dimensional Gaussian function is used for smoothing. Represents the convolution operation. Three-dimensional Gaussian function. Defined as:

[0149] ,

[0150] in: The scaling parameter controls the width of the Gaussian function; a larger value indicates a wider width. The value corresponds to a coarser scale.

[0151] Secondly, curvature flow evolution analysis is performed to track the curvature changes at each edge point at different scales. The curvature flow evolution uses the following partial differential equation:

[0152] ,

[0153] in: It represents the rate of change of the surface over time (scale), and is a vector field; The mean curvature is a scalar field representing the average degree of curvature of the surface at each point; The unit normal vector is a unit vector field perpendicular to the tangent plane of the surface.

[0154] This equation describes the evolution of a surface under the influence of mean curvature flow. Numerical methods can be used to calculate the position and curvature value of each edge point at different scales. 15–30 iterations are performed at each scale to ensure sufficient evolutionary effect. In practical calculations, an explicit Euler method is used for discretization.

[0155] ,

[0156] in: and They represent the first Step and the first The curved surface of the step; This is the time step, typically set to 0.01-0.05; and The first The average curvature and unit normal vector of the step.

[0157] Third, record the points where the curvature sign changes and the rate of change for each point. For each edge point, track its curvature value change trajectory in scale space, paying particular attention to points where the curvature sign changes and locations where the rate of change exceeds a threshold (usually set to 40%). (Curvature change rate) Defined as:

[0158] ,

[0159] in: For point From scale To scale The rate of change of curvature; and Points In scale and The average curvature under the given conditions.

[0160] Finally, based on curvature changes, edge points are classified into positive and negative edge points. The classification rules are as follows:

[0161] Positive edge point: an edge point where the curvature changes from positive to negative, usually corresponding to the edge of a synaptic protrusion;

[0162] Negative edge point: an edge point where the curvature changes from negative to positive, usually corresponding to the edge of a synaptic depression;

[0163] Neutral edge point: A point where the curvature remains the same sign but changes significantly, usually used as an auxiliary reference point.

[0164] Formal representation:

[0165] ,

[0166] ,

[0167] ,

[0168] in: , and These are the positive edge point set, the negative edge point set, and the neutral edge point set, respectively. For edge points; and For different scale parameters; For point In scale The average curvature under these conditions; This is a sign function that returns the sign of the argument. For point From scale To scale The rate of change of curvature; This is the threshold for the rate of change of curvature, which is usually set to 0.4.

[0169] Practice has shown that the edge point classification method based on manifold hierarchical decomposition can accurately identify the convex and concave regions of neuronal synaptic structures, providing a reliable foundation for subsequent morphological parameter calculations. This method is particularly suitable for distinguishing between the synaptic neck and head regions, which is of great significance for assessing the impact of aluminum exposure on synaptic morphology.

[0170] In one embodiment of the present invention, analysis was performed on hippocampal neuronal samples 24 hours after exposure to aluminum (10 μg / L). Positive marginal points were mainly distributed along the outline of the synaptic spines, while negative marginal points were mainly distributed in the synaptic neck region, which is highly consistent with the anatomical features of the synapse. The classification accuracy was verified by professionals to be over 92%, far exceeding the 60%–70% of traditional methods.

[0171] The geodesic network construction module 5 is communicatively connected to the edge point classification module 4. It is used to receive the positive edge point set and the negative edge point set sent by the edge point classification module 4, establish the connection relationship between each point in the positive edge point set and the negative edge point set, calculate the geodesic distance between edge points, construct the geodesic network based on the geodesic distance, and transmit the geodesic network to the morphological parameter quantization module 6.

[0172] Reference Figure 5 The processing flow of the geodesic network construction module 5 includes the following steps:

[0173] First, Delaunay triangulation of the edge points is constructed. Triangulation is performed on both the positive and negative edge point sets to generate an initial connectivity graph. Delaunay triangulation maximizes the minimum angle of the triangles, contributing to good mesh quality. In this embodiment, the triangulation is implemented using an incremental algorithm with a computational complexity of O(nlogn), where n is the number of edge points.

[0174] Secondly, the geodesic distance between edge points is calculated. Geodesic distance refers to the distance along the shortest path on the surface, more accurately reflecting the actual distance on the synaptic surface. This invention uses an improved FastMarching Method to calculate the geodesic distance:

[0175] ,

[0176] Where: T(x) is the arrival time from the starting point to point x, which is a scalar field; F(x) is the velocity function at point x, usually set to a constant 1, indicating that the wavefront propagates at the same speed in all directions; |T(x)| represents the gradient norm of the arrival time field, which is a scalar field representing the rate of change of arrival time in space.

[0177] This equation is an Eikonal equation, describing the propagation process of the wavefront in the medium. By solving this equation, the geodesic distance from any edge point to other points can be obtained. On a discrete grid, the following difference scheme is used for approximate solution:

[0178] ,

[0179] in: , Equal to the forward and backward differences along each coordinate axis at the point (i,j,k); Let be the velocity function value at point (i,j,k).

[0180] In practical calculations, a priority queue management approach is used, starting from the initial point and gradually updating the arrival times of surrounding points until the entire area is covered. Geodesic distance It can be calculated directly from the arrival time:

[0181] ,

[0182] in: Let be the geodesic distance from point p to point q; Let p be the arrival time of point q.

[0183] Third, an initial geodesic path is generated based on the shortest path algorithm. After obtaining the geodesic distance field, the shortest path connecting any two points, i.e., the geodesic, can be obtained by tracing along the reverse direction of the gradient. This invention uses the gradient descent method to calculate the geodesic path:

[0184] ,

[0185] in: The parametric geodesic path is a three-dimensional curve. This is a path parameter, and its value range is typically [value range missing]. ; The gradient of the arrival time field along the path is a vector field; Let be the gradient norm, which is a scalar field.

[0186] In the discrete implementation, the geodesic path is calculated iteratively:

[0187] ,

[0188] in: and The first Step and Path points of the steps; This is the step size parameter, typically set to 0.01-0.05. The iteration starts from the endpoint and traces in the reverse direction of the gradient until the starting point is reached.

[0189] Fourth, use tension spline interpolation to smooth the initial geodesic path. The initial path may contain localized unsmooth areas; tension spline interpolation can yield a smoother geodesic.

[0190] ,

[0191] in: The smoothed geodesic path is a three-dimensional curve. For parameters, the value range is usually 100%. ; Control points are key points on the path, and are usually selected evenly from the original path. As a basis function, it defines how the control points affect the shape of the curve; The number of control points.

[0192] basis functions of tension splines It can be represented as:

[0193] ,

[0194] in: It is a step function; Control points The corresponding parameter values. The tension parameter is usually set to 0.5-0.7 to achieve a smooth effect while keeping the path close to the original geodesic.

[0195] Finally, key nodes in the geodesic network are identified, including intersections and endpoints. Intersections are nodes connecting multiple geodesics, typically corresponding to critical locations in synaptic structures; endpoints are the termination points of geodesics, typically corresponding to synaptic boundaries. Key nodes are identified using connectivity analysis: nodes with a connectivity greater than 2 are identified as intersections, and nodes with a connectivity of 1 are identified as endpoints. Formalized as:

[0196] ,

[0197] ,

[0198] in: and These are the intersection set and the endpoint set, respectively. For all nodes in the network; For nodes The connectivity is the number of edges connected to that node.

[0199] Through the above steps, this invention constructs a representation of a neuronal synaptic network based on geodesics. This representation preserves the topological relationships and geometric characteristics of the synaptic structure, providing a solid foundation for subsequent morphological parameter quantification.

[0200] In one specific embodiment of the present invention, geodesic networks were constructed from normal and aluminum-exposed (20 μg / L, 48 hours) neuronal samples. The results showed that the topology of the geodesic network in the aluminum-exposed group was significantly simplified, the number of crossover points was reduced by about 30%, and the average geodesic distance between endpoints increased by about 15%. This reflects the reduction in synaptic complexity and connection reduction caused by aluminum exposure.

[0201] The morphological parameter quantification module 6 is communicatively connected to the geodesic network construction module 5. It is used to receive the geodesic network sent by the geodesic network construction module 5, calculate basic morphological parameters such as synaptic neck width, synaptic spine volume and synaptic length based on the geodesic network, as well as complex morphological parameters such as synaptic arrangement regularity, synaptic uniformity and synaptic heterogeneity, and output the neuronal synaptic morphological parameter results.

[0202] Reference Figure 6 The processing flow of the morphological parameter quantization module 6 includes the following steps:

[0203] First, the basic morphological parameters are calculated. The synaptic neck width is a key indicator for assessing the impact of aluminum exposure. This invention obtains this parameter by calculating the minimum geodesic distance between the positive and negative edge points:

[0204] ,

[0205] in: The width of the synaptic neck represents the minimum diameter of the synaptic neck. Positive edge point To the negative edge point The geodetic distance; This means taking all possible values. The minimum value in the pair; Points within the positive edge point set; These are points within the negative edge point set.

[0206] The volume of the synaptic spine is calculated by volume integral over the region enclosed by the geodesic:

[0207] ,

[0208] in: The volume of the synaptic spine represents the size of the synaptic spine structure. The area enclosed by the geodesics represents the spatial extent of the synaptic spine; Let be a volume element, representing a tiny volume unit in the integration process.

[0209] In practical calculations, the integral is approximated by dividing the region into small voxels and summing them:

[0210] ,

[0211] in: The volume of a single voxel is usually determined by the image resolution; The coordinates are voxel coordinates; the summation range is the area enclosed by the geodesics. All voxels within.

[0212] Synaptic length is determined by the maximum geodesic distance along the principal axis:

[0213] ,

[0214] in: The synaptic length represents the maximum dimension of the synaptic structure along the principal axis. Positive edge point arrive The geodetic distance; The angle between the geodesic and the principal axis is usually determined by principal component analysis; This means taking all possible values. The maximum value in the pair; and These are points in the positive edge point set.

[0215] Secondly, the parameters of complex morphology are calculated. The regularity of synaptic alignment is obtained through variance analysis of the principal directions of the geodesics:

[0216] ,

[0217] in: The regularity of synaptic arrangement, with a value range of [value range missing]. A larger value indicates a more regular arrangement; The variance of the geodesic direction angle is calculated as follows: ,in For the first The direction angle of the geodesic line, The average direction angle, This represents the total number of geodesics. This represents the theoretical maximum variance, corresponding to a completely random distribution, and is typically taken as a value of [value missing]. This is evenly distributed in The variance of a random variable over an interval.

[0218] Synaptic homogeneity is assessed by the homogeneity of the geodesic distance distribution:

[0219] ,

[0220] in: For synaptic uniformity, the value range is: A larger value indicates a more uniform distribution. The standard deviation of the geodetic distance is calculated as follows: ,in For the first Geodetic distance between points Total number of points; The mean of the geodetic distance is calculated as follows: .

[0221] Synaptic heteromorphism is calculated by the degree of deviation from the standard morphology:

[0222] ,

[0223] in: This represents synaptic heterogeneity and can be positive or negative. The shape angle is usually determined by the angle between two principal geodesics, and the unit is degrees; 45° is the reference angle, representing the standard shape. When When, it indicates that the synaptic structure is becoming heterogeneous towards a square structure; when When, it indicates a shift towards a circular structure; The larger the value, the higher the degree of heterogeneity.

[0224] Finally, the morphological parameter quantification module 6 organizes the calculated morphological parameters into structured data and outputs the neuronal synaptic morphological parameter results. These results can be directly used to assess the impact of aluminum exposure on neuronal synaptic morphology, and can also serve as indicators for the early diagnosis of neurodegenerative diseases.

[0225] In a specific application of this invention, by comparing and analyzing the synaptic morphology parameters of the normal group and the groups exposed to different concentrations of aluminum (5 μg / L, 10 μg / L, 20 μg / L), it was found that as the aluminum concentration increased, the synaptic neck width decreased significantly (up to 35%), the synaptic spine volume decreased significantly (up to 42%), and the synaptic atypia shifted to a negative value (indicating that the structure tended to become square). These changes are closely related to neuronal dysfunction and provide important clues for understanding the mechanism of aluminum neurotoxicity.

[0226] The present invention also includes a data storage module 7 and a result visualization module 8. The data storage module 7 is communicatively connected to the morphological parameter quantification module 6 and is used to store the neuronal synaptic morphological parameter results and raw image data. The result visualization module 8 is communicatively connected to the morphological parameter quantification module 6 and the data storage module 7 and is used to present the neuronal synaptic morphological parameter results in a graphical manner.

[0227] Data storage module 7 employs a hierarchical storage structure, managing raw data, intermediate processing results, and final parameter results in a tiered manner. Data is stored in standard formats (such as HDF5 or TIFF+XML) to ensure traceability and reusability. The storage module also provides data indexing and retrieval functions, allowing users to quickly locate and access historical data.

[0228] The results visualization module 8 offers multiple visualization methods, including generating three-dimensional structural models of neuronal synapses, using different colors to mark positive and negative edge point sets, displaying the topological structure of geodesic networks, and presenting statistical analysis results of synaptic morphological parameters in chart form. Furthermore, this module supports comparative displays of multiple sets of data, allowing researchers to intuitively compare changes in synaptic morphology under different treatment conditions.

[0229] In one specific embodiment, in the 3D view generated by the results visualization module 8, positive edge points are represented in red, negative edge points in blue, and geodesics in green. Through interactive operations such as rotation, scaling, and slicing, users can observe the synaptic structure and measurement results from different angles. For parameter statistical results, bar charts, scatter plots, and heatmaps are used for visualization, and significance analysis results are provided to help researchers quickly identify statistically significant changes.

[0230] Reference Figure 7 The present invention also provides a method for micro-measuring morphological changes of neuronal synapses after aluminum exposure. This method is implemented using the above-mentioned device and includes the following steps:

[0231] Step S1: Acquire three-dimensional image data of neuronal synapses;

[0232] Step S2: Preprocess the 3D image data to generate enhanced 3D image data;

[0233] Step S3: Convert the enhanced 3D image data into a continuous surface representation;

[0234] Step S4: Based on the continuous surface representation, calculate the curvature map and extract edge candidate points;

[0235] Step S5: Apply edge consistency constraints to the edge candidate points to generate an edge point set;

[0236] Step S6: Construct a multi-scale representation of the edge point set and perform curvature flow evolution analysis;

[0237] Step S7: Based on curvature changes, classify the edge point set into positive edge point set and negative edge point set;

[0238] Step S8: Establish the connection relationship between the points in the positive edge point set and the negative edge point set;

[0239] Step S9: Calculate the geodesic distances between edge points and construct a geodesic network;

[0240] Step S10: Based on the geodesic network, calculate and output the neuronal synaptic morphological parameters.

[0241] In specific implementation, in step S1, a confocal microscopy imaging module is used to acquire three-dimensional image data of neuronal synapses. Preferably, a 60× or 100× oil immersion microscope is used for imaging, with the Z-axis scanning step size set to 0.2-0.5 μm to obtain three-dimensional data with sufficient resolution. For fluorescently labeled samples, a laser matching the excitation wavelength of the fluorescent protein can be selected (e.g., 488 nm for GFP).

[0242] In step S2, anisotropic diffusion filtering is applied to the 3D image data for noise reduction, and adaptive histogram equalization is used to enhance contrast. The preferred noise reduction parameters are: diffusion coefficient 0.1-0.3, iteration count 10-20; the contrast enhancement parameters are: histogram equalization window size 16×16×16 pixels, cropping threshold 0.01-0.05.

[0243] Steps S3 to S5 correspond to the processing flow of the edge detection module. They convert the enhanced 3D image data into a continuous surface representation, calculate the curvature map, extract candidate edge points, and apply edge consistency constraints to generate an edge point set. A key innovation of this method is the use of differential geometry curvature analysis to identify edges, which offers higher accuracy and noise resistance compared to traditional gradient-based methods.

[0244] Steps S6 and S7 correspond to the processing flow of the edge point classification module, which constructs a multi-scale representation, performs curvature flow evolution analysis, and classifies edge points into positive and negative edge points based on curvature changes. This classification is crucial for accurately identifying different structural regions of the synapse (such as the neck and head).

[0245] Steps S8 to S9 correspond to the processing flow of the geodesic network construction module, establishing the connection relationships between edge points, calculating geodesic distances, and constructing the geodesic network. The geodesic network preserves the topological relationships of the synaptic structure, providing a foundation for the accurate calculation of morphological parameters.

[0246] Step S10 corresponds to the processing flow of the morphological parameter quantification module. It calculates various morphological parameters of the synapse based on the geodesic network and outputs the results. These parameters include basic morphological parameters (such as neck width and volume) and complex morphological parameters (such as arrangement regularity and irregularity), which together constitute a comprehensive description of the synaptic morphology.

[0247] In one practical application embodiment of the present invention, this method was used to study the effects of different concentrations of aluminum exposure (0, 5, 10, 20 μg / L) on the synaptic morphology of cultured hippocampal neurons. Experimental results showed that this method could detect minute changes in synaptic morphology induced by aluminum exposure as low as 5 μg / L, including a decrease in neck width (approximately 15%) and an increase in atypia (approximately 20%). These changes are difficult to detect reliably using conventional methods, demonstrating the high sensitivity and practical value of the method of the present invention.

[0248] In addition, this method has been applied to the study of Alzheimer's disease model mice. The results showed that synaptic morphological parameters had changed significantly 6-18 months before the onset of clinical symptoms, especially the decrease in synaptic spine density and the increase in atypia, providing new biomarkers for the early diagnosis of the disease.

[0249] As can be seen from the above embodiments, the micro-measurement device and method for morphological changes of neuronal synapses after aluminum exposure provided by the present invention, through the introduction of an innovative method combining differential geometry theory and image processing technology, achieves high-precision identification of neuronal synaptic edges and accurate quantification of morphological parameters, providing strong technical support for studying the effects of aluminum exposure on the nervous system and the early diagnosis of neurodegenerative diseases.

[0250] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A device for micro-measuring synaptic morphological changes in neurons after aluminum exposure, characterized by, The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device.

2. The apparatus of claim 1, wherein, The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device.

3. The apparatus of claim 1, wherein, The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device.

4. The apparatus of claim 1, wherein, The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse morphological parameter quantification method and device. The application relates to a neuron synapse A scale sequence is set from fine scale to coarse scale, with a scale ratio of 1.5 between adjacent scales; A surface smoothing operation is performed at each scale to generate a multi-scale surface sequence; The curvature sign change points and change rates of each point are recorded; And the edge point where the curvature changes from positive to negative is identified as a positive edge point, and the edge point where the curvature changes from negative to positive is identified as a negative edge point.

5. The apparatus of claim 1, wherein, The geodesic network construction module is further used to: Construct a Delaunay triangulation of the edge points; Generate an initial geodesic path based on the shortest path algorithm; Smooth the initial geodesic path using tension spline interpolation; And identify key nodes in the geodesic network, including intersection points and end points.

6. The apparatus of claim 1, wherein, The morphological parameter quantification module is further used to: Calculate the minimum geodesic distance between positive and negative edge points as the synaptic neck width; Calculate the volume integral of the region enclosed by the geodesic line as the synaptic spine volume; Analyze the variance of the main direction of the geodesic line as the synaptic arrangement regularity; And calculate the degree of deviation from the standard morphology as the synaptic heterotype.

7. The apparatus of claim 1, wherein, Further comprising: A data storage module, in communication with the morphological parameter quantification module, for saving the results of the morphological parameters of the neuron synapses and the original image data; And a result visualization module, in communication with the morphological parameter quantification module and the data storage module, for presenting the results of the morphological parameters of the neuron synapses in a graphical manner.

8. The apparatus of claim 1, wherein, The edge detection module is further used to: Set the edge detection window size to a value between 3x3x3 and 7x7x7, where the edge detection window size is adaptively adjusted according to the image resolution; And set the edge continuity threshold to 20% of the average curvature, where the edge continuity threshold can be adjusted according to the characteristics of the sample.

9. The apparatus of claim 7, wherein, The result visualization module is used to: Generate a three-dimensional structure model of the neuron synapses; Use different colors to mark the positive edge point set and the negative edge point set; Display the topological structure of the geodesic network; And display the statistical analysis results of the synaptic morphological parameters in chart form.

10. A method for micro-measuring the synaptic morphological changes in neurons after aluminium exposure, using a device as claimed in any one of claims 1 to 9, characterised in that, Comprising the following steps: Collecting three-dimensional image data of neuron synapses; Preprocessing the three-dimensional image data to generate enhanced three-dimensional image data; Converting the enhanced three-dimensional image data into a continuous surface representation; Based on the continuous surface representation, calculating the curvature atlas and extracting edge candidate points; Applying edge consistency constraints to the edge candidate points to generate an edge point set; Constructing a multi-scale representation of the edge point set and performing curvature flow evolution analysis; Based on the curvature change, classifying the edge point set into a positive edge point set and a negative edge point set; Establishing the connection relationship between each point in the positive edge point set and the negative edge point set; Calculating the geodesic distance between edge points to construct a geodesic network; And based on the geodesic network, calculating and outputting the results of the morphological parameters of the neuron synapses.

Images