Method and device for multi-scale morphological measurement of hippocampus of human brain based on clinical MRI image

Abstract

The application discloses a kind of based on clinical MRI image human brain hippocampus multiscale morphological measurement method and equipment, comprising: obtaining image sequence;Based on skeleton description method, baseline hippocampus corresponding skeleton representation model is constructed;Subregion label is added in baseline skeleton representation model, and multiscale skeleton representation model is obtained;According to the boundary surface of baseline hippocampus and the hippocampus of each time point, the multiscale skeleton representation model of hippocampus in the magnetic resonance image of each time point is obtained;The morphological measurement of each time point hippocampus of test object is carried out, and the morphological parameter of each time point hippocampus is obtained.The present application constructs the multiscale skeleton representation model of different time points of test object by using skeleton representation method, so as to provide the basis for accurately measuring the time characteristics of hippocampus morphology, solve the problem that the morphological difference between different individual hippocampus is large, and the morphological change in individual is small, improve the precision of hippocampus morphological change measurement.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and more specifically, to a method and device for multi-scale morphological measurement of the human brain hippocampus based on clinical MRI images. Background Technology

[0002] The hippocampus, also known as the hippocampal gyrus, hippocampal region, or cerebral hippocampus, is located between the thalamus and the medial temporal lobe of the brain and is part of the limbic system. Hippocampal atrophy is a typical structural change in the brain of many neurodegenerative diseases, such as Alzheimer's disease (AD), Parkinson's disease (PD), and amyotrophic lateral sclerosis (ALS). Therefore, hippocampal atrophy has become a structural imaging marker for studying early neurodegenerative diseases, and in clinical studies, it is usually quantified by a reduction in volume.

[0003] In related technologies, methods based on subregion volume typically use automatic segmentation techniques to obtain volume information of each subregion of the hippocampus in the subject, and determine the atrophy pattern of the hippocampus based on volume changes. Alternatively, surface-based methods can indirectly represent hippocampal atrophy through surface changes, providing quantitative indicators of localized atrophy.

[0004] The aforementioned sub-region volume-based analyses of hippocampal atrophy heavily rely on the accuracy of sub-region segmentation. Furthermore, since hippocampal sub-regions are elongated structures extending from anterior to posterior, sub-region volume analysis cannot provide a sufficiently precise description of localized atrophy. As for surface-based methods, hippocampal atrophy is essentially a three-dimensional anatomical change in thickness, width, and length. This makes the morphological parameters obtained from surface-based methods—such as curvature changes, spherical harmonic decomposition coefficients, Laplace transform decomposition coefficients, and the momentum and divergence of the surface transform field—clinically difficult to interpret and incomparable to histological findings. Consequently, it is challenging to apply existing consistent histological findings to clinical diagnosis. Summary of the Invention

[0005] The problem addressed by this invention is how to construct a hierarchical structure of the hippocampus to achieve hippocampal morphology measurement.

[0006] To address the aforementioned problems, in a first aspect, the present invention provides a method for multi-scale morphological measurement of the human hippocampus based on clinical MRI images, comprising:

[0007] An image sequence is acquired, which includes at least two magnetic resonance images of the same test object at different time points. Each magnetic resonance image includes segmentation information of the hippocampus, which is used to represent the boundary surface of the hippocampus in the magnetic resonance image.

[0008] Based on the skeleton description method, a skeleton representation model corresponding to the baseline hippocampus is constructed. The baseline hippocampus is the hippocampus in the earliest magnetic resonance image in the image sequence. The skeleton representation model includes a radial axis vector group, and the radial axis in the radial axis vector group represents the morphological structure of the baseline hippocampus.

[0009] By adding sub-region labels to the baseline skeleton representation model, a multi-scale skeleton representation model of the baseline hippocampus is obtained. The sub-region labels are used to indicate the location of each sub-region of the baseline hippocampus on the skeleton representation model.

[0010] Based on the boundary surface of the hippocampus in the baseline hippocampus and other time-point magnetic resonance images, the multi-scale skeleton representation model of the baseline hippocampus is transformed to obtain the multi-scale skeleton representation model of the hippocampus in the magnetic resonance images at other time points.

[0011] Based on the multi-scale skeleton representation model of the baseline hippocampus and the multi-scale skeleton representation model at other time points, morphological measurements of the hippocampus of the test subject at each time point were performed to obtain the morphological parameters of the hippocampus at each time point. These morphological parameters are used to characterize the temporal variation characteristics of the hippocampus morphology of the test subject.

[0012] Optionally, the multi-scale morphological measurement method of the human brain hippocampus based on clinical MRI images provided in this embodiment of the invention includes basic morphological parameters and trend morphological parameters. The basic morphological parameters include hippocampal length, overall thickness, thickness of each subregion, hippocampal width, and long axis curvature. The trend morphological parameters indicate the degree of hippocampal atrophy.

[0013] Optionally, the multi-scale morphological measurement method for the human hippocampus based on clinical MRI images provided in this embodiment of the invention includes an upper radial axis and a lower radial axis. The surface corresponding to the endpoint of the upper radial axis is the upper surface of the hippocampus, and the surface corresponding to the endpoint of the lower radial axis is the lower surface of the hippocampus. The longitudinal centerline of the multi-scale skeleton model is the long axis of the hippocampus, and the radial axis connected to the end of the long axis is the coronal radial axis.

[0014] Based on the multi-scale skeleton representation model of the baseline hippocampus and other multi-scale skeleton representation models at various time points, morphological measurements of the hippocampus of the test subject were performed at various time points, and the morphological parameters of the hippocampus at each time point were obtained, including:

[0015] The sum of the length of the major axis and the length of the coronal radial axis is determined as the length of the hippocampus;

[0016] The sum of the lengths of the upper radial axis and the lower radial axis of the hippocampus is determined as the overall thickness of the hippocampus;

[0017] The thickness of each subregion is determined as the average of the sum of the lengths of the upper and lower radial axes of the corresponding region, and is taken as the subregion thickness of the hippocampus.

[0018] Determine the curvature value of the major axis as the curvature of the hippocampus.

[0019] Optionally, the multi-scale morphological measurement method for the human brain hippocampus based on clinical MRI images provided in this embodiment of the invention, based on the multi-scale skeleton representation model of the baseline hippocampus and other multi-scale skeleton representation models at various time points, performs morphological measurements on the hippocampus of the test subject at various time points to obtain the morphological parameters of the hippocampus at each time point, including:

[0020] A two-dimensional space is generated, in which the first coordinate axis takes the value of each time point of the image sequence, and the second coordinate axis takes the value of each basic morphological parameter corresponding to each time point.

[0021] Linear regression was performed on the values ​​of the first and second coordinate axes to determine the corresponding slope, which is the trend morphology parameter of the hippocampus.

[0022] Optionally, the multi-scale morphological measurement method for the human hippocampus based on clinical MRI images provided in this embodiment of the invention, which is based on a skeleton description method, constructs a skeleton representation model corresponding to the baseline hippocampus, including:

[0023] Based on the skeleton representation method, an initial skeleton representation model of the baseline hippocampus is constructed.

[0024] Adjust the length and position of the axis in the initial skeleton representation model to obtain the adjusted initial skeleton representation model;

[0025] Based on the interpolation method, the mesh of the adjusted initial skeleton representation model is refined to obtain the refined initial skeleton representation model;

[0026] The refined initial skeleton representation model is optimized based on the central axis geometry to obtain the target skeleton representation model, which is the skeleton representation model of the baseline hippocampus.

[0027] Optionally, the multi-scale morphological measurement method for the human hippocampus based on clinical MRI images provided in this embodiment of the invention further includes a apical coronal-radial axis located at the end of the long axis, and adjusting the length and position of the radial axis in the initial skeletal representation model includes:

[0028] Adjust the length of the spoke axis so that the apex of the spoke axis is in contact with the boundary surface of the baseline hippocampus;

[0029] Move the tip of the coronal axis to the tail tip of the baseline hippocampus, so that the tip of the coronal axis points to the end of the baseline hippocampus.

[0030] Optionally, the multi-scale morphological measurement method for the human hippocampus based on clinical MRI images provided in this embodiment of the invention, which is based on an interpolation method, refines the mesh of the adjusted initial skeleton representation model to obtain a refined initial skeleton representation model, including:

[0031] The adjusted skeleton represents the face points and edge points corresponding to each initial mesh face in the mesh surface of the model.

[0032] Add edge points to the edges of the initial mesh surface to form a new mesh surface;

[0033] Determine the face points of the new grid surface;

[0034] Based on the face points and edge points of the new mesh surface, a subdivided mesh surface is formed;

[0035] Repeat the above steps of subdividing the initial mesh surface to form the subdivided mesh surface until the preset conditions are met, and obtain the target mesh surface. Each vertex in the target mesh surface is the spoke vertex after interpolation.

[0036] Based on each vertex in the target mesh surface, determine the new spoke axis after interpolation processing of the adjusted initial skeleton representation model, and obtain the refined initial skeleton representation model corresponding to the new spoke axis.

[0037] Optionally, the multi-scale morphological measurement method for the human hippocampus based on clinical MRI images provided in this embodiment of the invention optimizes the refined initial skeleton representation model according to the midline geometry to obtain the target skeleton representation model, including:

[0038] Adjust the vertex of the new spoke axis to the boundary surface of the baseline hippocampus, so that the implicit surface corresponding to the vertex of the new spoke axis coincides with the boundary surface of the baseline hippocampus;

[0039] Adjust the direction of the new spoke axis so that the angle between the new spoke axis and the boundary surface of the baseline hippocampus does not exceed a first preset threshold.

[0040] Adjust the lengths of the upper and lower spokes in the new spokes so that the difference between the lengths of the upper and lower spokes in the new spokes does not exceed a second preset threshold, and obtain the target skeleton representation model.

[0041] Optionally, the multi-scale morphological measurement method for the human brain hippocampus based on clinical MRI images provided in this embodiment of the invention, which involves converting the multi-scale skeleton representation model of the baseline hippocampus based on the boundary surface of the hippocampus in MRI images at other time points to obtain the multi-scale skeleton representation model of the hippocampus in MRI images at other time points, includes:

[0042] Determine the correspondence between the boundary surface of the baseline hippocampus and the boundary surfaces of the hippocampus at other time points;

[0043] Based on this correspondence, the boundary surface of the baseline hippocampus is smoothed to the boundary surface of the hippocampus at other time points, thus obtaining the deformation field of the hippocampus from the baseline hippocampus to the hippocampus at other time points.

[0044] The skeletal representation model of the baseline hippocampus is deformed based on the deformation field to obtain the skeletal representation model of the hippocampus at other time points.

[0045] In a second aspect, the present invention provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the multi-scale morphological measurement method of the human brain hippocampus based on clinical MRI images as described in the first aspect.

[0046] The present invention provides a method and computer device for multi-scale morphological measurement of the human hippocampus based on clinical MRI images. For the same test subject with MRI image sequences at different time points, firstly, based on skeleton representation, a baseline hippocampal skeleton representation model is constructed for the earliest time point corresponding to the MRI image. Then, corresponding subregion labels are added to the baseline hippocampal skeleton representation model to generate a multi-scale skeleton representation model of the baseline hippocampus. This multi-scale skeleton representation model can characterize the local characteristics of each subregion of the baseline hippocampus. Furthermore, the boundary surfaces of the hippocampus in the MRI images at all time points are used to transform the baseline hippocampal skeleton representation model, obtaining the hippocampal skeleton representation model corresponding to each time point. Finally, based on the constructed hippocampal skeleton representation models for each time point, the measurement of multi-scale hippocampal morphological changes, including the overall hippocampus, subregions, and local morphology, can be achieved.

[0047] This method utilizes skeletal representation to construct multi-scale skeletal representation models of test subjects at different time points, thereby providing a basis for accurately measuring the temporal changes in hippocampal morphology, improving the accuracy of hippocampal morphological change measurement, and obtaining morphological parameters that conform to histological interpretation, making them easier for clinical understanding. This provides a new approach for transforming hippocampal atrophy patterns discovered based on imaging into clinically usable disease analysis indicators. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of magnetic resonance images at multiple time points provided in some embodiments of the present invention;

[0049] Figure 2 This is a flowchart illustrating a multi-scale morphological measurement method for the human hippocampus based on clinical MRI images, provided in some embodiments of the present invention.

[0050] Figure 3 A schematic diagram of the flowchart of a multi-scale morphological measurement method for the human brain hippocampus based on clinical MRI images provided in some embodiments of the present invention;

[0051] Figure 4 This is a schematic diagram of the skeleton representation model provided in some embodiments of the present invention;

[0052] Figure 5 This is a flowchart illustrating a multi-scale morphological measurement method for the human hippocampus based on clinical MRI images, provided in some embodiments of the present invention.

[0053] Figure 6 This is a flowchart illustrating a multi-scale morphological measurement method for the human hippocampus based on clinical MRI images, provided in some embodiments of the present invention.

[0054] Figure 7 A schematic diagram of the structure of a multi-scale morphological measurement device for the human brain hippocampus based on clinical MRI images provided in some embodiments of the present invention;

[0055] Figure 8 This is a schematic diagram of the structure of a computer system provided in an embodiment of the present invention. Detailed Implementation

[0056] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0057] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0058] MRI technology can generate multimodal images that reflect the morphology of human organs and tissues, such as brain morphology, lung morphology, or bone morphology. Specifically, it can generate multimodal image data including T1-weighted imaging, T2-weighted imaging, T1ce imaging, and free water inhibition sequence (FLAIR).

[0059] For example, such as Figure 1As shown, this is brain imaging data of the same test subject at different time points acquired by MRI, with the highlighted part in the middle showing the outline of the hippocampus.

[0060] It is understandable that anatomical studies show that the axons of the main cells in the hippocampus are distributed in parallel, forming very thin bands, also known as "lamellae," that is, longitudinal interlayer connections. These lamellae are almost perpendicular to the long axis of the hippocampus. The information transmission patterns in the hippocampus correspond to different hippocampal subregions, such as the dentate gyrus (DG), the cornu ammonis (CA, specifically including CA1, CA2, and CA3), the subiculum (SUB), and the SRLM. In the head of the hippocampus, the hippocampal toes are distributed, and their uncinate gyri form vertical hooks.

[0061] It is also understandable that information about the changes in the shape and volume of the hippocampus in the brain over time can largely predict the risk of developing related diseases or infer the course of related diseases.

[0062] For example, hippocampal atrophy has become a structural imaging marker for diagnosing early neurodegenerative diseases, and in clinical studies it is usually quantified as a reduction in volume.

[0063] In practice, during the prodromal phase of neurodegenerative diseases such as Alzheimer's disease (AD), the overall volume of the hippocampus has low specificity for identifying individuals transitioning from mild cognitive impairment (MCI) to AD. This results in some potentially AD-prone patients going undetected, hindering early intervention for these patients. Evidence from pathological, histological, and high-field magnetic resonance imaging studies shows that the diffusion pattern of neurofibrillary tangles (NFTs) leads to selective neuronal degeneration in hippocampal subregions, macroscopically resulting in localized hippocampal atrophy. Similar patterns exist in other diseases. A major challenge in early intervention for neurodegenerative diseases is identifying early-affected localized hippocampal atrophy patterns to incorporate them into early structural biomarkers for the prediagnosis and prognosis of neurodegenerative diseases.

[0064] To accurately obtain hippocampal morphological information from medical images and comprehensively quantify local atrophy, related technologies employ methods based on sub-regional volume or surface. However, neither of these methods can provide a sufficiently accurate description of local atrophy during implementation.

[0065] Therefore, in this embodiment of the invention, to accurately characterize the hippocampal atrophy pattern over time in the test subjects, i.e., to analyze the longitudinal volume changes of the hippocampus, a multiscale skeletal representation (MS-rep) based method is used to describe the changes in hippocampal morphology over time using a global-to-local multiscale metric. Since MS-rep, through S-rep interpolation and surface subregion atlas projection, combined with the longitudinal MRI data of the test subjects, can capture the rich 3D intrinsic geometric properties of the hippocampus, it can more accurately characterize the hippocampal atrophy pattern and temporal characteristics of the test subjects. Furthermore, it has stronger clinical interpretability and statistical stability, and has important practical value for the early identification of neurodegenerative diseases.

[0066] To better understand the hippocampal morphology characterization method provided in the embodiments of the present invention, it will be described in detail below with reference to the accompanying drawings.

[0067] Figure 2 The diagram shown is a flowchart illustrating the hippocampal morphology measurement method according to an embodiment of the present invention. Figure 2 As shown, the method specifically includes:

[0068] S110, acquire an image sequence, which includes at least two magnetic resonance images of the same test object at different time points, each magnetic resonance image including hippocampal segmentation information, which is used to represent the boundary surface of the hippocampus in the magnetic resonance image.

[0069] S120, Based on the skeleton description method, construct the skeleton representation model corresponding to the baseline hippocampus. The baseline hippocampus is the hippocampus in the earliest time point of the magnetic resonance image in the image sequence. The skeleton representation model includes a radial axis vector group, and the radial axis in the radial axis vector group represents the morphological structure of the baseline hippocampus.

[0070] S130, add sub-region labels to the baseline skeleton representation model to obtain a multi-scale skeleton representation model of the baseline hippocampus. The sub-region labels are used to indicate the location of each sub-region of the baseline hippocampus in the skeleton representation model.

[0071] S140, based on the boundary surface of the hippocampus in the baseline hippocampus and other time-point magnetic resonance images, the multi-scale skeleton representation model of the baseline hippocampus is transformed to obtain the multi-scale skeleton representation model of the hippocampus in the other time-point magnetic resonance images.

[0072] S150, Based on the multi-scale skeleton representation model of the baseline hippocampus and the multi-scale skeleton representation model at other time points, morphological measurements of the hippocampus of the test subject at each time point are performed to obtain the morphological parameters of the hippocampus at each time point. These morphological parameters are used to characterize the temporal variation characteristics of the hippocampus morphology of the test subject.

[0073] Specifically, in combination Figure 3 The diagram illustrates the workflow framework for hippocampal measurement. In this embodiment of the invention, to accurately characterize the temporal changes in the volume and morphology of the hippocampus of the same test subject, and to accurately reflect the hippocampal atrophy, longitudinal analysis of the time-series magnetic resonance images is required. This involves first obtaining hippocampal edge images at multiple time points of a test subject, i.e., an image sequence. Each magnetic resonance image in this sequence includes hippocampal segmentation information to represent the location region of the hippocampus in the magnetic resonance image, such as... Figure 3 (a) The area enclosed in the dashed box is the hippocampus.

[0074] Each magnetic resonance image in this image sequence is a three-dimensional image, where the smallest unit is called a voxel. The segmentation information includes labels representing the region of the hippocampus added after segmenting the original magnetic resonance image.

[0075] The segmentation operation of the original MRI image involves assigning a label value to each voxel, with regions having the same label value belonging to the same brain region. The segmented label file is then binarized; for example, voxels with the same label as the hippocampus are set to 1, while the remaining voxels are set to 0. This yields an MRI image that includes segmentation information of the hippocampus.

[0076] It is understood that the image sequence may include two or more MRI images of the brain taken at different time points. Specifically, this could be MRI images acquired at intervals, such as multiple MRI images acquired one month or eight years apart.

[0077] In this image sequence, the earliest magnetic resonance image can be defined as the baseline magnetic resonance image, and the hippocampus in the baseline magnetic resonance image can be defined as the baseline hippocampus. For example... Figure 3 As shown, the magnetic resonance image corresponding to time point 0 is defined as the baseline magnetic resonance image.

[0078] Furthermore, in this embodiment of the invention, after obtaining the image sequence including the segmentation information, the baseline magnetic resonance image can be processed using skeleton representation (S-rep) to construct a skeleton representation model of the baseline hippocampus, i.e., the S-rep model of the baseline hippocampus.

[0079] It can be understood that the constructed skeletal representation model consists of a set of radial axis vectors (p, S), where p represents a point located inside the hippocampus, and S represents the corresponding vector of p, so that all radial axes in the radial axis vector set can represent the morphological structure of the baseline hippocampus.

[0080] For example, such as Figure 4 As shown, the union of the radial axis vector sets forms the interior of the hippocampus, the union at the top of the radial axis vector sets forms the boundary of the hippocampus, and the tail of the radial axis vector sets forms a folded double-sided surface, which can be called the skeletal trajectory. This skeletal trajectory divides the hippocampus into upper and lower parts; the surface corresponding to the upper part is called the upper surface, and the surface corresponding to the lower part is called the lower surface. No two radial axes in the radial axis vector set intersect each other, and each smooth point p is associated with two radial axes.

[0081] Then as Figure 4 As shown, the spokes pointing towards the upper surface of the hippocampus can be called the upper spokes, and the spokes pointing towards the lower surface of the hippocampus can be called the lower spokes; the spokes attached to the folding point of the skeletal trajectory can be called the crown spokes; the crown spokes attached to the end of the skeletal trajectory can be called the tip crown spokes; and the point with the highest average curvature on the surface of the hippocampus can be called the hippocampal tail tip.

[0082] like Figure 4 As shown, the two figures on the left show the side view and top view of the s-rep model.

[0083] Depend on Figure 4 It can be seen that the shape and structure of the hippocampus can be represented by the spoke-axis vector group, that is, the top of the spoke-axis vector group can be connected as the implicit surface of the hippocampus.

[0084] Furthermore, in this embodiment of the invention, in order to increase the morphological measurement scale of the model and more accurately reflect the characteristics of each subregion, after constructing the skeleton representation model of the baseline hippocampus through the above steps, subregion labels can be added to the skeleton representation model of the baseline hippocampus to obtain the multi-scale skeleton representation model of the baseline hippocampus.

[0085] In this context, the subregion labels in the multi-scale skeleton representation model of the baseline hippocampus can be used to represent the location of each subregion of the baseline hippocampus on the skeleton representation model.

[0086] For example, hippocampal subregion maps from anatomy and 7T-MRI can be mapped to the baseline hippocampal s-rep model to add hippocampal subregion labels to the baseline skeletal representation model, so that the mapped baseline hippocampal skeletal representation model includes subregion label information, thus constructing a multiscale skeletal representation (ms-rep) model of the baseline hippocampus.

[0087] Furthermore, after constructing the multi-scale skeleton representation model of the baseline hippocampus, the boundary surface of the baseline hippocampus, as well as the boundary surfaces at other time points, can be used to perform model transformation on the multi-scale skeleton representation model of the baseline hippocampus in the baseline magnetic resonance image, thereby obtaining the multi-scale skeleton representation model of the hippocampus at other time points of the test subject.

[0088] Finally, morphological measurements can be performed on the hippocampal ms-rep model at each time point to obtain the morphological parameters of the hippocampus at each time point, so as to characterize the temporal changes in volume or morphology of the hippocampus of the test subject at each time point.

[0089] It is understood that the multi-scale morphological measurement method of the human hippocampus based on clinical MRI images in this embodiment of the invention first constructs a baseline hippocampal skeleton representation model of the earliest time point of the MRI image sequence of the same test subject based on skeleton representation. Then, corresponding subregion labels are added to the baseline hippocampal skeleton representation model to generate a multi-scale skeleton representation model of the baseline hippocampus. This multi-scale skeleton representation model can characterize the local characteristics of each subregion of the baseline hippocampus. Then, the boundary surface of the hippocampus in the MRI images of all time points is used to transform the baseline hippocampal skeleton representation model to obtain the hippocampal skeleton representation model corresponding to each time point. Finally, based on the constructed hippocampal skeleton representation model of each time point, the measurement of multi-scale hippocampal morphological changes, including the overall hippocampus, subregions, and local morphology, can be realized.

[0090] This method utilizes skeletal representation to construct multi-scale skeletal representation models of test subjects at different time points, thereby providing a basis for accurately measuring the temporal changes in hippocampal morphology, improving the accuracy of hippocampal morphological change measurement, and obtaining morphological parameters that conform to histological interpretation, making them easier for clinical understanding. This provides a new approach for transforming hippocampal atrophy patterns discovered based on imaging into clinically usable disease analysis indicators.

[0091] Optionally, in some embodiments of the present invention, in S150, the hippocampus morphology of the s-rep model at each time point can be measured according to local and regional specific spatiotemporal measurement methods to obtain basic morphological parameters representing the hippocampus morphological characteristics of the test object at each time point and trend morphological parameters representing the temporal change characteristics of the hippocampus morphology.

[0092] like Figure 3 As shown in (d), the basic morphological parameters, namely the three-dimensional geometric feature measurement results, can specifically include the length (i.e., front-to-back span), width (i.e., inner-to-outer span), thickness, major axis, curvature, and annualized change rate (ACR) of the hippocampus.

[0093] Specifically, the thickness of the hippocampus can include the thickness specific to each subregion, as well as the overall thickness of the hippocampus. The subregion-specific thickness is the average of the measurements for each subregion.

[0094] For example, in some embodiments, the overall thickness of the hippocampus can be defined as the local thickness of each point on the upper surface of the hippocampus as the length of its corresponding upper spoke axis; and the local thickness of each point on the lower surface of the hippocampus as the length of its corresponding lower spoke axis. The overall thickness of the hippocampus is then obtained by summing the spoke axis lengths and the lower spoke axis lengths.

[0095] The length of the hippocampus can be determined by defining its long axis based on the centerline of the skeletal trajectory. The length of the hippocampus is then obtained by measuring this long axis and adding the lengths of the two coronal radial axes connected to its ends.

[0096] The width of the hippocampus can be obtained by measuring the width of the skeletal trajectory and adding the corresponding lengths of the two coronal axes.

[0097] The curvature of the hippocampus can be obtained by calculating the curvature of the longitudinal centerline of the skeletal contour surface, which gives the curvature of the major axis of the hippocampus.

[0098] Furthermore, such as Figure 3 As shown in (d), for the trend morphological parameters among the above morphological parameters, the above basic morphological parameters can be processed, that is, longitudinal measurement can be performed to obtain the longitudinal morphological parameters representing the changes in the hippocampal morphology and volume, that is, the annual change rate.

[0099] Specifically, a two-dimensional space can first be generated. The first coordinate axis in this two-dimensional space can be represented by various time points in the image sequence, and the second coordinate axis can be represented by the basic morphological parameters corresponding to each time point. Then, a linear regression can be performed on the values ​​of the first and second coordinate axes to determine the corresponding slope, which represents the trend morphological parameter of the hippocampus.

[0100] For example, to determine the rate of change in hippocampal volume, a two-dimensional space is first defined. The horizontal axis, such as the first coordinate axis, represents the longitudinal data of the subject arranged chronologically at various time points. The vertical axis, i.e., the second coordinate axis, represents the morphological parameters of the subject at that time point, such as length. A linear regression is then performed between the time vector on the horizontal axis and the morphological vector on the vertical axis. The slope of this linear regression is defined as the longitudinal shrinkage index of the hippocampus, ACR.

[0101] It is understood that in this embodiment, after constructing a multi-scale skeleton representation model at various time points using the skeleton representation method, the radial axes in the radial axis vector group of the multi-scale skeleton representation model at various time points can be used to calculate the morphological structure of the hippocampus and the changes in the hippocampus over time. That is, using the radial axes at various positions, basic parameters including the length, width, thickness, and curvature of the hippocampus, as well as longitudinal morphological parameters representing morphological changes, can be calculated. This allows for accurate determination of the volume shrinkage and changes of the hippocampus in the test subject, ultimately providing a new approach for transforming hippocampal shrinkage patterns discovered based on images into clinically applicable disease diagnostic indicators.

[0102] Optionally, in some embodiments of the present invention, in S110, when obtaining the image sequence of the test object, the original image sequence can be obtained first, and then the hippocampus of each magnetic resonance image in the original image sequence can be segmented and identified to obtain the image sequence.

[0103] For example, in some embodiments, the raw image sequence of the test subject to be processed can be obtained first. This raw image sequence may include two or more raw magnetic resonance imaging (MRI) images of the brain acquired at different time points, such as T1-weighted structural images. Specifically, this could be MRI images acquired at intervals, for example, multiple raw MRI images acquired one month or six months apart.

[0104] like Figure 1 As shown, the original image sequence of the brain includes three original magnetic resonance images acquired at different times. That is, from left to right, they are the original magnetic resonance images corresponding to time point 0, time point 1, and time point 3.

[0105] The first time point can be the starting moment in the time sequence, that is, the corresponding image, which is the original magnetic resonance image corresponding to the earliest acquisition time point in the original image sequence.

[0106] It is understandable that the original magnetic resonance images in the acquired original image sequence, as images to be processed and analyzed, include the contour information of the hippocampus, that is, each image has a corresponding hippocampal region.

[0107] Furthermore, after obtaining the original magnetic resonance imaging sequence of the brain, the hippocampal contour can be identified and segmented. For example, each original magnetic resonance image in the original magnetic resonance imaging sequence can be input into a segmentation model based on a neural network to obtain the segmentation result, which is the image sequence.

[0108] It is understood that in some embodiments, the above-mentioned magnetic resonance image sequence can also be obtained directly. That is, after the original image sequence is segmented and identified by other means or computer devices, a magnetic resonance image including the tag information of the hippocampus is obtained. This image can be input to the computing device executing the embodiments of the present invention by transmission or other means, so that the computer device can directly perform subsequent processing on the segmentation results after segmentation and identification.

[0109] Optional, such as Figure 5 As shown, in some embodiments of the present invention, in S120, based on the skeleton description method, the hippocampus in the earliest time point magnetic resonance image in the image sequence is used as the baseline hippocampus. When constructing the baseline hippocampus skeleton representation model, the specific steps include the following:

[0110] S121. Based on the skeleton representation method, an initial skeleton representation model of the baseline hippocampus is constructed.

[0111] Specifically, for the hippocampus in the magnetic resonance image corresponding to the earliest time point in the image sequence, i.e. the baseline hippocampus, the continuous internal filling skeleton representation model of s-rep can be used to perform morphological modeling on the segmented baseline hippocampus image to form an initial skeleton representation model.

[0112] Understandable, such as Figure 4 As shown, in this initial skeleton representation model, the union of the tops of the spoke-axis vector groups forms the boundary of the hippocampus. Connecting the tops of the spoke-axis vector groups yields the implicit surface of the hippocampus. The tails of the spoke-axis vector groups form a folded double-sided surface, called the skeleton trajectory. The skeleton trajectory divides the hippocampus into upper and lower parts; the surface corresponding to the upper part is called the upper surface, and the surface corresponding to the lower part is called the lower surface. Thus, the morphological structure of the hippocampus can be represented by the spoke-axis vector groups.

[0113] It is understandable that, due to the limitations of the s-rep model, the generated radial axis is at the end of the hippocampus, and the orientation of the crown-radial axis at the tip will also deviate. Therefore, in order to achieve accurate measurement of the hippocampus, the skeletal trajectory of the initial s-rep model can be adjusted, i.e., S122 can be executed.

[0114] S122, Adjust the length and position of the axis in the initial skeleton representation model to obtain the adjusted initial skeleton representation model.

[0115] Specifically, firstly, the length of the spokes in the spokes vector group can be adjusted so that the vertex of the spokes is in contact with the boundary surface of the baseline hippocampus; then, the tip crown spoke in the spokes vector group can be moved to the tail tip of the baseline hippocampus so that the tip crown secondary axis points to the end of the baseline hippocampus.

[0116] It is understandable that the above adjustments can ensure that the implicit surface corresponding to s-rep accurately matches the boundary surface corresponding to the segmentation information of the baseline hippocampus. Furthermore, by forcibly positioning the crown-radial axis of the high tip to the tail tip of the baseline hippocampus, it can be ensured that the crown-radial axis of the high tip always points to the end of the baseline hippocampus.

[0117] It is understandable that, due to the limitations of the s-rep model, the number of spokes generated by the initial s-rep model is insufficient to fully describe the morphological characteristics of the hippocampus. Therefore, the skeleton trajectory of the adjusted s-rep model can be interpolated to refine the mesh surface corresponding to the skeleton representation model and obtain a sufficient number of spokes.

[0118] S123, Based on the interpolation method, the mesh of the adjusted initial skeleton representation model is refined to obtain the refined initial skeleton representation model.

[0119] Specifically, in this embodiment of the invention, after obtaining the adjusted initial skeleton representation model described above, the spoke axis of the adjusted baseline hippocampus initial s-rep model can be interpolated based on an interpolation algorithm, such as the Catmull-Clark algorithm.

[0120] For example, when using the Catmull-Clark algorithm for interpolation, the specific steps include:

[0121] S01, determine the face points and edge points corresponding to each initial mesh surface in the mesh surface corresponding to the adjusted skeleton representation model.

[0122] S02, add edge points to the edges of the initial mesh surface to form a new mesh surface.

[0123] S03, determine the face points of the new grid surface.

[0124] S04, based on the face points and edge points of the new mesh surface, form a subdivided mesh surface.

[0125] S05, repeat the above steps of subdividing the initial mesh surface to form the subdivided mesh surface until the preset conditions are met, and obtain the target mesh surface, where each vertex of the target mesh surface is the interpolated spoke vertex.

[0126] S06. Based on each vertex in the target mesh surface, determine the new spoke axis after interpolation processing of the adjusted initial skeleton representation model, and obtain the refined initial skeleton representation model corresponding to the new spoke axis.

[0127] Specifically, for the initial skeleton representation model of the adjusted baseline hippocampus, the face points and edge points of its corresponding initial mesh surface can be used first.

[0128] For example, for a face consisting of four vertices V0, V1, V2, and V3, let f be the face. Then the face points f are... p It can be calculated using the following formula:

[0129]

[0130] Furthermore, for edge points, we can first calculate the center point of the edge.

[0131] For example, given 6 vertices V0, V1, V2, V3, V4, and V5, V0, V1, V3 arranged clockwise form face f1; V2, V3, V4, and V5 arranged counterclockwise form face f2. The edges formed by vertices V2 and V3... Let the center point be denoted as ec. Then:

[0132]

[0133] Secondly, calculate the edges. Point f of two adjacent faces f1 and f2 p1 f p2 center point f pc ,but:

[0134]

[0135] Finally, calculate the edges. edge ep:

[0136]

[0137] Further, calculate the average surface area.

[0138] For example, if point O has four adjacent faces f1, f2, f3, and f4, then the average face value of all adjacent faces of O is defined as avg_f. p :

[0139]

[0140] Furthermore, edge points can be added to the edges of the initial mesh surface to form a new mesh surface; then the face points of the new mesh surface are determined, and further subdivided mesh surfaces can be formed based on the face points and edge points of the new mesh surface.

[0141] Specifically, it can update the coordinates of the original point. For example, assuming the original point P has four adjacent faces, the updated point is P′, then:

[0142]

[0143] The values ​​of α and β can be manually set based on experience.

[0144] Furthermore, in actual processing, after completing the above update, the above iterative process can be repeated, that is, the initial mesh surface is subdivided to form the subdivided mesh surface.

[0145] It is understandable that each iteration generates a new face point fp and a new edge point ep. By repeating the above steps until the preset conditions are met, the skeleton trajectory mesh can be subdivided, and the target mesh surface is finally obtained. Each vertex in the target mesh surface is the interpolated spoke vertex.

[0146] Furthermore, based on the subdivided mesh, the new spoke axes at the vertices of the interpolated mesh can be calculated; the length and direction of the new spoke axes are defined as the average of the four adjacent spoke axes. That is, based on each vertex in the target mesh surface, the new spoke axes after interpolation of the adjusted initial skeleton representation model can be determined, and the refined initial skeleton representation model corresponding to the new spoke axes can be obtained.

[0147] For example, using the Catmull-Clark interpolation described above, a hippocampus can be represented by 1738 spoke axes, which meets the accuracy requirements for hippocampal measurement. Figure 3 As shown in (b).

[0148] It is understandable that the skeletal adjustment process in S122 only corrects some obvious representational errors in the length of the spokes and the tip of the hippocampus, and the resulting spokes do not fully meet the set central axis geometry. Furthermore, the interpolated spokes generated in step S123 depend on the spokes adjusted in the previous step.

[0149] Therefore, in this embodiment of the invention, S124 can also be executed, that is, all spokes can be refined by an optimization method to eliminate errors in the length and direction of the spokes.

[0150] S124. The refined initial skeleton representation model is optimized according to the central axis geometry to obtain the target skeleton representation model, which is the skeleton representation model of the baseline hippocampus.

[0151] Specifically, in this embodiment of the invention, in order to construct a more realistic hippocampal skeleton and enable the hippocampal skeleton representation model constructed by the spokes to accurately represent the morphological features of the hippocampus, multiple central axis geometric conditions are set when optimizing the initial skeleton representation model after interpolation, so that the central axis geometric conditions are satisfied as constraints during the optimization process.

[0152] The central axis geometry can include:

[0153] First condition: The implicit curved surface formed by the tips of the spokes coincides with the boundary surface of the hippocampus;

[0154] Second condition: The radial axis direction is perpendicular to the surface of the hippocampal boundary;

[0155] Third condition: The lengths of the upper spokes and the corresponding lower spokes are equal.

[0156] Then, corresponding to the above-mentioned central axis geometric conditions, the skeleton representation model after interpolation can be optimized, specifically including:

[0157] S001, adjust the vertex of the new spoke axis to the boundary surface of the baseline hippocampus, so that the implicit surface corresponding to the vertex of the new spoke axis coincides with the boundary surface of the baseline hippocampus;

[0158] S002, Adjust the direction of the new spoke axis so that the angle between the new spoke axis and the boundary surface of the baseline hippocampus does not exceed the first preset threshold.

[0159] S003, adjust the lengths of the upper and lower spokes in the new spokes so that the difference between the lengths of the upper and lower spokes in the new spokes does not exceed the second preset threshold, and obtain the target skeleton representation model.

[0160] Specifically, to ensure the accuracy of the reconstructed shape, the first condition can be forced to be strictly met, so that the tip of the spoke axis is on the original hippocampal boundary surface.

[0161] Furthermore, based on the second condition, an appropriate threshold (such as an included angle of 90°±5°) is set to restrict the non-orthogonality between the spoke axis and the implicit surface. The included angle between the spoke axis and the boundary surface is calculated using 1-cos(θp), where θp represents the angle between the spoke axis direction of vertex p on the skeleton trajectory and the normal vector of the implicit surface at the vertex closest to the spoke axis tip.

[0162] Finally, according to the third condition, since the spokes at the bend of the hippocampus head often pass through the hidden surface and point to the wrong position, the direction of the spokes can be gradually adjusted by forcibly applying the third condition and then gradually adjusting it to be approximately orthogonal to the hidden surface.

[0163] It is understandable that, in practice, the above three steps can be iterated and the first condition can be met again to obtain the final accurate spoke axis.

[0164] Optionally, in S130 of this embodiment of the invention, when adding corresponding subregion labels to the baseline hippocampal skeletal representation model, a hippocampal subregion atlas from anatomy and 7T-MRI can be used.

[0165] Specifically, firstly, the automated morphological analysis tool SPHARM-PDM can be used to establish a correspondence between points on the surface of the hippocampal subregion map and points on the hidden surface of the baseline hippocampus corresponding to the s-rep model of that baseline hippocampus. Then, the hippocampal subregion map can be mapped onto the hidden surface of the baseline hippocampus corresponding to the s-rep model of that baseline hippocampus. Finally, based on the nearest point of each radial axis on the hidden surface, the subregion label of the point on the surface closest to the endpoint of each radial axis can be assigned to the corresponding radial axis.

[0166] Finally, by projecting the subregion map onto the radial axis, the radial axis can be recombined by region, with radial axes of the same label grouped into one category, so that the same subregion label is used to represent the location of each subregion of the baseline hippocampus on the skeletal representation model.

[0167] For example, such as Figure 3 As shown in (b), tags including CA1, CA2, CA3, DG, SUB and SRLM subregions can be added.

[0168] It is understood that in this embodiment, by adding subregion labels of the hippocampus to the skeleton representation model, the morphological changes of each subregion can be accurately described during hippocampal morphology measurement. Specifically, the mean value of the radial axis measurement index can be used as the subregion measurement value.

[0169] Optionally, in some embodiments of the present invention, when converting the baseline hippocampal skeleton representation model using the hippocampal boundary information at each time point in S140 to obtain the hippocampal skeleton representation model corresponding to each time point, the deformation field generated by the hippocampal boundary information at each time point can be used.

[0170] That is, Figure 6 As shown, the process includes the following steps:

[0171] S141, determine the correspondence between the boundary surface of the baseline hippocampus and the boundary surfaces of the hippocampus at other time points.

[0172] S142, according to the correspondence, smooth the boundary surface of the baseline hippocampus to the boundary surface of the hippocampus at other time points to obtain the deformation field of the hippocampus from the baseline hippocampus to the hippocampus at other time points.

[0173] S143, deform the skeletal representation model of the baseline hippocampus according to the deformation field to obtain the skeletal representation model of the hippocampus at other time points.

[0174] Specifically, since there are large morphological differences between different individuals' hippocampi, while the morphological changes within an individual are small, in order to further improve the accuracy of hippocampal morphological change measurement, the multi-scale skeleton representation model of the baseline hippocampus processed by the above steps can be converted to other time points for the longitudinal data of the test subjects, i.e., magnetic resonance images at various time points, so as to obtain the multi-scale skeleton representation model corresponding to other time points.

[0175] It can be understood that, in this embodiment of the invention, the implicit surface is the surface constructed from the spoke axes of the model built using the skeleton representation method. The boundary surface refers to the surface directly constructed from the segmented image, that is, the surface corresponding to the segmentation information in the image sequence.

[0176] Through the above steps, the skeleton representation model of the baseline hippocampus is constructed, thus obtaining the implicit surface of the baseline hippocampus. Furthermore, the implicit surface of the s-rep of the baseline hippocampus obtained through the above steps coincides with the boundary surface of the baseline hippocampus, so the deformation field from the boundary surface of the baseline hippocampus to the boundary surface at each time point can be used for transformation.

[0177] In this embodiment, the correspondence between the hippocampal boundary surface at various time points of the test object and the baseline hippocampal boundary surface can first be established using SPHARM-PDM. This boundary surface is the original hippocampal boundary surface constructed from the hippocampal images at each time point.

[0178] Furthermore, based on the obtained surface correspondence, dense control points are set on each vertex of the surface mesh, and the surface boundary of the baseline hippocampus is smoothly deformed to the surface of the hippocampus boundary at other time points through thin-plate spline (TPS), thus obtaining the deformation field from the baseline hippocampus to each time point.

[0179] Furthermore, the obtained baseline hippocampal multi-scale skeletal model is deformed based on the deformation field to adapt to the hippocampal surface at other time points. For example... Figure 3 As shown in (c), the multi-scale skeleton representation model of the baseline hippocampus at time point 0 can be transformed by surface transformation to obtain the multi-scale skeleton representation models of the hippocampus at time points 1 and 2.

[0180] It is understood that in some embodiments, the multi-scale skeleton representation models generated through the above steps at other time points can be optimized by performing the above steps on the ms-rep models at each time point, so that they are more closely matched with the boundary surfaces corresponding to the segmentation information in the magnetic resonance images, so as to better adapt to the boundary surfaces and maintain the geometry of the s-rep.

[0181] On the other hand, such as Figure 7 As shown, this embodiment of the invention also provides a multi-scale morphological measurement device for the human hippocampus based on clinical MRI images, the device comprising:

[0182] The acquisition module 210 is used to acquire an image sequence, which includes at least two magnetic resonance images of the same test object at different time points. Each magnetic resonance image includes segmentation information of the hippocampus, and the segmentation information is used to represent the boundary surface of the hippocampus in the magnetic resonance image.

[0183] Module 220 constructs a skeleton representation model corresponding to the baseline hippocampus based on the skeleton description method. The baseline hippocampus is the hippocampus in the earliest time point of the magnetic resonance image in the image sequence. The skeleton representation model includes a radial axis vector group, and the radial axis in the radial axis vector group represents the morphological structure of the baseline hippocampus.

[0184] The mapping module 230 is used to add sub-region labels to the baseline skeleton representation model to obtain a multi-scale skeleton representation model of the baseline hippocampus. The sub-region labels are used to represent the regions where each sub-region of the baseline hippocampus is located on the skeleton representation model.

[0185] The conversion module 240 is used to convert the multi-scale skeleton representation model of the baseline hippocampus based on the boundary surface of the hippocampus in magnetic resonance images at other time points, so as to obtain the multi-scale skeleton representation model of the hippocampus in magnetic resonance images at other time points.

[0186] The measurement module 250 is used to perform morphological measurements on the hippocampus of the test subject at various time points based on the multi-scale skeleton representation model of the baseline hippocampus and other multi-scale skeleton representation models at various time points, and to obtain the morphological parameters of the hippocampus at each time point. The morphological parameters are used to characterize the temporal variation characteristics of the hippocampus morphology of the test subject.

[0187] Optionally, the multi-scale morphological measurement device for the human brain hippocampus based on clinical MRI images provided in this embodiment of the invention includes basic morphological parameters and trend morphological parameters. The basic morphological parameters include hippocampal length, overall thickness, thickness of each subregion, hippocampal width, and long axis curvature. The trend morphological parameters represent the atrophy of the hippocampus.

[0188] Optionally, the multi-scale morphological measurement device for the human hippocampus based on clinical MRI images provided in this embodiment of the invention includes an upper radial axis and a lower radial axis. The surface corresponding to the endpoint of the upper radial axis is the upper surface of the hippocampus, and the surface corresponding to the endpoint of the lower radial axis is the lower surface of the hippocampus. The longitudinal centerline of the multi-scale skeleton model is the long axis of the hippocampus, and the radial axes connected to the end of the long axis are the coronal radial axes.

[0189] The measurement module is specifically used for:

[0190] The sum of the length of the major axis and the length of the coronal axis is determined as the length of the hippocampus;

[0191] The sum of the lengths of the upper and lower radial axes of the hippocampus is determined as the overall thickness of the hippocampus.

[0192] The thickness of each subregion is determined as the average of the sum of the lengths of the upper and lower radial axes of the corresponding region, and is taken as the subregion thickness of the hippocampus.

[0193] Determine the curvature value of the major axis as the curvature of the hippocampus.

[0194] Optionally, the multi-scale morphological measurement device for the human hippocampus based on clinical MRI images provided in this embodiment of the invention has a measurement module specifically used for:

[0195] A two-dimensional space is generated, where the first coordinate axis of the two-dimensional space takes the values ​​of each time point in the image sequence, and the second coordinate axis takes the values ​​of each basic morphological parameter corresponding to each time point.

[0196] Linear regression was performed on the values ​​of the first and second coordinate axes to determine the corresponding slopes, which are the trend morphology parameters of the hippocampus.

[0197] Optionally, the multi-scale morphological measurement device for the human hippocampus based on clinical MRI images provided in this embodiment of the invention is specifically configured to:

[0198] Based on the skeleton representation method, an initial skeleton representation model of the baseline hippocampus is constructed.

[0199] Adjust the length and position of the axis in the initial skeleton representation model to obtain the adjusted initial skeleton representation model;

[0200] Based on the interpolation method, the mesh of the adjusted initial skeleton representation model is refined to obtain the refined initial skeleton representation model;

[0201] The initial skeleton representation model is optimized based on the central axis geometry to obtain the target skeleton representation model, which is the skeleton representation model of the baseline hippocampus.

[0202] Optionally, the multi-scale morphological measurement device for the human hippocampus based on clinical MRI images provided in this embodiment of the invention further includes a apical coronal-radial axis located at the end of the long axis, and the construction module is specifically used for:

[0203] Adjust the length of the spoke axis so that the apex of the spoke axis is in contact with the boundary surface of the baseline hippocampus;

[0204] Move the tip of the coronal axis to the tail tip of the baseline hippocampus so that the tip of the coronal axis points to the end of the baseline hippocampus.

[0205] Optionally, the multi-scale morphological measurement device for the human hippocampus based on clinical MRI images provided in this embodiment of the invention is specifically used to construct the model for:

[0206] Determine the face points and edge points corresponding to each initial mesh face in the mesh surface corresponding to the adjusted skeleton representation model;

[0207] Add edge points to the edges of the initial mesh surface to form a new mesh surface;

[0208] Determine the face points of the new mesh surface;

[0209] Based on the face points and edge points of the new mesh surface, a subdivided mesh surface is formed;

[0210] Repeat the above steps of subdividing the initial mesh surface to form a subdivided mesh surface until the preset conditions are met, and obtain the target mesh surface. Each vertex in the target mesh surface is the spoke vertex after interpolation.

[0211] Based on each vertex in the target mesh surface, determine the new spoke axis after interpolation processing of the adjusted initial skeleton representation model, and obtain the refined initial skeleton representation model corresponding to the new spoke axis.

[0212] Optionally, the multi-scale morphological measurement device for the human hippocampus based on clinical MRI images provided in this embodiment of the invention is specifically configured to:

[0213] Adjust the vertex of the new spoke axis to the boundary surface of the baseline hippocampus so that the implicit surface corresponding to the vertex of the new spoke axis coincides with the boundary surface of the baseline hippocampus.

[0214] Adjust the direction of the new spoke axis so that the angle between the new spoke axis and the boundary surface of the baseline hippocampus does not exceed the first preset threshold.

[0215] Adjust the lengths of the upper and lower spokes in the new spokes so that the difference between the lengths of the upper and lower spokes in the new spokes does not exceed a second preset threshold, and obtain the target skeleton representation model.

[0216] Optionally, in the multi-scale morphological measurement device for the human hippocampus based on clinical MRI images provided in this embodiment of the invention, the conversion module is specifically used for:

[0217] Determine the correspondence between the boundary surface of the baseline hippocampus and the boundary surfaces of the hippocampus at other time points;

[0218] Based on the correspondence, the boundary surface of the baseline hippocampus is smoothed to the boundary surface of the hippocampus at other time points, thus obtaining the deformation field from the baseline hippocampus to the hippocampus at other time points.

[0219] The baseline hippocampal skeleton representation model is deformed based on the deformation field to obtain the hippocampal skeleton representation models at other time points.

[0220] On the other hand, the computer device provided in the embodiments of the present invention further includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the multi-scale morphological measurement method of the human brain hippocampus based on clinical MRI images as described above.

[0221] The following is for reference. Figure 8 , Figure 8 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention.

[0222] like Figure 8As shown, the electronic device includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage section 308 into a random access memory (RAM) 303. The RAM 303 also stores various programs and data required for the operation of the electronic device 300. The CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304. In some embodiments, the following components are connected to the I / O interface 305: an input section 306 including a keyboard, mouse, etc.; an output section 307 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN card, modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A driver 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are mounted on drive 310 as needed so that computer programs read from them can be installed into storage section 308 as needed. In particular, according to embodiments of the invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the invention include a computer program product comprising a computer program carried on a machine-readable medium containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable media 311. When the computer program is executed by central processing unit (CPU) 301, the functions defined in the electronic device of the invention are performed.

[0223] It should be noted that the computer-readable medium shown in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electronic device, apparatus, or device that is electrical, magnetic, optical, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an electronic device, apparatus, or device that executes instructions. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit programs for use by or in connection with an electronic device, apparatus, or device whose instructions are executed. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0224] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of electronic devices, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based electronic device that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0225] The units or modules described in the embodiments of the present invention can be implemented in software or hardware. The described units or modules can also be housed in a processor; for example, they can be described as: a processor including: an acquisition module, a construction module, a mapping module, a conversion module, and a measurement module. The names of these units or modules do not necessarily limit the specific unit or module itself. For example, the measurement module can also be described as "for performing morphological measurements on the hippocampus of the test object at various time points based on the multi-scale skeleton representation model of the baseline hippocampus and other multi-scale skeleton representation models at various time points, obtaining morphological parameters of the hippocampus at each time point, wherein the morphological parameters characterize the temporal variation characteristics of the hippocampus morphology of the test object."

[0226] In another aspect, the present invention also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device. The computer-readable storage medium stores one or more computer programs that, when used by one or more processors, execute the multi-scale morphological measurement method for the human hippocampus based on clinical MRI images described in the present invention:

[0227] An image sequence is acquired, the image sequence including at least two magnetic resonance images of the same test object at different time points, each magnetic resonance image including hippocampal segmentation information, the segmentation information being used to represent the boundary surface of the hippocampus in the magnetic resonance image;

[0228] Based on the skeleton description method, a skeleton representation model corresponding to the baseline hippocampus is constructed. The baseline hippocampus is the hippocampus in the earliest magnetic resonance image in the image sequence. The skeleton representation model includes a radial axis vector group, and the radial axis in the radial axis vector group represents the morphological structure of the baseline hippocampus.

[0229] Subregion labels are added to the baseline skeleton representation model to obtain a multi-scale skeleton representation model of the baseline hippocampus. The subregion labels are used to represent the regions where each subregion of the baseline hippocampus is located on the skeleton representation model.

[0230] Based on the boundary surface of the hippocampus in the baseline hippocampus and other time-point magnetic resonance images, the multi-scale skeleton representation model of the baseline hippocampus is transformed to obtain the multi-scale skeleton representation model of the hippocampus in the magnetic resonance images at other time points.

[0231] Based on the multi-scale skeleton representation model of the baseline hippocampus and the multi-scale skeleton representation models at other time points, morphological measurements of the hippocampus of the test subject at each time point are performed to obtain morphological parameters of the hippocampus at each time point. These morphological parameters are used to characterize the temporal variation characteristics of the hippocampus morphology of the test subject.

[0232] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the foregoing disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

Claims

1. A method for multi-scale morphological measurement of the human hippocampus based on clinical MRI images, characterized in that, include: An image sequence is acquired, the image sequence including at least two magnetic resonance images of the same test object at different time points, each magnetic resonance image including hippocampal segmentation information, the segmentation information being used to represent the boundary surface of the hippocampus in the magnetic resonance image; Based on the skeleton description method, a skeleton representation model corresponding to the baseline hippocampus is constructed, including: Based on the skeleton representation method, an initial skeleton representation model of the baseline hippocampus is constructed; Adjust the length and position of the axis in the initial skeleton representation model to obtain the adjusted initial skeleton representation model; Based on the interpolation method, the mesh of the adjusted initial skeleton representation model is refined to obtain a refined initial skeleton representation model; The refined initial skeleton representation model is optimized based on the central axis geometry to obtain the target skeleton representation model, which is the skeleton representation model of the baseline hippocampus. The baseline hippocampus is the hippocampus in the earliest time point magnetic resonance image in the image sequence. The skeleton representation model includes a radial axis vector group, and the radial axis in the radial axis vector group represents the morphological structure of the baseline hippocampus. Subregion labels are added to the baseline skeleton representation model to obtain a multi-scale skeleton representation model of the baseline hippocampus. The subregion labels are used to represent the regions where each subregion of the baseline hippocampus is located on the skeleton representation model. Based on the boundary surface of the hippocampus in the baseline hippocampus and other time-point magnetic resonance images, the multi-scale skeleton representation model of the baseline hippocampus is transformed to obtain the multi-scale skeleton representation model of the hippocampus in the magnetic resonance images at other time points. Based on the multi-scale skeleton representation model of the baseline hippocampus and the multi-scale skeleton representation models of other time points, morphological measurements of the hippocampus of the test subject at each time point are performed to obtain the morphological parameters of the hippocampus at each time point. The morphological parameters are used to characterize the temporal variation characteristics of the hippocampus morphology of the test subject. The morphological parameters include basic morphological parameters and trend morphological parameters. The basic morphological parameters include length, overall thickness, thickness of each subregion, width and curvature. The trend morphological parameters represent the atrophy of the hippocampus. The spokes include an upper spoke and a lower spoke. The surface corresponding to the endpoint of the upper spoke is the upper surface of the hippocampus, and the surface corresponding to the endpoint of the lower spoke is the lower surface of the hippocampus. The multi-scale skeleton represents the longitudinal centerline of the model as the long axis of the hippocampus, and the spokes connected to the end of the long axis are called coronal spokes. The multi-scale skeleton representation model based on the baseline hippocampus and other multi-scale skeleton representation models at various time points are used to perform morphological measurements on the hippocampus of the test subject at various time points, obtaining the morphological parameters of the hippocampus at each time point, including: The sum of the length of the major axis and the length of the coronal radial axis is determined as the length of the hippocampus; The sum of the lengths of the upper radial axis and the lower radial axis of the hippocampus is determined as the overall thickness of the hippocampus; The thickness of each subregion is determined as the average of the sum of the lengths of the upper and lower radial axes of the corresponding region, and is taken as the subregion thickness of the hippocampus; The curvature value of the major axis is determined as the curvature of the hippocampus.

2. The method for multi-scale morphological measurement of the human hippocampus based on clinical MRI images according to claim 1, characterized in that, The multi-scale skeleton representation model based on the baseline hippocampus and other multi-scale skeleton representation models at various time points, used to perform morphological measurements on the hippocampus of the test subject at various time points, and to obtain the morphological parameters of the hippocampus at each time point, further includes: A two-dimensional space is generated, wherein the first coordinate axis in the two-dimensional space takes the value of each time point of the image sequence, and the second coordinate axis takes the value of each basic morphological parameter corresponding to each time point; Linear regression is performed on the values ​​of the first and second coordinate axes to determine the corresponding slope, which is the trend morphology parameter of the hippocampus.

3. The method for multi-scale morphological measurement of the human hippocampus based on clinical MRI images according to claim 1, characterized in that, The spoke axis also includes a pointed crown-spoke axis located at the end of the long axis, and adjusting the length and position of the spoke axis in the initial skeleton representation model includes: Adjust the length of the spokes so that the apex of the spokes contacts the boundary surface of the baseline hippocampus; Move the tip coronal axis to the tail tip of the baseline hippocampus, so that the tip coronal axis points to the end of the baseline hippocampus.

4. The method for multi-scale morphological measurement of the human hippocampus based on clinical MRI images according to claim 1, characterized in that, The method based on interpolation refines the mesh of the adjusted initial skeleton representation model to obtain a refined initial skeleton representation model, including: Determine the face points and edge points corresponding to each initial mesh surface in the mesh surface corresponding to the adjusted skeleton representation model; Add edge points to the edges of the initial mesh surface to form a new mesh surface; Determine the face points of the new mesh surface; Based on the face points and edge points of the new mesh surface, a subdivided mesh surface is formed; Repeat the above steps of subdividing the initial mesh surface to form the subdivided mesh surface until the preset condition is met to obtain the target mesh surface, where each vertex in the target mesh surface is the interpolated spoke vertex. Based on each vertex in the target mesh surface, determine the new spoke axis after interpolation processing of the adjusted initial skeleton representation model, and obtain the refined initial skeleton representation model corresponding to the new spoke axis.

5. The method for multi-scale morphological measurement of the human hippocampus based on clinical MRI images according to claim 4, characterized in that, The optimization of the refined initial skeleton representation model based on the central axis geometry to obtain the target skeleton representation model includes: The vertex of the new spoke axis is adjusted to the boundary surface of the baseline hippocampus, so that the implicit surface corresponding to the vertex of the new spoke axis coincides with the boundary surface of the baseline hippocampus. Adjust the direction of the new spoke axis so that the angle between the new spoke axis and the boundary surface of the baseline hippocampus does not exceed a first preset threshold. Adjust the lengths of the upper and lower spokes in the new spokes so that the difference between the lengths of the upper and lower spokes in the new spokes does not exceed a second preset threshold, thereby obtaining the target skeleton representation model.

6. The method for multi-scale morphological measurement of the human hippocampus based on clinical MRI images according to any one of claims 1-5, characterized in that, The step of transforming the multi-scale skeleton representation model of the baseline hippocampus based on the boundary surface of the hippocampus in magnetic resonance images at other time points to obtain the multi-scale skeleton representation model of the hippocampus in the magnetic resonance images at other time points includes: Determine the correspondence between the boundary surface of the baseline hippocampus and the boundary surfaces of the hippocampus at other time points; Based on the correspondence, the boundary surface of the baseline hippocampus is smoothed to the boundary surface of the hippocampus at other time points to obtain the deformation field of the hippocampus from the baseline hippocampus to the hippocampus at other time points. The baseline hippocampal skeleton representation model is deformed based on the deformation field to obtain the hippocampal skeleton representation models at other time points.

7. A computer device, characterized in that, The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the multi-scale morphological measurement method of the human brain hippocampus based on clinical MRI images as described in any one of claims 1-6.

Images