Method and system for locating myocardial infarction area combined with image

Abstract

The application relates to the technical field of heart image positioning, in particular to a myocardial infarction region positioning method and system combined with images, which comprises the following steps: extracting a left ventricular wall thickness direction image region, combining frame sequence signal changes, arranging according to a space level, identifying an abnormal boundary according to a tension stress direction reversal, extracting a high response rate point in a perfusion image, identifying a level trend change, intersecting and screening an abnormal region to generate a positioning mask, in the application, wall thickness direction coordinates are extracted, combined with continuous frame signal strength, a tension stress time sequence is constructed, structure dynamic characteristics are captured, a boundary is extracted according to a tension stress direction reversal, abnormal region positioning is realized, a high response rate point is identified by matching a perfusion signal, the sensitivity of function change is enhanced, a gradient surface is generated by combining image level distribution, response continuity and directionality are highlighted, positioning indexes are constructed by fusing structure and function characteristics, a mask is generated by screening a target region, and the consistency and recognition precision of space positioning are improved.

Description

Technical Field

[0001] This invention relates to the field of cardiac image localization technology, and in particular to a method and system for localizing myocardial infarction areas by combining images. Background Technology

[0002] The field of cardiac image localization technology involves methods for identifying and spatially locating cardiac structure, function, and diseased areas using medical imaging techniques. Core aspects include quantitative analysis and precise annotation of the myocardium, cardiac chambers, blood vessels, and abnormal areas through acquired cardiac imaging data. This technology primarily employs image acquisition, image processing, and image feature extraction methods for visual diagnosis and localization analysis of the heart. Traditional methods for locating myocardial infarction areas using imaging involve manual identification and contour drawing of the grayscale distribution and texture features of the myocardial region in CT or MRI images. For the identification and localization of myocardial infarction areas, static image grayscale thresholding or pixel classification based on image segmentation algorithms combined with manual anatomical annotation are commonly used for analysis and localization.

[0003] Using static image grayscale segmentation or manual annotation based on texture features for myocardial lesion identification is limited by the feature representation capabilities of a single image frame. It struggles to reflect the temporal development and changes of lesions and cannot meet the dynamic feature recognition requirements during the rapid systolic and diastolic cycles of the heart. In practice, image processing relies on single-frame grayscale differences for region segmentation, which can lead to blurred boundaries between lesion areas and normal tissue due to differences in illumination, insufficient local contrast, or overlapping tissue signals, resulting in decreased accuracy. Image segmentation methods often lack coordinated analysis of the spatial hierarchical structure and temporal behavior of images, frequently resulting in insufficient judgment of structural continuity and prone to location drift or mislabeling. For example, when stress abnormalities occur in the mid-myocardium but do not show obvious grayscale changes, static segmentation techniques struggle to detect this area, leading to a decreased lesion detection rate. Manual contour drawing is inefficient and subject to individual experience differences, making it difficult to maintain annotation consistency in high frame rate sequences. This reduces the stability and repeatability of structural recognition, resulting in insufficient image anatomical consistency and limiting the ability to quantitatively describe the extent and location of lesions in precision medicine. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a method for locating myocardial infarction regions by combining imaging, comprising the following steps: To achieve the above objectives, the present invention employs the following technical solution: a method for locating myocardial infarction regions using imaging, comprising the following steps: S1: Extract the image coordinate region in the wall thickness direction of the left ventricle image, combine the signal intensity changes of the points in the continuous frame sequence, construct the trend sequence of tensile stress change over time, and arrange them according to spatial hierarchy to generate a tensile stress change structure array map. S2: Based on the direction of tensile stress change in the tensile stress change structure array map, identify the coordinates of the direction reversal points, summarize them to form a continuous distribution area, extract the boundary shape, and generate the structural anomaly image annotation area. S3: Based on the spatial coordinates within the marked area of ​​the structural anomaly image, extract the corresponding signal time change in the perfusion image sequence, extract the set of points whose rate exceeds a set threshold, mark the response area, and generate functional partition image segments. S4: Utilize the distribution of points in the functional partitioned image segments at the image structure hierarchy to identify regions that exhibit trend changes in hierarchical continuity, construct a set of locations in the image space, and generate a set of interlayer response gradient surface coordinates. S5: Based on the intersection coordinates of the interlayer response gradient surface coordinate set and the structural anomaly region, extract the point structure and functional feature values, construct the positioning reference index, screen the regions that meet the conditions, and mark them in the form of a mask to generate the positioning mask output region.

[0005] As a further aspect of the present invention, the tensile stress variation structure array map includes a cluster of time-series tensile stress curves, a spatial hierarchy index, and a pixel mapping relationship; the structural anomaly image annotation region includes a set of direction reversal points, a continuous distribution boundary, and anomaly region label attributes; the functional partition image segment includes an injection signal time series, a set of suprathreshold velocity points, and a response region label layer; the interlayer response gradient surface coordinate set includes a hierarchical continuity segment, a cluster of segmented spatial locations, and gradient surface fitting parameters; and the positioning mask output region includes a structural functional composite index, an intersection point index, and a mask pixel set.

[0006] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Extract the image coordinate region in the wall thickness direction of the left ventricle image, identify the spatial distribution of the inner and outer wall edges in the image sequence, determine the central axis coordinate region based on the connection between the edges, record the signal intensity changes of the pixels in the region in consecutive image frames, and obtain the image signal intensity sequence value in the wall thickness direction. S102: Call the image signal intensity sequence value in the wall thickness direction, construct the signal curve sequence of pixel points changing with time, extract key change nodes in the curve, calculate the degree of signal change between nodes, and obtain the trend value of the tensile stress change rate in the wall thickness region. S103: Call the trend value of the tensile stress change rate in the wall thickness region, construct a spatial mapping matrix of the tensile stress change trend according to the spatial arrangement order in the wall thickness direction, arrange the spatiotemporal correspondence of the trend values, and obtain the tensile stress change structure array map.

[0007] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Based on the tensile stress change trend value of the spatial level in the tensile stress change structure array map, determine the positive and negative relationship of the corresponding tensile stress change direction of adjacent levels in the time series, filter the position points where the direction change is reversed, mark the map coordinate value of the position points, and obtain the tensile stress direction reversal coordinates. S202: Call the tensile stress direction reversal coordinates, determine the connection relationship between adjacent points according to the continuity of spatial distribution, extract the boundary of the connected region formed by the densely connected region, generate the contour line of the closed shape according to the connection sequence of the boundary line, and obtain the tensile stress reversal connected boundary contour. S203: Call the tensile stress inversion connected boundary contour, map the boundary contour line to the original image coordinate area, draw the contour line shape according to the image position relationship of each closed boundary, mark the covered image area, distinguish the label number corresponding to different closed boundaries, and obtain the structural anomaly image label area.

[0008] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Based on the image spatial coordinates of the points in the annotated region of the structural anomaly image, extract the signal change values ​​of the points in the corresponding perfusion image sequence at different times, and generate a perfusion signal time series; S302: Call the perfusion signal time series, calculate the signal change rate value of the point based on the signal change trend between consecutive time points, and filter the points whose rate exceeds the perfusion signal rate threshold to obtain a set of abnormal perfusion response rate points. S303: Call the set of abnormal points of the perfusion response rate, mark the boundary of the response area according to the distribution relationship of the abnormal points in the image space, complete the area drawing on the image, and obtain the functional partition image fragment.

[0009] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Based on the distribution information of all points in the functional partition image segment at the image structure level, identify the structural level label of the point, compare the positions according to the continuity trend of point distribution between adjacent levels, determine the continuous level segments with order differences, and establish a sequence of structural level trend change segments. S402: Call the sequence of trend change segments of the structure hierarchy, combine it with the coordinate information in the image space, extract the two-dimensional position value of the point in the image space within each segment, aggregate the point coordinates within the same segment, and generate a set of segment position coordinates. S403: Call the set of location coordinates of the layer segment, calculate the spatial change trend value between the location sets based on the spatial gradient difference of the point coordinates in the location sets between adjacent layer segments, and map the trend value into three-dimensional coordinate form to obtain the inter-layer response gradient surface coordinate set.

[0010] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Based on the intersection coordinates between the interlayer response gradient surface coordinate set and the annotated area of ​​the structural anomaly image, extract the structural feature values ​​and functional feature values ​​corresponding to the coordinate points, organize them into a unified structure, and generate a set of structural and functional feature values. S502: Call the set of structural and functional feature values, and based on the numerical performance of the structural and functional feature values ​​at coordinate points, filter the points that simultaneously meet the structural feature threshold and the functional feature threshold to obtain the coordinate set of the threshold-satisfied region. S503: Based on the threshold satisfying the regional coordinate set, identify the related adjacent points, generate a mask matrix for the corresponding image positions, superimpose it onto the original image space, and obtain the positioning mask output area.

[0011] As a further aspect of the present invention, the wall thickness direction refers to the perpendicular anatomical direction along the left ventricular wall to the epicardium; The stress variation trend refers to the stress estimation curve calculated by inter-frame variation based on the change of image signal intensity of myocardium during the contraction cycle. The tensile stress variation structure array map refers to a matrix map formed by hierarchically arranging the pixel positions in the image space with the corresponding tensile stress estimation results. The structural anomaly image annotation region refers to the region in the image space marked with a closed boundary where the direction of tensile stress change is abnormally concentrated, serving as an image localization reference for suspected myocardial injury or lesion areas.

[0012] As a further aspect of the present invention, the perfusion imaging sequence refers to a set of continuous image frames that reflect the distribution process of contrast agents in myocardial tissue, acquired using magnetic resonance or CT technology. The signal time refers to the rate of change of signal intensity with respect to time at each image point in the perfusion sequence; The rate exceeding the set threshold means that the value of the signal time derivative is greater than the reference standard value of the perfusion rise rate set by the system. The threshold can be obtained through clinical sample statistics or literature settings. The functional partition image segment refers to a set of regions in the image that show significant changes in perfusion response and high signal derivative values, which are used as functionally active regions for spatial segmentation and labeling; The hierarchical continuity refers to the sequential relationship of the slice sequences of myocardial images arranged from the inside to the outside in the wall thickness direction; The interlayer response gradient surface coordinate set refers to a spatial set composed of pixel positions in multiple image layers that exhibit response change trends, forming a reference surface for judging the depth distribution of regional changes. The structural and functional characteristic values ​​refer to the tensile stress estimation value, the infusion time delay value, and the infusion rise rate corresponding to the image points; The positioning reference index refers to a quantitative standard for judging positioning, formed by structural and functional characteristic values; The masking method refers to converting the recognition result area into a binary region in the image layer by using image masking.

[0013] A system for locating myocardial infarction regions based on imaging includes: The tensile stress trend construction module obtains the coordinate region of the wall thickness direction in the left ventricular image sequence, extracts the signal intensity value of the coordinate point in the continuous image frame, calculates the change trend of the signal in the time series, constructs the change trend set corresponding to the slice according to the spatial arrangement order of the image slice, and generates a tensile stress change structure array map. Based on the tensile stress change structure array map, the structural anomaly identification module extracts the trend direction of the same coordinate points in adjacent image slices, identifies the points where the direction is reversed, summarizes the regions formed by the spatially continuous reversal points, extracts the region boundaries and marks them on the image, and generates the structural anomaly image annotation region. The perfusion response extraction module calls the points in the annotated area of ​​the structural anomaly image, extracts the signal time change of the corresponding points in the perfusion image sequence, calculates the signal change rate, extracts the set of points whose rate exceeds the set threshold, marks them in the image to form regional segments, and generates functional partition image segments. The interlayer gradient generation module identifies region segments with continuous numbers based on the image slice numbers of the points in the functional partition image segments, extracts the spatial distribution positions and records the corresponding coordinates, and generates an interlayer response gradient surface coordinate set. The positioning index filtering module extracts the structural and functional feature values ​​of each point based on the intersection points of the interlayer response gradient surface coordinate set and the structural anomaly image annotation area, calculates the index values, filters the set of points that meet the set criteria, marks them on the image in the form of a mask, and generates a positioning mask output area.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, tensile stress time series is constructed by extracting wall thickness direction coordinates and combining them with continuous frame signal intensity to capture structural dynamic features. Boundaries are extracted based on the reverse of tensile stress direction to achieve abnormal area localization. High response rate points are identified by matching perfusion signals to enhance the sensitivity to functional changes. Gradient surfaces are generated by combining image hierarchical distribution to highlight response continuity and directionality. Location indicators are constructed by fusing structural and functional features. Target areas are selected to generate masks, thereby improving the consistency and recognition accuracy of spatial localization. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention; Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] Please see Figure 1 This invention provides a method for locating myocardial infarction regions using imaging, comprising the following steps: S1: Extract the image coordinate region in the wall thickness direction of the left ventricle image, combine the image signal intensity changes of the points in the continuous frame sequence, construct the tensile stress change trend in the time series, and arrange the trend sequence according to the spatial hierarchy to generate a tensile stress change structure array map. The wall thickness direction refers to the vertical anatomical direction along the left ventricular wall to the epicardium. This direction is used to delineate the hierarchical structure of the myocardium in cardiac images and is a commonly used directional label for locating myocardial regions in magnetic resonance imaging. The stress variation trend refers to the stress estimation curve calculated by inter-frame variation based on the change of image signal intensity of myocardium during the contraction cycle, which is used to reflect the dynamic stress response of tissue during the cardiac cycle. Tension stress variation structure array map refers to a matrix map formed by hierarchically arranging the pixel positions in the image space with the corresponding tension stress estimation results. The map reflects the stress distribution structure of the myocardial region and can be used for boundary identification. S2: Based on the tensile stress change direction of adjacent layers in the tensile stress change structure array map, identify the coordinates of points where the change direction is reversed, summarize them to form a continuous distribution area, extract the boundary shape in the image space and label the image area to generate the structural anomaly image labeling area. Reversal of change direction refers to the behavior of the change trend of tensile stress between adjacent layers switching between positive and negative directions. This behavior is used to determine the discontinuous change of the local mechanical state and is the basis for judging the location of potential boundaries. Structural anomaly image annotation regions refer to areas in the image space marked with closed boundaries where the direction of tensile stress changes is abnormally concentrated, serving as image localization references for suspected areas of myocardial injury or lesions; S3: Based on the spatial coordinates of all points in the annotated area of ​​the structural anomaly image, extract the signal change over time of each point in the corresponding perfusion image sequence, extract the set of points whose rate exceeds the set threshold based on the time derivative of the perfusion signal, mark the response area in the image, and generate functional partition image segments. Perfusion imaging sequence refers to a collection of continuous image frames acquired using magnetic resonance or CT technology, reflecting the distribution process of contrast agents in myocardial tissue. The sequence is used to analyze the myocardial perfusion function status. The signal time derivative refers to the rate of change of signal intensity with respect to time at each image point in the perfusion sequence, and is used to quantify the blood flow response velocity in a local area. A rate exceeding the set threshold means that the value of the signal time derivative is greater than the reference standard value of the perfusion rise rate set by the system. The threshold can be obtained through clinical sample statistics or literature settings. Functional image segments refer to a set of regions in an image that show significant changes in perfusion response and high signal derivative values, which are used for spatial segmentation and labeling as functionally active regions; S4: Based on the distribution information of all points in the functional partition image segment at the image structure level, identify the segmented regions that show a trend change in hierarchical continuity, construct the position set of the segment in the image space, and generate the inter-layer response gradient surface coordinate set. Hierarchical continuity refers to the sequential relationship of slices arranged from the inside to the outside in the wall thickness direction of myocardial images, and is used to describe the tissue connectivity of spatial structures between layers; The interlayer response gradient surface coordinate set refers to the spatial set of pixel positions in multiple image layers that show the trend of response changes, which constitutes a reference surface for judging the depth distribution of regional changes. S5: Based on the intersection coordinates between the interlayer response gradient surface coordinate set and the labeled area of ​​the structural anomaly image, extract the structural and functional feature values ​​corresponding to each point, construct the localization reference index, select the areas that meet the requirements and mark them in the image in the form of a mask, and generate the localization mask output area. Structural and functional characteristic values ​​refer to the tensile stress estimation value, perfusion time delay value, and perfusion rise rate corresponding to image points, which can be used for region identification, point selection, and region scoring. Positioning reference indicators refer to quantitative standards formed by structural and functional characteristic values ​​used to determine positioning, and are used for the selection and output of positioning areas. Masking refers to the process of converting the recognition result area into a binary region in the image layer by masking the image, so that the localization result can be displayed and processed.

[0023] The tensile stress variation structure array map includes a cluster of time-series tensile stress curves, a spatial hierarchy index, and pixel mapping relationships. The structural anomaly image annotation area includes a set of direction reversal points, continuous distribution boundaries, and anomaly region label attributes. The functional partition image segments include a time series of injection signals, a set of suprathreshold velocity points, and a response region label layer. The interlayer response gradient surface coordinate set includes hierarchical continuity segments, a cluster of segmented spatial locations, and gradient surface fitting parameters. The localization mask output area includes structural functional composite indices, intersection point indices, and a mask pixel set.

[0024] Please see Figure 2 The specific steps of S1 are as follows: S101: Extract the image coordinate region in the wall thickness direction of the left ventricle image, identify the spatial distribution of the inner and outer wall edges in the image sequence, determine the central axis coordinate region based on the connection between the edges, record the signal intensity changes of the pixels in the region in consecutive image frames, and obtain the image signal intensity sequence value in the wall thickness direction. When extracting the image coordinate region in the wall thickness direction of the left ventricle image, it is necessary to process the original image sequence data of cardiac MRI, preferably in DICOM format. The region where the left ventricle is located is initially segmented and identified by a deep learning image segmentation model such as U-Net. After obtaining the approximate outline of the left ventricle in the image, image processing methods such as edge detection algorithms or active contour models are used to extract the coordinate information of the inner and outer wall edges of the left ventricle in consecutive frames. Then, the matched inner and outer wall edges in each frame are paired in pixel order to establish a set of lines in the wall thickness direction. Each line represents the pixel connection region in the wall thickness direction. The position information of all pixels contained in each line is recorded according to the image coordinate system. Taking a scanning resolution of 0.5 mm / pixel as an example, if the line length is 8 mm, it contains 16 pixels. At each pixel location, the grayscale signal value of the corresponding position in each frame of the image is extracted sequentially to form a continuous frame signal sequence. If the image frame rate is 30 frames per second and the analysis period is 2 seconds, then each point generates 60 signal values. A signal matrix is ​​then established for all pixels along each line, with each row representing the signal intensity change sequence of a pixel over time. For example, the signal values ​​of the 8th pixel on a certain line from frame 1 to frame 60 are 85, 88, 92, etc., indicating its brightness change over time. The above extraction operation is repeated for each line along the wall thickness direction, ultimately forming a set of grayscale changes of multiple pixels in the time domain, which is used for subsequent analysis of the signal change characteristics along the wall thickness direction.

[0025] S102: Call the image signal intensity sequence value in the wall thickness direction, construct the signal curve sequence of pixel points changing with time, extract key change nodes in the curve, calculate the degree of signal change between nodes, and obtain the trend value of the tensile stress change rate in the wall thickness region. When using the image signal intensity sequence values ​​along the wall thickness direction, a time-corresponding signal change curve is constructed for each pixel along the wall thickness direction. By using frame-by-frame differencing, the signal value change amplitude between consecutive frames is compared, and abrupt change points in the signal curve are extracted. For example, if the grayscale value changes from 80 to 120, and a change of more than 10 grayscale units between frames is identified, it is marked as a critical change node. The total signal change amplitude between every two critical nodes is statistically analyzed, evenly divided by time intervals, and the average rate of change within each segment is further calculated. For example, if the total signal change is 45 grayscale units between nodes with a 3-frame interval, the average rate of change is 15 grayscale units per frame. The signal change rates of all pixels along the entire line are summarized to obtain the average rate trend for each wall thickness direction. All the calculation results for all lines form a sequence set reflecting the local tensile stress change trend. This sequence can be used to characterize the distribution of contraction and expansion intensity of the myocardial wall thickness at each location. In an example, if there are 30 wall thickness lines, 30 rate values ​​will be formed. If the signal change value of one of the lines is high, exceeding 2500 units per second, it can be regarded as a high change region; while the line with a change value less than 1500 units per second is classified into a low change range. The numerical range can be defined according to the upper and lower quartiles of the statistical sample.

[0026] S103: Call the trend value of the rate of change of tensile stress in the wall thickness region, construct the spatial mapping matrix of the trend of tensile stress change according to the spatial arrangement order in the wall thickness direction, arrange the spatiotemporal correspondence of the trend values, and obtain the structural array map of the tensile stress change. The trend values ​​of the tensile stress change rate in the wall thickness region are arranged spatially in the wall thickness direction. Each connection line is numbered sequentially from the inside to the outside of the image, constructing a spatial mapping matrix of trend values. The trend values ​​of each connection line within the entire analysis time window are filled in chronologically. The values ​​of the same connection line in multiple time periods constitute its time series. With the time series of all connections, a trend matrix graph is formed. The horizontal axis of this graph is the time series, and the vertical axis is the position of each connection line number in the wall thickness direction. Each matrix element represents the trend rate value of that connection line in a certain time period. For example, analyzing 30 frames of images in 1 second, assuming there are 20 connections in the wall thickness direction, the final graph dimension is 20 rows and 30 columns. To enhance the resolution of the graph, the values ​​can be normalized, mapping all trend values ​​to a standard range between 0 and 1. For example, when the minimum value is 1000 and the maximum value is 3000, a trend value of 2000 is normalized to 0.5. For trend value interval division, the low-rate zone can be set to 0 to 0.3, the medium-rate zone to 0.3 to 0.7, and the high-rate zone to 0.7 to 1. The final trend map represents the distribution trajectory of tensile stress intensity at the locations of the lines connecting different wall thicknesses throughout the entire observation period, providing basic data support for subsequent spatial distribution analysis or pattern recognition.

[0027] Please see Figure 3The specific steps of S2 are as follows: S201: Based on the tensile stress variation trend value of the spatial level in the tensile stress variation structure array map, determine the positive and negative relationship of the corresponding tensile stress variation direction of adjacent levels in the time series, filter the positions where the direction change is reversed, mark the map coordinate value of the position, and obtain the tensile stress direction reversal coordinate. Based on the tensile stress trend values ​​corresponding to each spatial level in the tensile stress variation structure array map, it is necessary to extract the map data of each frame in chronological order and compare the trend values ​​of two spatially adjacent levels at the same time point. By judging the positive and negative signs of the trend values ​​between adjacent levels, it is possible to identify whether there is a reversal of direction. For example, in frame 25, the trend values ​​of layer 10 and layer 11 are 2300 and -2100, respectively. The former is a positive value and the latter is a negative value, indicating that the trend direction has reversed. When performing the judgment, it is necessary to traverse all vertical adjacent level combinations in the map frame by frame, record the trend value sign status of each group in the same frame, and use the product of trend values ​​to determine whether there is a reversal. If it is negative, it is identified as a reversal position, and the row and column coordinate information of the corresponding map is recorded. To eliminate isolated reversal misjudgments caused by noise among all possible reversal points, a spatial density threshold needs to be set to filter out areas with excessively sparse reversal points. The number of reversal points can be counted within each sliding window. For example, a square window with 5 rows and 5 columns can be used to traverse the entire map. If the number of reversal points in the window is less than 3, the location is considered discontinuous and excluded. This threshold should be flexibly set according to the actual size of the map; for example, a threshold between 2 and 5 is reasonable when the map is 100 rows and 100 columns. After completing the reversal relationship judgment and density filtering, the final set of points obtained is the tensile stress direction reversal coordinate point, containing both spatial location and time series index information.

[0028] S202: Call the tensile stress direction inversion coordinates, determine the connection relationship between adjacent points according to the continuity of spatial distribution, extract the boundary of the connected region formed by the densely connected region, generate the contour line of the closed shape according to the connection order of the boundary line, and obtain the tensile stress inversion connected boundary contour. Based on the tensile stress direction reversal coordinate information, the positional relationships of all points in the map are first analyzed. An eight-neighborhood approach is used to identify whether there are connected paths between each reversal point. During processing, neighboring points within a distance of no more than 1 pixel around each reversal point are searched. If a neighboring point is also a reversal point, the two points are considered related, and this process is expanded layer by layer to form connected regions. All interconnected points are grouped into the same group. For each group of connected regions, its spatial boundary is extracted, and a contour extraction method is used to obtain the edge coordinate sequence. These are then organized into closed curves according to the order of the connecting points. When judging the closure of the boundary, the distance between the starting and ending points is used as the standard. If the straight-line distance between two points is less than or equal to the square root of 2 pixels, it is considered closed. For example, if a connected region contains approximately 30 points, with the starting point (x=45, y=60) and the ending point (x=46, y=59) connected sequentially, and the distance between them is less than or equal to 1.5 pixels, this connected region can be considered to form a valid closed contour. After all contours have been extracted, the number of points contained within each contour needs to be recorded, and the region number to which each contour belongs, such as R1, R2, etc., needs to be marked for subsequent mapping in the original image. The final result is a set of tensile stress inversion connected boundary contours composed of multiple closed region boundaries. Each contour contains a list of coordinate points and the basic attributes of the connected region.

[0029] S203: Call the tensile stress inversion connected boundary contour, map the boundary contour line to the original image coordinate area, draw the contour line shape according to the image position relationship of each closed boundary, mark the covered image area, distinguish the label number corresponding to different closed boundaries, and obtain the structural anomaly image label area. After obtaining the tensile stress inversion connected boundary contour, the contour coordinates need to be converted from the index coordinate system in the tensile stress map to the actual pixel coordinate system of the original image. Coordinate mapping is performed according to the scaling ratio and position offset parameters used when the map was generated. For example, the pixel position in the original image corresponding to row 50, column 80 in the map is (x=200, y=160). This conversion process is performed on all coordinate points in each contour to form a complete sequence of original image contour coordinates. On the original image, closed curves are drawn sequentially according to the contour coordinates to form closed graphic regions. The regions inside the graphics are labeled, which can include filling with color or highlighting the boundaries. Each region is assigned a unique number, such as region A1, A2, etc. For the image region covered by each closed graphic, its position coordinate range and estimated area value need to be recorded. The area can be calculated based on the polygonal contour formed by the contour points. For example, a boundary region composed of 40 points encloses an area of ​​approximately 9 mm². The area value is recorded after conversion according to pixel density. Finally, multiple anomaly regions are formed on the original image, each with a clear number, location, contour coordinates and area data, which serve as the output of the anomaly image annotation region.

[0030] Please see Figure 4 The specific steps of S3 are as follows: S301: Based on the image spatial coordinates of points in the annotated region of the structural anomaly image, extract the signal change values ​​of the corresponding points in the perfusion image sequence at different times, and generate a perfusion signal time series; Based on the image coordinates of points in the annotated regions of structural anomalies, the process first involves traversing all pixels within each closed region to extract their specific positions in the two-dimensional image coordinate system. These pixel coordinates are then mapped to the corresponding frame positions in the entire perfusion image sequence. In each frame, the signal intensity or grayscale value of these points is extracted to construct a time-series signal set corresponding to each frame. Taking a point (120, 85) in an anomaly region as an example, in a total of 40 frames of the perfusion sequence, the signal intensities of this point in each frame are 145, 152, 158, 164, and so on, until the last frame, forming a complete signal sequence. This process requires point-by-point extraction within all annotated regions and organizing them into a multi-dimensional matrix structure. Rows represent different pixels, and columns represent signal values ​​at different times. For example, if there are 25 pixels, a 25x40 matrix is ​​formed, with each element representing the signal value of the corresponding point in the corresponding frame. The time interval between each frame acquisition also needs to be recorded. If the acquisition interval is 2 seconds, the entire sequence coverage time is 80 seconds. Ultimately, through this method, each point within a structural anomaly region can form a corresponding time series data set of perfusion signals, possessing both spatial coordinates and time series information attributes.

[0031] S302: Call the injection signal time series, calculate the signal change rate value of the point based on the signal change trend between consecutive time points, and filter the points whose rate exceeds the injection signal rate threshold to obtain the set of abnormal injection response rate points. After acquiring the injection signal time series, the trend of signal value changes over time at each point needs to be calculated. This is done by comparing and analyzing the rate of change of signal value at each moment frame by frame. The difference between the signal values ​​at each consecutive time point is calculated and divided by the time interval to obtain the rate of change for that point. For example, if a point has a signal value of 180 in frame 10 and 195 in frame 11, with a frame interval of 2 seconds, the rate is 7.5 units per second. Each point forms a rate sequence, which is then filtered to identify points with rates of change exceeding the normal range. To achieve this, a threshold for the injection signal rate needs to be set. This value can be set based on the average rate and dispersion of all points. For example, if the average rate for the entire image is 5.2 units per second and the dispersion range is 2.1 units per second, the rate threshold can be set to 8.3 units per second. When the rate of a point exceeds 8.3 units per second in any frame, it is determined that the point has an abnormal rate, and the corresponding frame number and image coordinates are recorded. This process requires a full traversal and judgment of all points, and finally a set of all points that meet the rate anomaly criteria are selected. The points are then structured according to image coordinates, frame number, and abnormal rate value to form a complete set of perfusion response rate anomaly points.

[0032] S303: Call the set of abnormal points of the perfusion response rate, mark the boundary of the response area according to the distribution relationship of the abnormal points in the image space, complete the area drawing on the image, and obtain the functional partition image fragment; The system calls upon a set of abnormal points in the perfusion response rate and performs distribution analysis based on their positional relationships in the image coordinate system. First, all abnormal points are marked as valid pixels on the image. Then, an eight-neighborhood search is used to find connectivity between adjacent points, merging points belonging to the same region to form connected regions. Boundary detection is performed on each connected region to extract contour edge points. These contour points are then connected sequentially along the coordinate axes to construct a closed shape, forming the contour boundary line of the functional response region. All contour boundary points must be converted to integer pixel values ​​in the image coordinate system to ensure accurate drawing of closed regions on the original image. Boundary closure is determined by a rule that the distance between the first and last points is less than 2 pixels. For each region, the number of points and approximate area contained in its contour are calculated. For example, a connected region containing 120 pixels has an area of ​​approximately 30 mm². The region is assigned the number F1, and its relevant attributes are recorded. This process is repeated to number, label, and draw all connected regions, ultimately forming multiple visualized functional partition image segments on the image. Each region possesses complete contour coordinates, pixel count, area size, and a unique identifier.

[0033] Please see Figure 5 The specific steps of S4 are as follows: S401: Based on the distribution information of all points in the functional partition image segment at the image structure level, identify the structural level label of the point, compare the positions according to the continuity trend of point distribution between adjacent levels, determine the continuous level segments with order differences, and establish a sequence of structural level trend change segments. Based on the distribution information of all points in the functional partition image fragment at the image structural level, the first step is to identify the structural level label of each point according to the image structural template. This template can preset the grayscale range, texture features, or spatial location distribution corresponding to each level. In anatomical images, for example, L1 to L6 are set as the cortical region, L7 as the white matter region, and L8 as the ventricle region. For a point with coordinates such as 180, 95, its grayscale features and structural matching results can be used to identify that it belongs to the L3 level. After assigning labels to all points, the points are grouped and statistically analyzed according to structural level. The number of points and spatial coverage within each group are counted. Then, the continuity analysis of the point distribution between adjacent levels is performed by comparing the number of points in adjacent levels. The magnitude of change and the spatial continuity of the points are used to determine whether they constitute a continuous hierarchical segment with small sequential differences. For example, if the number of points between L3 and L4 is similar, their distribution is close, and there are no obvious gaps, then they are considered a continuous hierarchical segment; otherwise, they are considered a broken or discontinuous segment. In the analysis process, judgment parameters such as the ratio of the number of points and the product of the average distance between points can be introduced to set a continuity judgment threshold. If the value of this parameter is greater than 0.5, it can be classified as a continuous segment. Finally, all adjacent levels are traversed one by one, and the combination of levels that meets the continuity requirements is marked. A trend change segment sequence is constructed according to its arrangement order in the structural hierarchy. Each segment includes the starting level, the ending level, the number of points contained, and the average spatial density, which are used for further processing of subsequent image structural changes.

[0034] S402: Call the sequence of trend change segments in the structural hierarchy, combine it with the coordinate information in the image space, extract the two-dimensional position values ​​of the points in the image space within each segment, aggregate the coordinates of the points within the same segment, and generate a set of segment position coordinates. After calling the structural hierarchy trend change segment sequence, it is necessary to extract the two-dimensional coordinate information of all points within each segment in the image space. First, the range of points included is determined according to the range of hierarchy numbers given in the segment sequence. For example, if a segment consists of L3 to L5, then all points belonging to L3, L4, and L5 are selected from the identified structural hierarchy labels. The x and y coordinate values ​​of these points in the image are read and organized into a data set. To avoid aggregation bias caused by inconsistent coordinate scales, the coordinates of all points need to be processed to a uniform pixel scale to ensure consistency between segments. Then, all points are... Point coordinates within the same level segment are merged into a unified set. During the aggregation process, duplicate point coordinates are removed. For example, two points that differ by less than 1 pixel are considered to be the same location and are merged. In the final generated set of segment position coordinates, each element contains the horizontal and vertical coordinates of the point and its corresponding segment label information. After aggregation, spatial sorting or numbering of the set can further enhance the manageability of the data. For example, 325 points were extracted from segments L3 to L5. After merging and deduplication, 290 valid points remain, forming the final set of segment position coordinates used for structural analysis.

[0035] S403: Call the set of location coordinates of the layer segment, calculate the spatial change trend value between the location sets based on the spatial gradient difference of the point coordinates in the location sets between adjacent layer segments, and map the trend value into three-dimensional coordinate form to obtain the coordinate set of the inter-layer response gradient surface; The specific formula for calculating the spatial gradient difference of point coordinates in the location set between adjacent layers is as follows: ; Calculate the spatial gradient change characteristic value, call the layer position coordinate set, calculate the spatial change trend value between the position sets based on the spatial gradient difference of the point coordinates in the position sets between adjacent layer segments, and map the trend value into three-dimensional coordinate form to obtain the inter-layer response gradient surface coordinate set; in, The spatial gradient change characteristic value represents the relationship between layer i and layer j. The x-coordinate represents the k-th point in segment i. This represents the x-coordinate of the k-th point in segment j. This represents the y-coordinate of the k-th point in segment i. This represents the y-coordinate of the k-th point in segment j. The z-coordinate represents the k-th point in segment i. The z-coordinate represents the k-th point in segment j. The spatial gradient value of the k-th point in layer i along the local normal direction is represented by . The spatial gradient value of the k-th point in segment j along the local normal direction is represented by . This represents the number of points involved in the calculation within the two layer location sets. For point index numbering; The spatial coordinates of point k in layer i are collected using a 3D laser scanning device (such as RIEGLVZ-400i) in the mine tunnel or geological section, with a point spacing of 0.5 meters and the coordinate accuracy of each point controlled within ±2 millimeters. The coordinates corresponding to point P1 are selected as follows: rice, rice, rice; The spatial coordinates of point P2 in segment j are obtained using the same method as described above. rice, rice, rice; The spatial gradient value along the normal direction at point k in the layer segment is calculated by converting the slope of the normal tangent of the interpolation model in the spatial section plane. Based on the interpolation gradient data of the section plane, the unit is meters (m⁻¹), and it is measured as follows: , ; Number of points: n=1 pairs of points are included in the calculation within the sampling range. This is only a single-point example. Substituting into the formula, the calculation is as follows: Part 1: Calculating the Euclidean distance between points: ; ; The second part calculates the gradient difference multiplied by the squared difference of the planar coordinates: ; ; ; Substitute both parts into the formula calculation terms: ; Since only one point is involved in the calculation, n=1, therefore: ; The results indicate that, under the combined influence of overall coordinate offset and gradient variation, the spatial response characteristics between points P1 and P2 exhibit a comprehensive gradient spatial difference value of approximately 0.815. This value represents the level of spatial structural difference between the two adjacent segments on the central coordinate plane, providing the basis for the initial gradient response quantity for subsequent spatial trend vector direction mapping. This parameter serves as one of the input values ​​for the response gradient intensity of cells in the trend mapping matrix in subsequent steps, participating in determining the basis for reconstructing the response surface morphology function of each spatial unit.

[0036] This formula reflects the spatial response differences between adjacent layer points by constructing a difference structure between two terms. The first term is the three-dimensional Euclidean distance, representing the actual geometric displacement of corresponding points in space within two layers. It uses the square root of the sum of squares to reflect the cumulative effect of the distance between coordinates, ensuring that the composite effect of displacements in each direction has a true geometric meaning. The second term is the product of the gradient change amplitude and the square of the planar projection displacement, representing the weighted influence of the spatial gradient difference within the local coordinate projection range. Its structure uses the absolute value of the gradient difference as a scale adjustment factor, and the squared coordinate difference reflects the local change trend amplitude brought about by the point distribution density. The multiplication structure realizes the expansion or convergence adjustment of the influence range of local gradient changes on the overall response. The two terms are used to construct a difference function to measure the degree of inconsistency between geometric displacement and gradient-dominated response. The absolute value operation eliminates directional interference, ensuring that the overall output only expresses the intensity change level. Finally, summation is introduced and the average is taken over the number of participating points to avoid the influence of outliers at individual points, so as to obtain a stable feature value that reflects the spatial structure differences between layers. Spatial gradient variation eigenvalues ​​are used to quantify the degree of difference in local response in spatial structure between adjacent layers. They reflect the offset characteristics of corresponding points in different layers in both geometric position and spatial gradient direction. This eigenvalue combines the actual three-dimensional coordinate displacement and the amplitude of local gradient change. By comprehensively calculating the degree of difference between the two, it can characterize the sensitive areas of structural distortion, deformation or uneven interface response in a continuous spatial region. The larger the value, the more significant the spatial morphological change between layers. In spatial modeling, response surface reconstruction and geological interface identification, this value can serve as an important indicator for judging the trend of structural change, the range of fault displacement or the boundary of material deformation.

[0037] Please see Figure 6 The specific steps of S5 are as follows: S501: Based on the intersection coordinates between the interlayer response gradient surface coordinate set and the labeled area of ​​the structural anomaly image, extract the structural feature values ​​and functional feature values ​​at the corresponding coordinate points, organize them into a unified structure, and generate a set of structural and functional feature values. Based on the intersection coordinates between the interlayer response gradient surface coordinate set and the annotated regions of the structural anomaly image, it is necessary to first unify the two data sources to a two-dimensional coordinate system in the image pixel space and ensure consistent coordinate resolution. By reading the annotated boundary information of the structural anomaly region (e.g., top left corner 160, 110, bottom right corner 180, 130), a rectangular boundary range of the anomaly region is constructed. Then, each set of three-dimensional coordinates in the interlayer response gradient surface coordinate set is read one by one, and the horizontal and vertical coordinates are extracted. These are then compared with the anomaly annotated region to determine the coordinate range. If the horizontal coordinate of a point is between 160 and 180, and the vertical coordinate is between 110 and 130, the point is considered an intersection point and marked as a candidate coordinate. Subsequently, based on these coordinates, the anomaly region is located in the original structural image. The structural feature values ​​are obtained from the location, such as reading pixel grayscale and texture direction changes. If the structural value is 145, the functional signal value at the same coordinate in the functional image data channel is read. For example, the functional activation signal is 0.87. The structural value and functional value are extracted. Each pair of coordinates and feature values ​​is recorded as an independent data item. The format includes the horizontal coordinate, vertical coordinate, layer position, structural value and functional value. For example, the coordinates 165, 115, 5 correspond to a structural value of 145 and a functional value of 0.87, which is recorded as a single entry. After traversing all intersection points to complete feature extraction, all records are integrated into a unified data set for subsequent processing. This set has structural consistency and complete spatial location information and multi-dimensional feature description.

[0038] S502: Call the set of structural and functional feature values, and based on the numerical performance of structural and functional feature values ​​at coordinate points, filter the points that simultaneously meet the structural feature threshold and the functional feature threshold to obtain the coordinate set of the threshold-satisfied area. After calling the set of structural and functional feature values, a dual-condition filtering operation needs to be performed based on the structural and functional feature values ​​of each point. Before filtering, the structural feature threshold is first set based on the statistical results of the data distribution of the overall structural image. For example, if the structural feature value is distributed in the range of 90 to 170 in the whole image with an average of 130, the filtering range can be set to 115 to 145. The functional feature threshold is set based on the difference between the signal background value and the activation value in the functional image. For example, if the functional value in the whole image is between 0.4 and 1.0 with an average of 0.65, the lower limit of the filtering can be set to 0.75. After setting, each record in the set of structural and functional feature values ​​is read one by one to determine whether its structural value is between 115 and 145 and whether its functional value is higher than 0.75. If both conditions are met, the point is considered to be qualified. The filtering criteria are used to retain points. If a point has a structure value of 138 and a function value of 0.79, it is considered a valid point. If another point has a structure value of 149 or a function value of 0.68, it is discarded. During the filtering process, a marker field is added to each point to record whether it passes the judgment. After processing all points, the coordinates of all points that meet the filtering requirements are extracted to generate a new set of regional coordinates, which serves as the threshold set for satisfying regional coordinates. This set of coordinates only includes points whose structural and functional characteristics simultaneously meet the specified interval, and is suitable for further spatial structure aggregation processing. In addition, the threshold setting can be optimized and adjusted by combining the statistical results of multiple sample images, or the impact of different threshold combinations on the density of filtered points and the spatial distribution of regions can be tested by scrolling and sliding, which helps in the selection of parameters for the filtering strategy.

[0039] S503: Based on the threshold-satisfied region coordinate set, identify the related adjacent points, generate a mask matrix for the corresponding image positions, and superimpose it onto the original image space to obtain the positioning mask output area; Based on the threshold-compliant region coordinate set, it is necessary to identify combinations of points with spatial adjacency relationships and construct connection relationships for region generation. First, spatial adjacency rules are defined, typically using a four-neighbor or eight-neighbor standard. That is, if a point has another point that meets the filtering criteria in the horizontal, vertical, or diagonal direction, it is determined that the two are spatially related and merged into the same connection region. A traversal method is used to scan all points in the coordinate set, establishing an association lookup table for each point to record whether there are adjacent points that meet the criteria. If so, the connection label is updated, ultimately forming several relatively independent associated regions. Based on this, a mask matrix of uniform image size is constructed. The matrix is ​​initially... The value is set to 0, and each point that meets the threshold is assigned a value of 1 in the matrix. If there are 50 adjacent points in a region, they form a dense block of regions with a value of 1 in the matrix. The matrix dimension must be consistent with the original image to ensure that the position mapping is unbiased. After the mask is constructed, the mask matrix is ​​superimposed on the original image pixel by pixel. That is, each pixel value in the original image is multiplied by the mask value. If the mask value is 1, the original pixel is retained. If it is 0, the position is cleared. Finally, the localization mask output area is generated, which presents an image area that meets the dual threshold screening conditions of structure and function and has spatial continuity, providing basic localization results for further image recognition or region feature extraction.

[0040] Please see Figure 7 A system for locating myocardial infarction regions based on imaging, including: The tensile stress trend construction module obtains the coordinate region of the wall thickness direction in the left ventricular image sequence, extracts the signal intensity value of the coordinate point in the continuous image frame, calculates the change trend of the signal in the time series, constructs the change trend set corresponding to the slice according to the spatial arrangement order of the image slice, and generates a tensile stress change structure array map. The structural anomaly identification module is based on the tensile stress change structural array map. It extracts the trend direction of the same coordinate points in adjacent image slices, identifies the points where the direction is reversed, summarizes the regions formed by the reversal points that are continuously distributed in space, extracts the region boundaries and marks them on the image, and generates the structural anomaly image annotation region. The perfusion response extraction module calls the points in the annotated area of ​​the structural anomaly image, extracts the signal time change of the corresponding points in the perfusion image sequence, calculates the signal change rate, extracts the set of points whose rate exceeds the set threshold, marks them in the image to form regional segments, and generates functional partition image segments. The interlayer gradient generation module identifies region segments with continuous numbering based on the image slice numbers of points in the functional partition image segments, extracts the spatial distribution positions and records the corresponding coordinates, and generates an interlayer response gradient surface coordinate set. The location index screening module extracts the structural and functional feature values ​​of each point based on the intersection of the interlayer response gradient surface coordinate set and the structural anomaly image annotation area, calculates the index value, screens the set of points that meet the set criteria, marks them on the image in the form of a mask, and generates the location mask output area.

[0041] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for locating myocardial infarction regions using imaging, characterized in that, Includes the following steps: S1: Extract the image coordinate region in the wall thickness direction of the left ventricle image, combine the signal intensity changes of the points in the continuous frame sequence, construct the tensile stress change trend sequence, and arrange it according to the spatial hierarchy to generate a tensile stress change structure array map. S2: Based on the direction of tensile stress change in the tensile stress change structure array map, identify the coordinates of the direction reversal points, summarize them to form a continuous distribution area, extract the boundary shape, and generate the structural anomaly image annotation area. S3: Based on the spatial coordinates within the marked area of ​​the structural anomaly image, extract the corresponding signal time change in the perfusion image sequence, extract the set of points whose rate exceeds a set threshold, mark the response area, and generate functional partition image segments. S4: Utilize the distribution of points in the functional partitioned image segments at the image structure hierarchy to identify regions that exhibit trend changes in hierarchical continuity, construct a set of locations in the image space, and generate a set of interlayer response gradient surface coordinates. S5: Based on the intersection coordinates of the interlayer response gradient surface coordinate set and the structural anomaly region, extract the point structure and functional feature values, construct the positioning reference index, screen the regions that meet the conditions, and mark them in the form of a mask to generate the positioning mask output region.

2. The method for locating myocardial infarction regions by combining imaging according to claim 1, characterized in that, The tensile stress variation structure array map includes a cluster of temporal tensile stress curves, a spatial hierarchy index, and a pixel mapping relationship. The structural anomaly image annotation region includes a set of direction reversal points, continuous distribution boundaries, and anomaly region label attributes. The functional partition image segment includes a time series of injection signals, a set of suprathreshold velocity points, and a response region label layer. The interlayer response gradient surface coordinate set includes hierarchical continuity segments, a cluster of segmented spatial locations, and gradient surface fitting parameters. The positioning mask output region includes structural functional composite indices, intersection point indexes, and a mask pixel set.

3. The method for locating myocardial infarction regions by combining imaging according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Extract the image coordinate region in the wall thickness direction of the left ventricle image, identify the spatial distribution of the inner and outer wall edges in the image sequence, determine the central axis coordinate region based on the connection between the edges, record the signal intensity changes of the pixels in the region in consecutive image frames, and obtain the image signal intensity sequence value in the wall thickness direction. S102: Call the image signal intensity sequence value in the wall thickness direction, construct the signal curve sequence of pixel points changing with time, extract key change nodes in the curve, calculate the degree of signal change between nodes, and obtain the trend value of the tensile stress change rate in the wall thickness region. S103: Call the trend value of the tensile stress change rate in the wall thickness region, construct a spatial mapping matrix of the tensile stress change trend according to the spatial arrangement order in the wall thickness direction, arrange the spatiotemporal correspondence of the trend values, and obtain the tensile stress change structure array map.

4. The method for locating myocardial infarction regions by combining imaging according to claim 3, characterized in that, The specific steps of S2 are as follows: S201: Based on the tensile stress change trend value of the spatial level in the tensile stress change structure array map, determine the positive and negative relationship of the corresponding tensile stress change direction of adjacent levels in the time series, filter the position points where the direction change is reversed, mark the map coordinate value of the position points, and obtain the tensile stress direction reversal coordinates. S202: Call the tensile stress direction reversal coordinates, determine the connection relationship between adjacent points according to the continuity of spatial distribution, extract the boundary of the connected region formed by the densely connected region, generate the contour line of the closed shape according to the connection sequence of the boundary line, and obtain the tensile stress reversal connected boundary contour. S203: Call the tensile stress inversion connected boundary contour, map the boundary contour line to the original image coordinate area, draw the contour line shape according to the image position relationship of each closed boundary, mark the covered image area, distinguish the label number corresponding to different closed boundaries, and obtain the structural anomaly image label area.

5. The method for locating myocardial infarction regions by combining imaging according to claim 4, characterized in that, The specific steps for S3 are as follows: S301: Based on the image spatial coordinates of the points in the annotated region of the structural anomaly image, extract the signal change values ​​of the points in the corresponding perfusion image sequence at different times, and generate a perfusion signal time series; S302: Call the perfusion signal time series, calculate the signal change rate value of the point based on the signal change trend between consecutive time points, and filter the points whose rate exceeds the perfusion signal rate threshold to obtain a set of abnormal perfusion response rate points. S303: Call the set of abnormal points of the perfusion response rate, mark the boundary of the response area according to the distribution relationship of the abnormal points in the image space, complete the area drawing on the image, and obtain the functional partition image fragment.

6. The method for locating myocardial infarction regions by combining imaging according to claim 5, characterized in that, The specific steps of S4 are as follows: S401: Based on the distribution information of all points in the functional partition image segment at the image structure level, identify the structural level label of the point, compare the positions according to the continuity trend of point distribution between adjacent levels, determine the continuous level segments with order differences, and establish a sequence of structural level trend change segments. S402: Call the sequence of trend change segments of the structure hierarchy, combine it with the coordinate information in the image space, extract the two-dimensional position value of the point in the image space within each segment, aggregate the point coordinates within the same segment, and generate a set of segment position coordinates. S403: Call the set of location coordinates of the layer segment, calculate the spatial change trend value between the location sets based on the spatial gradient difference of the point coordinates in the location sets between adjacent layer segments, and map the trend value into three-dimensional coordinate form to obtain the inter-layer response gradient surface coordinate set.

7. The method for locating myocardial infarction regions by combining imaging according to claim 6, characterized in that, The specific steps of S5 are as follows: S501: Based on the intersection coordinates between the interlayer response gradient surface coordinate set and the annotated area of ​​the structural anomaly image, extract the structural feature values ​​and functional feature values ​​corresponding to the coordinate points, organize them into a unified structure, and generate a set of structural and functional feature values. S502: Call the set of structural and functional feature values, and based on the numerical performance of the structural and functional feature values ​​at coordinate points, filter the points that simultaneously meet the structural feature threshold and the functional feature threshold to obtain the coordinate set of the threshold-satisfied region. S503: Based on the threshold satisfying the regional coordinate set, identify the related adjacent points, generate a mask matrix for the corresponding image positions, superimpose it onto the original image space, and obtain the positioning mask output area.

8. The method for locating myocardial infarction regions by combining imaging according to claim 1, characterized in that, The wall thickness direction refers to the perpendicular anatomical direction along the left ventricular wall to the epicardium; The stress variation trend refers to the stress estimation curve calculated by inter-frame variation based on the change of image signal intensity of myocardium during the contraction cycle. The tensile stress variation structure array map refers to a matrix map formed by hierarchically arranging the pixel positions in the image space with the corresponding tensile stress estimation results. The structural anomaly image annotation region refers to the region in the image space marked with a closed boundary where the direction of tensile stress change is abnormally concentrated, serving as an image localization reference for suspected myocardial injury or lesion areas.

9. The method for locating myocardial infarction regions by combining imaging according to claim 1, characterized in that, The perfusion imaging sequence refers to a set of continuous image frames acquired using magnetic resonance or CT technology that reflects the distribution process of contrast agents in myocardial tissue. The signal time refers to the rate of change of signal intensity with respect to time at each image point in the perfusion sequence; The rate exceeding the set threshold means that the value of the signal time derivative is greater than the reference standard value of the perfusion rise rate set by the system. The threshold can be obtained through clinical sample statistics or literature settings. The functional partition image segment refers to a set of regions in the image that show significant changes in perfusion response and high signal derivative values, which are used as functionally active regions for spatial segmentation and labeling; The hierarchical continuity refers to the sequential relationship of the slice sequences of myocardial images arranged from the inside to the outside in the wall thickness direction; The interlayer response gradient surface coordinate set refers to a spatial set composed of pixel positions in multiple image layers that exhibit response change trends, forming a reference surface for judging the depth distribution of regional changes. The structural and functional characteristic values ​​refer to the tensile stress estimation value, the infusion time delay value, and the infusion rise rate corresponding to the image points; The positioning reference index refers to a quantitative standard for judging positioning, formed by structural and functional characteristic values; The masking method refers to converting the recognition result area into a binary region in the image layer by using image masking.

10. A myocardial infarction region localization system combining imaging, characterized in that, The system is used to implement the method for locating myocardial infarction regions by combining imaging as described in any one of claims 1-9, the system comprising: The tensile stress trend construction module obtains the coordinate region of the wall thickness direction in the left ventricular image sequence, extracts the signal intensity value of the coordinate point in the continuous image frame, calculates the change trend of the signal in the time series, constructs the change trend set corresponding to the slice according to the spatial arrangement order of the image slice, and generates a tensile stress change structure array map. Based on the tensile stress change structure array map, the structural anomaly identification module extracts the trend direction of the same coordinate points in adjacent image slices, identifies the points where the direction is reversed, summarizes the regions formed by the spatially continuous reversal points, extracts the region boundaries and marks them on the image, and generates the structural anomaly image annotation region. The perfusion response extraction module calls the points in the annotated area of ​​the structural anomaly image, extracts the signal time change of the corresponding points in the perfusion image sequence, calculates the signal change rate, extracts the set of points whose rate exceeds the set threshold, marks them in the image to form regional segments, and generates functional partition image segments. The interlayer gradient generation module identifies region segments with continuous numbers based on the image slice numbers of the points in the functional partition image segments, extracts the spatial distribution positions and records the corresponding coordinates, and generates an interlayer response gradient surface coordinate set. The positioning index filtering module extracts the structural and functional feature values ​​of each point based on the intersection points of the interlayer response gradient surface coordinate set and the structural anomaly image annotation area, calculates the index values, filters the set of points that meet the set criteria, marks them on the image in the form of a mask, and generates a positioning mask output area.

Images