Neurology data analysis method and device and medium

Abstract

The invention provides a neurology data analysis method and device and a medium, and relates to a data analysis technology. A database is constructed by combining data of a plurality of cerebral palsy patients, and medical images such as CT images and symptom description data of the patients are collected; therefore, the pathological part of the current patient can be predicted in combination with the data in the database and the comparison result of the image data acquired by the acquisition device for the current patient, and the display image is generated to display the predicted illness state for the patient, so that doctors can be assisted in diagnosis, and more suitable diagnosis and treatment suggestions are provided for the patient.

Description

Technical Field

[0001] The present invention relates to data analysis technology, and in particular to a neurology data analysis method, equipment and medium. Background Art

[0002] In today's medical field, especially in the diagnosis and treatment of neurological diseases, early and accurate prediction of the disease is extremely critical. Predicting the pathological site and the development trend of the disease as early as possible and accurately can win valuable treatment opportunities for patients, significantly improving the treatment effect and the patient's quality of life. In actual clinical scenarios, doctors face complex and diverse patient symptoms and data. Currently, when patients seek medical attention, traditional diagnostic methods rely primarily on the physician's experience. When a patient presents with symptoms like numbness in the hands or facial paralysis, doctors can infer the likely condition based on their experience. However, due to the complexity and diversity of symptoms, different patients may present with different symptoms for the same disease, and many early symptoms are atypical. This method is prone to misdiagnosis or missed diagnosis, and cannot accurately predict the location of the pathology.

[0003] Therefore, there is an urgent need for an efficient and accurate prediction method to provide a reference for formulating personalized diagnosis and treatment plans for patients. Summary of the Invention

[0004] The present invention provides a neurology data analysis method, device and medium, which can improve the accuracy and efficiency of prediction and provide a reference basis for formulating personalized diagnosis and treatment plans for patients.

[0005] A first aspect of the present invention provides a neurology data analysis method, comprising: Acquire patient data, replace the real structure in the patient data with a virtual structure, add data tags and store in a database; Acquire a sensing image and a visible image corresponding to a target part of the current patient, and determine a region of interest in the visible image based on temperature distribution information in the sensing image; Identify spatial metric parameters corresponding to structural points in the region of interest; According to the comparison result of the spatial measurement parameters and the historical storage parameters in the database, the predicted pathological site of the current patient is determined, and a predicted display image is generated.

[0006] Optionally, in a possible implementation of the first aspect, obtaining patient data, replacing a real structure in the patient data with a virtual structure, adding a data tag, and storing the data in a database includes: Obtaining a facial image of the patient according to the patient data, and extracting the coordinates of key points of each real structure in the facial image; Based on the key point coordinates, calculating key parameters of each real structure and generating a virtual structure corresponding to the key parameters, where the key parameters include at least shape parameters, position parameters, and angle parameters; After the real structure in the facial image is replaced with the virtual structure, a data label is added and stored in a database.

[0007] Optionally, in a possible implementation of the first aspect, acquiring a sensing image and a visible image corresponding to a target part of the current patient, and determining a region of interest in the visible image based on temperature distribution information in the sensing image includes: Dividing the visible image into a plurality of symmetrical structural regions, the visible image including an initial image and a dynamic image of the current patient under various action instructions; determining a temperature difference between symmetrical structural regions based on temperature distribution information in a sensing image corresponding to the visible image; The structural area with a temperature difference greater than the temperature threshold is determined as a suspicious area. The abnormal value of the suspicious area in different visible images is calculated based on the abnormal level of the temperature difference. The suspicious area with an abnormal value greater than the preset abnormal value is determined as the focus area.

[0008] Optionally, in a possible implementation manner of the first aspect, dividing the visible image to obtain a plurality of symmetrical structural regions includes: Extracting the facial contour of the visual image and generating a vertical centerline of the facial contour; Identifying a plurality of structural extreme points in the facial contour, and determining an offset point at a preset distance from the structural extreme point according to a preset direction corresponding to each structural extreme point; A dividing line passing through the offset point and perpendicular to the vertical center line is generated, and multiple bilaterally symmetrical structural regions are obtained according to the dividing lines corresponding to the structural extreme points with the same structural attributes.

[0009] Optionally, in a possible implementation of the first aspect, the temperature difference between the symmetrical structural regions includes: The temperature difference between all symmetrical structural regions in the initial image and the temperature difference between structural regions corresponding to the action instructions in the dynamic image.

[0010] Optionally, in a possible implementation of the first aspect, calculating abnormality values of suspicious areas in different visible images based on the abnormality levels of the temperature differences includes: Traverse the temperature difference intervals corresponding to each preset level of the temperature difference, determine the preset level where the temperature difference is located and its corresponding abnormal level, and configure each abnormal level with a corresponding preset abnormal value; The preset abnormal values of the suspicious areas in different visual images are superimposed and calculated to obtain their corresponding abnormal values.

[0011] Optionally, in a possible implementation manner of the first aspect, identifying the spatial metric parameters corresponding to the structure points in the region of interest includes: Extracting structural points in the region of interest, and obtaining the distance between each structural point and the vertical midline of the facial contour; Arrange the structure points on the corresponding side from small to large according to the spacing to obtain a structure point sequence, and determine the connection order of each structure point according to the comparison result of the preset sequence on the corresponding side of the focus area and the structure point sequence, wherein each preset structure point in the preset sequence is configured with a corresponding connection number; Multiple determination lines are obtained by connecting the structural points with the same connection number, and the spatial measurement parameters are obtained according to the angles of the determination lines.

[0012] Optionally, in a possible implementation of the first aspect, determining the predicted pathological site of the current patient based on a comparison result of the spatial measurement parameter and the historically stored parameters in the database, and generating a predicted display image, includes: Determining structural attributes of the region of interest, and extracting reference regions corresponding to the structural attributes in the facial images included in the data of each patient; Acquiring a historical angle of the reference area, wherein the historical storage parameters include the historical angle; Comparing the angle corresponding to the spatial measurement parameter with the historical angle, and determining the historical angle whose angle difference between the two is less than an angle difference threshold as the target angle; The historical pathological site corresponding to the target angle is determined to be the predicted pathological site of the current patient, a pathological display image is retrieved, and the predicted pathological site in the pathological display image is framed to obtain the predicted display image.

[0013] Optionally, in a possible implementation of the first aspect, determining a historical pathological site corresponding to a target angle as a predicted pathological site of a current patient, retrieving a pathological display image, and framing the predicted pathological site in the pathological display image to obtain the predicted display image includes: When the number of regions of interest is greater than a reference constant, obtaining a correlation region group corresponding to each predicted pathological site; Determining association labels corresponding to the structural attributes of the association region group, and determining a severity level of the predicted pathological site according to a preset level of the association labels; The predicted pathological part corresponding to the pathological display image is updated based on the area of the reference part corresponding to the severity level, and is framed to obtain the predicted display image.

[0014] Optionally, in a possible implementation of the first aspect, determining the reference area corresponding to the severity level by the following steps includes: Acquire multiple medical image data corresponding to the associated labels, and extract the pathological area of the part corresponding to the corresponding predicted pathological part in each medical image data; The average of each pathological area was calculated to obtain the baseline area.

[0015] According to a second aspect of the present invention, an electronic device is provided, comprising: a memory, a processor, and a computer program, wherein the computer program is stored in the memory, and the processor runs the computer program to execute the first aspect of the present invention and various methods that may be involved in the first aspect.

[0016] According to a third aspect of the present invention, a readable storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the first aspect of the present invention and various methods that may be involved in the first aspect.

[0017] The beneficial effects of the present invention are as follows: 1. The present invention can perform multi-dimensional data integration and precise analysis. The present invention uses acquisition equipment to simultaneously acquire thermal imaging images and optical images, combining the temperature distribution information in the thermal imaging images with the clear facial structure display of the optical images. By comparing the temperature differences in multiple areas of the face on both sides, it is possible to accurately locate areas of concern where pathological abnormalities may exist, effectively eliminating interfering factors such as congenital facial deviation, and greatly improving the accuracy of data prediction. For example, when judging facial deviation caused by acute cerebral infarction, thermal imaging images can detect temperature changes caused by abnormal local blood circulation or neural regulation. Combined with optical images, the abnormal area can be more accurately determined, providing a reliable basis for subsequent analysis.

[0018] 2. This invention protects patient privacy by extracting key coordinates of real facial structures from facial images, calculating key parameters, and generating virtual structures. This virtual structure is then replaced with a data tag and stored in a database. This process not only comprehensively collects patient information but also effectively protects the patient's real facial information throughout the entire data storage, transmission, and analysis process, meeting the stringent requirements for medical data security and compliance, and providing strong guarantees for the standardized use of medical data.

[0019] 3. The present invention uses historical data to improve prediction accuracy, builds a database, and collects medical images, symptom descriptions and other data from many cerebral palsy patients. When determining the predicted pathological site of the current patient, the spatial measurement parameters of the structural points in the area of interest are identified and carefully compared with the historical storage parameters in the database. For example, the angle information of the area of interest is compared with the historical angle, and the target angle with an angle difference less than a threshold is found, and the corresponding historical pathological site is determined as the predicted pathological site of the current patient. When the number of areas of interest is greater than the benchmark constant, the associated area group is obtained, the severity level is determined in combination with the associated label, and the pathological display image is updated and framed by calculating the area of the benchmark site. This method makes full use of historical data, greatly improves the accuracy of pathological site prediction and disease assessment, and provides an accurate basis for doctors to formulate scientific and reasonable diagnosis and treatment plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a schematic diagram of an application scenario provided by an embodiment of the present invention; Figure 2 This is a flowchart of a neurology data analysis method provided by an embodiment of the present invention; Figure 3 The figure is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0022] See also Figure 1 , is a schematic diagram of an application scenario provided by an embodiment of the present invention. The embodiment of the present invention will build a database based on the data of multiple cerebral palsy patients, collect the patients' medical images such as CT images, symptom descriptions and other data, so that the data in the database and the comparison results of the image data collected by the acquisition device for the current patient can be combined to predict the pathological site of the current patient, and generate a display image to show the predicted condition to the patient, thereby assisting the physician in diagnosis and providing the patient with more appropriate diagnosis and treatment suggestions. Among them, the acquisition device can be a camera, and the acquisition device in this embodiment can simultaneously acquire thermal imaging images and optical images, so that it can combine image data of multiple dimensions for predictive analysis, improve the accuracy of data acquisition, and provide more accurate prediction data.

[0023] See also Figure 2, is a flow chart of a neurology data analysis method provided by an embodiment of the present invention, Figure 2 The execution subject of the method shown may be a software and / or hardware device. The execution subject of the present application may include but is not limited to at least one of the following: user equipment, network equipment, etc. Among them, the user equipment may include but is not limited to computers, smart phones, personal digital assistants (PDAs) and the electronic devices mentioned above. Network equipment may include but is not limited to a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of computers or network servers, wherein cloud computing is a type of distributed computing, a super virtual computer composed of a group of loosely coupled computers. This embodiment does not limit this. It includes steps S101 to S104, as follows: S101, obtaining patient data, replacing the real structure in the patient data with a virtual structure, adding data tags and storing the data in a database.

[0024] It is understandable that in order to comprehensively collect information on cerebral palsy patients while protecting their privacy, the patient data can be privacy-processed before being stored.

[0025] Patient data includes facial images, CT images, symptoms such as hand numbness and facial paralysis, and detailed information about medical history. Real structures refer to actual body structures, such as facial organs, while virtual structures refer to simulated body structures. For example, virtual models of facial organs, such as eyes and nose, replace the corresponding organs in a real facial image. Data labels are used to identify various types of data, such as patient number, data collection time, symptom type, and disease diagnosis. This facilitates data classification, retrieval, and management.

[0026] By building a database, a rich, comprehensive and standardized data foundation can be provided, and patient privacy can be effectively protected, preventing the patient's real information, such as real facial information, from being leaked during data flow, storage and analysis, thereby meeting the requirements of medical data security and compliance.

[0027] In some embodiments, step S101 may be implemented in the following manner: A facial image of the patient is obtained based on the patient data, and the key point coordinates of each real structure in the facial image are extracted; based on the key point coordinates, key parameters of each real structure are calculated, and virtual structures corresponding to the key parameters are generated, where the key parameters include at least shape parameters, position parameters, and angle parameters; after replacing the real structure in the facial image with the virtual structure, data tags are added and stored in a database.

[0028] It is understandable that the purpose of obtaining the patient's facial image and extracting the coordinates of key points is to accurately capture the position and morphological information of various real facial structures such as eyes, nose, mouth, etc., to provide basic data for generating virtual structures, thereby realizing the quantification and analysis of the patient's facial features.

[0029] Specifically, facial image data can be obtained from patient data, which may come from a hospital's image storage system, electronic medical record system, or other sources. Deep learning algorithms, such as facial landmark detection algorithms based on convolutional neural networks (CNNs), can be used to extract keypoint coordinates. Before inputting the facial image into the model, preprocessing can be performed, including resizing the image to meet the model's input size requirements. The preprocessed image is then fed into the model, which extracts and analyzes image features layer by layer to output the coordinates of key points of various facial structures, such as the eyes, nose, and mouth. For example, for the eyes, the coordinates of key points such as the corners of the eyes and the edge of the eyelids can be precisely located; for the mouth, the coordinates of key points such as the corners of the mouth and the lip contour can be determined. Keypoint coordinates are the coordinates of representative points in the facial image, selected to accurately describe the position and morphology of various real-world structures.

[0030] Based on the detected keypoint coordinates, key parameters of each real-world structure are calculated. For example, for the eyes, the aspect ratio can be calculated (determined by measuring the horizontal and vertical lengths of the eyes); the center position of the eyeball can be calculated by averaging the coordinates of the eye's keypoints; and the tilt angle of the eye can be determined by the angle between the line connecting the inner and outer corners of the eye and the horizontal. For the mouth, parameters such as the position of the mouth corners (coordinates of the mouth corner keypoints) and the thickness of the lips can be calculated (measured by measuring the distance between the keypoints of the upper and lower lip edges); and the degree of mouth opening (measured by the vertical distance between the keypoints of the upper and lower lips). These parameters can comprehensively describe the shape, position, and posture of each real-world structure. Key parameters are parameters used to describe the characteristics of each real-world facial structure, primarily including shape parameters, position parameters, and angle parameters. Position parameters can be used to determine the position of a real-world structure relative to the face or other structures, such as the coordinates of an organ's center point or its distance from the facial midline. Angle parameters indicate the orientation or tilt of a real-world structure, such as the tilt angle of the eyes or the angle of mouth opening. Shape parameters can be used to describe the shape of a real-world structure.

[0031] When generating a virtual structure based on the extracted parameters, computer graphics techniques, such as modeling software, can be used to generate the corresponding virtual model. Taking the eye as an example, based on the calculated parameters such as the eye's aspect ratio, eyeball center position, and tilt angle, an eye model with the corresponding shape, size, and orientation can be constructed in the modeling software.

[0032] When replacing real structures in a facial image with virtual structures, image segmentation algorithms, such as semantic segmentation algorithms based on fully convolutional neural networks like U-Net, can be used to segment real structures in the facial image, such as the eyes, nose, and mouth, from the entire image. Then, based on the previously calculated key point coordinates and key parameters, the generated virtual structures are transformed and matched to precisely match the original real structures in position, angle, and size. For example, the virtual eyes can be translated, rotated, and scaled to align their position and angle with the real eyes. Finally, image fusion techniques, such as Poisson fusion, are used to fuse the transformed virtual structures into the original facial image, replacing the segmented real structures to generate the replaced facial image. This ensures a natural transition between the fused images and no noticeable splicing artifacts.

[0033] This approach helps protect patient privacy and prevents the leakage of patients' real facial information during data storage, transmission, and analysis. Adding data tags also helps classify, manage, and retrieve data, facilitating rapid data analysis and application.

[0034] S102 , collecting a sensing image and a visible image corresponding to a target part of the current patient, and determining a region of interest in the visible image according to temperature distribution information in the sensing image.

[0035] In practice, congenital facial deviations can be easily confused with facial deviations and related symptoms caused by acute cerebral infarction due to bony interference and compensatory soft tissue changes. However, thermal imaging technology can detect facial temperature distribution. Cerebral infarctions can cause temperature changes due to abnormal local blood circulation or neural regulation, and these temperature changes are not directly affected by congenital facial bony and soft tissue abnormalities. Therefore, by clearly displaying the facial structure through visual images and combining them with the temperature information from the sensor images, it is possible to accurately locate the abnormal areas caused by the disease, effectively eliminating the interference of congenital facial deviation and improving the accuracy of data prediction.

[0036] The sensed image refers to a thermal image, and temperature distribution information refers to the temperature differences between different areas of the face. It's worth noting that under normal circumstances, facial temperatures are relatively symmetrical on both sides, but illness can cause temperature imbalances. The visual image refers to an optical image that can be used to clearly display the structure, morphology, and features of the face. The area of interest (ROI) is an area in the visual image identified based on the temperature distribution information in the sensed image that may contain pathological abnormalities, such as areas with significant temperature differences from surrounding areas.

[0037] Specifically, the sensing image and the visible image can be compared, and multiple areas on both sides of the face can be divided based on the visible image. Then, based on the temperature distribution between the various areas in the sensing image, the areas with large temperature differences on both sides of the face can be found, thereby determining the areas of interest.

[0038] Based on the above embodiment, the specific implementation of step S103 may be: The visible image is divided into a plurality of symmetrical structural areas, and the visible image includes an initial image and a dynamic image of the current patient under various action instructions.

[0039] The initial image is a visual image of the patient's face captured in a naturally relaxed state without any specific facial expressions or movements, reflecting the basic facial form and natural state. The dynamic image is a visual image of the patient's face captured while they respond to specific movements, such as smiling, frowning, or closing their eyes, and then make the corresponding facial expressions. Different dynamic images demonstrate the changes in facial muscles under different motion states, helping to observe changes in temperature distribution and structural characteristics of the face under various facial expressions and movements, thereby identifying potential abnormalities.

[0040] Symmetrical structural regions divide the visible face into multiple corresponding left and right regions based on the left-right symmetry of the face, such as the left and right cheeks, the left and right eye areas, and the upper and lower lips. These symmetrical regions are similar in position and structure, facilitating comparative analysis of temperature, morphology, and other characteristics of corresponding facial regions on both sides.

[0041] It is understandable that dividing the visible image into multiple symmetrical structural regions facilitates comparative analysis of temperature differences in corresponding areas on both sides of the face. Due to the bilateral symmetry of the face, this division allows for more intuitive identification of areas with potential temperature anomalies. Furthermore, considering both the initial image and the dynamic images generated by each action command allows for a comprehensive observation of the face in different states, as different facial expressions and movements may make temperature changes in areas of potential neurological dysfunction or lesions more pronounced, thereby improving the accuracy and comprehensiveness of detection.

[0042] In some embodiments, the visible image may be divided into multiple symmetrical structural regions by the following steps: The facial contour of the visual image is extracted to generate a vertical centerline of the facial contour; multiple structural extreme points in the facial contour are identified, and offset points at preset distances from the structural extreme points are determined based on preset directions corresponding to the respective structural extreme points; dividing lines passing through the offset points and perpendicular to the vertical centerline are generated, and multiple bilaterally symmetrical structural regions are obtained based on the dividing lines corresponding to the structural extreme points having the same structural attributes.

[0043] It's understandable that facial contour extraction is intended to determine the facial boundaries and provide a basis for regional segmentation. Generating the vertical midline of the facial contour leverages the left-right symmetry of the face, dividing it into two relatively symmetrical parts. This facilitates comparative analysis of the left and right sides, allowing for more accurate detection of potential anomalies, such as temperature differences. The vertical midline is a straight line generated based on the facial contour, perpendicular to the horizontal direction of the image. It can be determined using key landmarks such as the tip of the nose and the center of the eyebrows.

[0044] Identifying structural extreme points is crucial for vertically defining facial features. These points reflect the boundaries of their corresponding organs and are crucial for region delineation. Determining offset points based on preset directions and distances allows for the identification of reference points for dividing facial regions based on the structural extreme points. This allows for the generation of appropriate demarcation lines, dividing the face into multiple symmetrical structural regions. For example, offsetting the structural extreme points of the eyes allows for a more comprehensive analysis of the features of the eyes and their surroundings.

[0045] Among them, the structural extreme points refer to the maximum and minimum points of the corresponding organs in the vertical direction, such as the upper and lower boundary points of the eyes, which are the highest and lowest points of the eyes in the vertical direction. The preset direction corresponding to the maximum point is from bottom to top, and the preset direction corresponding to the minimum point is from top to bottom. The preset distance refers to the preset fixed distance moving from the structural extreme point along the preset direction, which can be expressed in pixels, for example, setting the distance to 5 pixels, 10 pixels, etc., to determine the specific position of the offset point. The offset point refers to the reference point obtained by extending the structural extreme point in a specific direction and distance on the basis of the structural extreme point, which is used to generate the division of the facial area.

[0046] A demarcation line is a line that passes through the offset point and is perpendicular to the vertical midline of the facial contour. This demarcation line allows for horizontal segmentation of the facial image. Structural attributes refer to characteristic properties of facial structure. For example, the upper and lower boundary points of the eyes correspond to eye structural attributes. Demarcation lines corresponding to structural extreme points with the same structural attributes are used to demarcate regions with similar structural features, such as the eye and lip regions.

[0047] This method allows facial images to be meticulously divided into multiple regions with clear structural attributes and symmetry, such as the left and right cheeks, left and right eye areas, and left and right lip areas. These symmetrical regions facilitate bilateral comparative analysis, allowing for more intuitive and clear observation of changes in facial features across different regions.

[0048] The temperature difference between the symmetrical structural regions is determined according to the temperature distribution information in the sensing image corresponding to the visible image.

[0049] The temperature difference is calculated by comparing the temperature values of corresponding areas on the left and right sides of a symmetrical structure. For example, the difference between the average temperature of the left cheek and the average temperature of the right cheek is used. The magnitude of the temperature difference reflects the degree of temperature difference between corresponding areas on both sides of the face and is an important indicator for determining whether there is a temperature abnormality on the face.

[0050] The temperature differences between the symmetrical structural regions include the temperature differences between all the symmetrical structural regions in the initial image and the temperature differences between the structural regions corresponding to the action instructions in the dynamic image.

[0051] It's understandable that the initial image reflects the natural state of the face. Comparing the temperature differences across all symmetrical structural regions can reveal the overall temperature distribution of the face. However, in dynamic images, not all regions experience significant temperature changes due to facial expressions. Therefore, by comparing only those regions that exhibit significant temperature differences under a particular expression, unnecessary temperature difference calculations and analysis of numerous unrelated regions can be avoided, improving analysis efficiency. Furthermore, different facial expressions activate different facial muscle groups, which are innervated by different nerves. When nerve function is abnormal, changes in blood circulation and metabolism in the corresponding muscles are primarily manifested in the muscle areas associated with that expression. For example, smiling primarily involves the cheek and lip muscles, and patients with facial paralysis experience more pronounced temperature changes in these areas when smiling. Therefore, by specifically comparing the temperature differences in these regions, we can more accurately locate areas of temperature abnormality associated with neural dysfunction, improving detection accuracy.

[0052] In practical applications, each action command can be configured with a corresponding facial area in advance. For example, for a smile command, the left and right cheek areas and the left and right lip areas can be matched with it. When detecting the dynamic image corresponding to the smile command, only the temperature difference between the left and right cheek areas and the left and right lip areas can be calculated, while the temperature difference between the areas where the temperature change is not obvious under the smile action is not calculated.

[0053] The structural area with a temperature difference greater than the temperature threshold is determined as a suspicious area. The abnormal value of the suspicious area in different visible images is calculated based on the abnormal level of the temperature difference. The suspicious area with an abnormal value greater than the preset abnormal value is determined as the focus area.

[0054] The temperature threshold refers to a pre-set temperature difference standard used to screen out structural areas with significant temperature differences. When the temperature difference between symmetrical structural areas exceeds the temperature threshold, the corresponding structural area is considered to have a potential anomaly, or a suspicious area. For example, if the temperature threshold is set to 0.5°C, if the temperature difference in a symmetrical area reaches 0.6°C, the area meets the criteria for being a suspicious area.

[0055] The temperature difference abnormality level is used to more carefully assess the degree of abnormality reflected by the temperature difference. Temperature differences are divided into different level intervals, with each interval corresponding to an abnormality level. Different abnormality levels reflect the magnitude of the temperature difference and the possible severity of the pathology, and are used to calculate and evaluate abnormal values in suspicious areas.

[0056] The anomaly value is a pre-set anomaly value for each suspicious area, based on the abnormality level of the temperature difference. The pre-set anomaly values for the same suspicious area are then superimposed and calculated across different visual images to form the anomaly value for that suspicious area. The anomaly value comprehensively considers the temperature anomaly of the suspicious area under different expression states and can be used to identify areas of particular concern.

[0057] The preset abnormality value is a pre-set numerical standard used to determine whether the abnormality value of a suspicious area is large enough to determine whether the area is a region of concern. When the abnormality value of a suspicious area is greater than the preset abnormality value, the suspicious area is determined to be a region of concern, indicating that the area is more likely to contain a true pathological abnormality.

[0058] It's understandable that setting a temperature threshold is intended to screen out structural areas with significant temperature differences, identifying these areas as suspicious, narrowing the scope of analysis and focusing on areas of possible pathological changes. The system calculates abnormality values for suspicious areas in different visual images based on the abnormality level of temperature differences and sets preset abnormality values to screen out areas of interest. This is done to comprehensively consider temperature anomalies in suspicious areas under different facial expressions, improving the accuracy and reliability of judgments.

[0059] In some embodiments, the following steps may be performed to calculate abnormal values of suspicious areas in different visible images based on the abnormal levels of temperature differences: The temperature difference intervals corresponding to each preset level of temperature difference are traversed to determine the preset level of temperature difference as its corresponding abnormal level. Each abnormal level is configured with a corresponding preset abnormal value; the preset abnormal values of the suspicious areas in different visible images are superimposed and calculated to obtain their corresponding abnormal values.

[0060] The preset levels are pre-set based on the degree of difference between facial temperature and normal standards, used to measure the degree of abnormality represented by the temperature difference. For example, four preset levels can be set: no abnormality (Level 0), slight abnormality (Level 1), moderate abnormality (Level 2), and severe abnormality (Level 3).

[0061] The temperature difference range is the temperature difference range corresponding to each preset level. For example, the temperature difference range for no abnormality (Level 0) is within 0.5°C; the temperature difference range for slight abnormality (Level 1) is 0.5-1°C; the temperature difference range for moderate abnormality (Level 2) is 1-2°C; and the temperature difference range for severe abnormality (Level 3) is more than 2°C.

[0062] A preset anomaly value is a pre-configured value for each anomaly level, used to quantify the severity of that anomaly level. Different anomaly levels correspond to different preset anomaly values. For example, no anomaly (level 0) might correspond to a preset anomaly value of 0, a minor anomaly (level 1) to a preset anomaly value of 1, a moderate anomaly (level 2) to a preset anomaly value of 2, and a severe anomaly (level 3) to a preset anomaly value of 3.

[0063] Specifically, for the temperature differences calculated between symmetrical structural regions in the visual image, the temperature difference intervals corresponding to each preset level are traversed. For example, if the temperature difference in a symmetrical structural region is 1.3°C, and the traversal reveals that 1.3°C falls within the temperature difference interval of 1-2°C, then the preset level for this temperature difference is moderate anomaly (level 2), and its corresponding anomaly level is determined to be moderate. In this way, the corresponding anomaly level is determined for the temperature differences in symmetrical structural regions in all visual images. Based on the determined anomaly level, the corresponding preset anomaly value is found. For each suspicious region, the preset anomaly values corresponding to the suspicious region in different visual images are superimposed and calculated. For example, the temperature difference in a suspicious region in the initial image corresponds to a mild anomaly level with a preset anomaly value of 1. In the dynamic image of a smiling action, the temperature difference in the suspicious region corresponds to a moderate anomaly level with a preset anomaly value of 2. In the dynamic image of a frowning action, the temperature difference in the suspicious region also corresponds to a mild anomaly level with a preset anomaly value of 1. Therefore, the anomaly value for the suspicious region is 4. If the abnormality value of the suspicious area is greater than the preset abnormality value, it is determined to be a focus area.

[0064] S103: Identify spatial metric parameters corresponding to the structural points in the region of interest.

[0065] Structural points are landmark points within the region of interest. For example, they can be key facial features such as the corners of the eyes, mouth, nose, and eyebrows. Structural points can reflect important positional and morphological information about facial structure. Spatial metric parameters are calculated by calculating the relative positional relationships between structural points. These parameters include angle parameters. It can be understood that these parameters describe the characteristics of the region of interest from a spatial perspective.

[0066] Understandably, the purpose of quantifying the structural features of the region of interest is to convert them into numerical parameters that can be used for data analysis and comparison. This provides a precise basis for comparison with historical data in the database, allowing for accurate judgment of the similarities and differences between the current patient's condition and previous cases. Precise structural feature quantification allows for a more objective and accurate assessment of changes in a patient's facial structure, providing a more precise basis for prediction.

[0067] Based on the above embodiment, the specific implementation of step S103 may be: The structural points in the focus area are extracted, and the distance between each structural point and the vertical midline in the facial contour is obtained; the structural points on the corresponding side are arranged from small to large according to the distance to obtain a structural point sequence; the connection order of each structural point is determined according to the comparison result of the preset sequence and the structural point sequence on the corresponding side of the focus area, and each preset structural point in the preset sequence is configured with a corresponding connection number; the structural points with the same connection number are connected to obtain multiple determination lines, and the spatial measurement parameters are obtained according to the angles of the determination lines.

[0068] The structure point sequence refers to an ordered sequence of points formed by arranging the structure points on the corresponding side, such as the left or right side, from smallest to largest, based on their distance from the vertical midline of the facial contour. A preset sequence refers to the pre-defined order of the preset structure points on the corresponding side of the region of interest, where each preset structure point is assigned a corresponding connection number. A decision line refers to the straight line formed by connecting the corresponding structure points according to the specified connection number.

[0069] Specifically, structure points are extracted from the determined region of interest. For example, if the region of interest is the eye, structure points such as the canthus and eyelid margin are extracted. Then, using image processing and coordinate positioning techniques, the distance between each structure point and the vertical midline of the facial contour is obtained. For example, the distance between the structure point and the vertical midline can be obtained by calculating the absolute value of the difference between the horizontal coordinate and the horizontal coordinate of the vertical midline. Based on the obtained distances between each structure point and the vertical midline, the structure points on the corresponding side, such as the left or right side, are arranged in ascending order to obtain a structure point sequence. Points with the same arrangement order in the preset sequence and the structure point sequence are then mapped to each other, and the connection numbers assigned to the corresponding preset structure points in the preset sequence are used as the connection numbers of the corresponding structure points. The preset structure points in the preset sequence can also be arranged in ascending order of their distance from the facial midline. The structure points with the same connection numbers are then connected to obtain a determination line. The angle of the determination line can be used to determine the degree of skewness of the corresponding organ.

[0070] In actual applications, when setting connection numbers, the connection numbers of preset structural points that can reflect organ characteristic information, such as skew, can be set to the same, so that the degree of skewness of the corresponding organ can be determined by combining the angle corresponding to the straight line formed by the corresponding structural points. For example, for the eye area, the point numbers corresponding to the inner and outer corners of the eye can be set to the same. Under normal circumstances, the line connecting the inner and outer corners of the eyes should be basically parallel to the horizontal midline of the face. If the angle between the inner and outer corners of an eye and the horizontal midline is too large or too small, it may indicate that the eye area is skewed in the up-down or left-right direction.

[0071] S104: Determine the predicted pathological site of the current patient based on the comparison result of the spatial measurement parameter and the historically stored parameters in the database, and generate a predicted display image.

[0072] Leveraging a vast database of historical patient data, such as facial structural features corresponding to different locations of cerebral infarction blood clots in different patients, and utilizing data comparison and analysis methods, we can predict the likely location of the current patient's pathology, providing a reference for clinical diagnosis. Historical data contains various characteristic parameters of patients with different conditions. By comparing these parameters, we can identify patterns similar to those of the current patient, and then combine this historical data to predict the current patient's pathological location. Generating a prediction display image visually demonstrates the prediction results, providing clearer reference data.

[0073] Among them, historical storage parameters refer to the spatial measurement parameters of past patients stored in the database, as well as the corresponding pathological sites, disease diagnosis results, treatment plans, and other information. The predicted pathological site refers to the predicted pathological location of the current patient's hematoma. For example, if the spatial measurement parameters of a historical case in the database are very similar to those of the current patient, and the pathological site of the historical case is a specific area of the brain, then it can be predicted that the current patient's hematoma may also be located in that area. The predicted display image refers to an image that presents the predicted pathological site in a visual manner. The predicted pathological site can be highlighted by marking, framing, etc.

[0074] Based on the above embodiment, the specific implementation of step S104 may be: The structural attributes of the region of interest are determined, and a reference region corresponding to the structural attributes is extracted from the facial image included in the data of each patient.

[0075] The facial images contained in the data of each patient can also be divided into multiple regions in the aforementioned manner, and corresponding structural attributes can be added to each region, so that the reference region corresponding to the current patient's focus area in the facial image contained in the data of each patient can be found, so that the current patient's condition can be predicted in combination with historical data.

[0076] A historical angle of the reference area is obtained, where the historical storage parameters include the historical angle.

[0077] The historical angle refers to the angle data of the facial structure area with the same structural attributes and matching position as the focus area stored in the historical patient data, which can be the angle information obtained by connecting the corresponding structural points in the reference area.

[0078] The angle corresponding to the spatial metric parameter is compared with the historical angle, and the historical angle whose angle difference between the two is less than an angle difference threshold is determined as the target angle.

[0079] The angle difference threshold is a pre-set value used to determine whether the angle difference between the current angle and historical angles is within an acceptable range, thereby filtering out historical angles similar to the current situation. For example, it can be set to 5 degrees. The calculated angle difference is compared with the threshold. If the angle difference is less than the set threshold, the historical angle is determined to meet the screening requirements and is selected as the target angle.

[0080] The historical pathological site corresponding to the target angle is determined to be the predicted pathological site of the current patient, a pathological display image is retrieved, and the predicted pathological site in the pathological display image is framed to obtain the predicted display image.

[0081] After determining the target angle, the pathological site corresponding to that target angle can be searched in the database and determined as the predicted pathological site for the current patient. For example, the historical angle of the left eye of patient D is the target angle. Patient D's medical records clearly document structural abnormalities in the left eye due to nerve compression caused by a blood clot in a specific area of the brain due to cerebral infarction. This specific brain area is the pathological site for patient D. Based on this, and considering the similarities in the eye structural angles between the current patient and patient D, it is speculated that the current patient's hematoma may also be located in the same or a similar area of the brain, thus determining this area as the predicted pathological site for the current patient.

[0082] The pathology display image can be a pre-set image designed to display the pathology site, such as a pre-set brain CT image. Within the pathology display image, the predicted pathology site is precisely framed by drawing a rectangular or irregular polygonal box, or using a specific marking tool, so that it is highlighted within the image, thereby generating the predicted display image. For example, if the predicted pathology site is a certain area of the brain, an irregular polygonal box can be drawn on the brain CT image using image processing software to closely fit the area, clearly marking it and forming the predicted display image.

[0083] In some embodiments, the predicted display image may be obtained by the following steps: When the number of areas of the area of interest is greater than a reference constant, the associated area groups corresponding to each predicted pathological site are obtained; the associated labels corresponding to the structural attributes of the associated area groups are determined, and the severity level of the predicted pathological site is determined according to the preset level of the associated labels; the corresponding predicted pathological site in the pathological display image is updated based on the reference site area corresponding to the severity level, and the predicted display image is obtained after framing it.

[0084] It's understandable that when the number of regions of interest exceeds the baseline constant, it indicates a significant number of abnormal facial areas. Obtaining the associated region groups corresponding to each predicted pathological site allows for in-depth exploration of the connection between multiple facial abnormalities and the same pathological site, providing a basis for accurately assessing the complexity of the condition. Because multiple facial abnormalities can be caused by the same pathological site, identifying the associated region groups clearly identifies the specific abnormal facial areas involved.

[0085] The baseline constant is 1, and the associated region group is a set of facial regions associated with the predicted pathological site. Because abnormalities in multiple facial regions may be caused by the same pathological site, facial regions corresponding to the same predicted pathological site can be grouped together. For example, if a vascular region in the brain is diseased, the associated facial regions may include areas around the ipsilateral eye and near the corner of the mouth where muscle weakness and sensory abnormalities are present.

[0086] Associated labels are pre-assigned identifiers for different facial region combinations corresponding to pathological sites. Preset levels are pre-assigned to associated labels and are used to measure the severity of the lesion associated with the facial region combination represented by the label, ranging from mild, moderate, to severe. Different facial region combinations present different abnormalities, often reflecting the varying severity of the lesions. Therefore, assigning different labels allows for accurate classification and assessment of these combinations. For example, the severity of an abnormality caused by a pathological site in both the eyes and mouth may differ from that caused by a pathological site in both the eyes. When the lesion affects only the neurovascular branches supplying the eyes, symptoms such as ptosis may occur. When the lesion further extends to affect the neurovascular branches supplying the mouth, symptoms such as droopy of the mouth corners may develop. However, relatively isolated abnormalities in the eye region initially have a relatively limited impact on overall facial function, so the severity of the latter combination may be less severe than the former.

[0087] The reference area refers to the standard area value corresponding to different severity levels. The higher the severity level, the larger the corresponding reference area is usually. It is used to measure the size range of the predicted pathological site and reflect the degree of abnormality from the perspective of area. For example, a severe severity level may correspond to a larger reference area, while a mild severity level corresponds to a smaller area. For example, for the predicted pathological site of the brain, a severe severity may correspond to a larger area of brain tissue lesions, while a mild severity corresponds to a smaller area. This reflects the difference in the range of lesions at different severity levels for the same pathological site. By adjusting the display size of the predicted pathological site through the reference area, the severity and range of lesions of the same pathological site under different facial area combinations can be intuitively reflected.

[0088] In some embodiments, the reference part area corresponding to the severity level can be determined by the following steps, including: obtaining multiple medical imaging data corresponding to the associated labels, extracting the pathological area of the part corresponding to the corresponding predicted pathological part in each medical imaging data; and calculating the average value of each pathological area to obtain the reference part area.

[0089] Specifically, multiple medical imaging data sets corresponding to associated labels can be found in the database, such as CT images. Each medical imaging data set can be pre-assigned to the label corresponding to its corresponding facial region combination, as well as information such as the abnormal angle of the corresponding facial region. Through this medical imaging data set, multiple sample data sets similar to the current patient's condition can be obtained. Combined with these sample data sets, the area of the pathological site corresponding to the current patient can be inferred, improving the accuracy of the prediction. Calculating the average pathological area of each medical image to obtain the reference site area can integrate multiple imaging information to make the resulting area data more representative and reliable.

[0090] See also Figure 3 , is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention, wherein the electronic device 30 includes: a processor 31, a memory 32 and a computer program; The memory 32 is used to store the computer program, which may also be a flash memory. The computer program is, for example, an application program or a functional module for implementing the above method.

[0091] The processor 31 is configured to execute the computer program stored in the memory to implement the various steps performed by the device in the above method. For details, please refer to the relevant description in the above method embodiment.

[0092] Optionally, the memory 32 may be independent or integrated with the processor 31 .

[0093] When the memory 32 is a device independent of the processor 31, the device may further include: The bus 33 is used to connect the memory 32 and the processor 31 .

[0094] The present invention also provides a readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the methods provided in the various embodiments described above.

[0095] The readable storage medium may be a computer storage medium or a communication medium. Communication media include any medium that facilitates the transfer of computer programs from one location to another. Computer storage media may be any available medium that can be accessed by a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium may also be an integral part of the processor. The processor and the readable storage medium may be located in an application-specific integrated circuit (ASIC). In addition, the ASIC may be located in a user device. Of course, the processor and the readable storage medium may also exist as discrete components in a communication device. The readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.

[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A neurology data analysis method, characterized in that: include: Acquire patient data, replace the real structure in the patient data with a virtual structure, add data tags and store in a database; Acquire a sensing image and a visible image corresponding to a target part of the current patient, and determine a region of interest in the visible image based on temperature distribution information in the sensing image; Identify spatial metric parameters corresponding to structural points in the region of interest; According to the comparison result of the spatial measurement parameters and the historical storage parameters in the database, the predicted pathological site of the current patient is determined, and a predicted display image is generated.

2. The method according to claim 1, characterized in that Obtaining patient data, replacing the real structure in the patient data with a virtual structure, adding data tags and storing them in a database, including: Obtaining a facial image of the patient according to the patient data, and extracting the coordinates of key points of each real structure in the facial image; Based on the key point coordinates, calculating key parameters of each real structure and generating a virtual structure corresponding to the key parameters, where the key parameters include at least shape parameters, position parameters, and angle parameters; After the real structure in the facial image is replaced with the virtual structure, a data label is added and stored in a database.

3. The method according to claim 1, characterized in that Acquiring a sensing image and a visible image corresponding to a target part of the current patient, and determining a region of interest in the visible image based on temperature distribution information in the sensing image, including: Dividing the visible image into a plurality of symmetrical structural regions, the visible image including an initial image and a dynamic image of the current patient under various action instructions; determining a temperature difference between symmetrical structural regions based on temperature distribution information in a sensing image corresponding to the visible image; The structural area with a temperature difference greater than the temperature threshold is determined as a suspicious area. The abnormal value of the suspicious area in different visible images is calculated based on the abnormal level of the temperature difference. The suspicious area with an abnormal value greater than the preset abnormal value is determined as the focus area.

4. The method according to claim 3, characterized in that The visible image is divided into a plurality of symmetrical structural regions, including: Extracting the facial contour of the visual image and generating a vertical centerline of the facial contour; Identifying a plurality of structural extreme points in the facial contour, and determining an offset point at a preset distance from the structural extreme point according to a preset direction corresponding to each structural extreme point; A dividing line passing through the offset point and perpendicular to the vertical center line is generated, and multiple bilaterally symmetrical structural regions are obtained according to the dividing lines corresponding to the structural extreme points with the same structural attributes.

5. The method according to claim 3, characterized in that Temperature differences between symmetrical structural areas, including: The temperature difference between all symmetrical structural regions in the initial image and the temperature difference between structural regions corresponding to the action instructions in the dynamic image.

6. The method according to claim 3, characterized in that The abnormal values of suspicious areas in different visual images are calculated based on the abnormal level of temperature difference, including: Traverse the temperature difference intervals corresponding to each preset level of the temperature difference, determine the preset level where the temperature difference is located and its corresponding abnormal level, and configure each abnormal level with a corresponding preset abnormal value; The preset abnormal values of the suspicious areas in different visual images are superimposed and calculated to obtain their corresponding abnormal values.

7. The method according to claim 1, characterized in that Identifying spatial metric parameters corresponding to structural points in the region of interest, including: Extracting structural points in the region of interest, and obtaining the distance between each structural point and the vertical midline of the facial contour; Arrange the structure points on the corresponding side from small to large according to the spacing to obtain a structure point sequence, and determine the connection order of each structure point according to the comparison result of the preset sequence on the corresponding side of the focus area and the structure point sequence, wherein each preset structure point in the preset sequence is configured with a corresponding connection number; Multiple determination lines are obtained by connecting the structural points with the same connection number, and the spatial measurement parameters are obtained according to the angles of the determination lines.

8. The method according to claim 1, characterized in that Determine the predicted pathological site of the current patient based on the comparison results of the spatial measurement parameters and the historically stored parameters in the database, and generate a predicted display image, including: Determining structural attributes of the region of interest, and extracting reference regions corresponding to the structural attributes in the facial images included in the data of each patient; Acquiring a historical angle of the reference area, wherein the historical storage parameters include the historical angle; Comparing the angle corresponding to the spatial measurement parameter with the historical angle, and determining the historical angle whose angle difference between the two is less than an angle difference threshold as the target angle; The historical pathological site corresponding to the target angle is determined to be the predicted pathological site of the current patient, a pathological display image is retrieved, and the predicted pathological site in the pathological display image is framed to obtain the predicted display image.

9. The method according to claim 8, characterized in that Determine the historical pathological site corresponding to the target angle as the predicted pathological site of the current patient, retrieve the pathological display image, and frame the predicted pathological site in the pathological display image to obtain the predicted display image, including: When the number of regions of interest is greater than a reference constant, obtaining a correlation region group corresponding to each predicted pathological site; Determining association labels corresponding to the structural attributes of the association region group, and determining a severity level of the predicted pathological site according to a preset level of the association labels; The predicted pathological part corresponding to the pathological display image is updated based on the area of the reference part corresponding to the severity level, and is framed to obtain the predicted display image.

10. The method according to claim 9, characterized in that The following steps are used to determine the baseline area corresponding to the severity level, including: Acquire multiple medical image data corresponding to the associated labels, and extract the pathological area of the part corresponding to the corresponding predicted pathological part in each medical image data; The average of each pathological area was calculated to obtain the baseline area.

Images