Intelligent Screening Method and System for Peripheral Circulation Abnormalities Based on Temporal Variation of Skin Temperature Field
By constructing a skin temperature field time-series data chain and feature group, the problem of misjudgment of the compression relief and warming process in peripheral circulation screening by infrared thermal imaging technology was solved, and more stable and accurate screening results were achieved.
Patent Information
- Application Number
- CN202611086802.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-21
- Publication Date
- 2026-11-17
AI Technical Summary
Existing infrared thermal imaging technology has difficulty distinguishing between the short-term warming process after compression relief and the temporal changes in skin temperature related to peripheral circulatory abnormalities in handheld continuous acquisition and peripheral circulation screening scenarios, resulting in insufficient stability and accuracy of screening results.
By constructing a multi-source skin temperature field time-series data chain, we can perform limb partitioning, neighborhood temperature benchmark analysis, compression residue identification, inter-frame correlation, and temperature change point detection to generate screening feature groups, remove transient temperature interference, and improve the accuracy of peripheral circulation screening.
It improves the stability and reliability of peripheral circulation screening results, reduces the risk of false alarms caused by short-term recovery and warming, outputs retest prompts and abnormal area heat maps, and enhances the traceability of screening results.
Smart Images

Figure CN123101092A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of skin temperature detection technology, and in particular to an intelligent screening method and system for peripheral circulation abnormalities based on temporal changes in skin temperature field. Background Technology
[0002] With the development of health monitoring equipment, infrared thermal imaging equipment, and portable smart terminals, non-contact body surface temperature acquisition has gradually become a common technical means in human condition observation, basic health screening, and home monitoring. Infrared thermal imaging can present the temperature distribution of the human body surface in the form of images, and has the characteristics of fast acquisition speed, no direct contact required, and high visualization. It has been applied to scenarios such as body surface temperature observation, local temperature difference analysis, exercise recovery monitoring, postoperative observation assistance, rehabilitation process recording, and primary care screening. Compared with traditional single-point temperature measurement methods, infrared thermograms can simultaneously reflect the temperature distribution of a large area of the skin surface, and can observe the trend of temperature change over time through continuous acquisition. Therefore, it has certain application value in observing changes in limb surface temperature, distribution of local heat zones, and the state of body surface circulation. Currently, related equipment usually combines infrared sensors, image display terminals, and data processing software to display, store, and perform basic analysis on the acquired thermograms, and output the temperature distribution results in the form of pseudo-color images, area annotations, or prompt information, providing operators with intuitive basis for body surface temperature observation.
[0003] For example, the invention patent with announcement number CN109394180B discloses a remote diagnosis and treatment system for knee osteoarthritis based on infrared imaging. This system includes a detection chamber equipped with a temperature control device, a timer, an angle calibration component, a distance calibration component, markers, and an infrared thermal imager; a computer; a thermal image processing module; a network; and a diagnosis and treatment terminal. The thermal image processing module is used to extract the position of the markers and, using the marker positions as a reference, calculates and extracts the temperatures of the medial tibial plateau region, the lateral tibial plateau region, the medial femoral condyle region, the lateral femoral condyle region, and the suprapatellar region. The network transmits the temperature results extracted by the thermal image processing module to the diagnosis and treatment terminal so that physicians can make further diagnoses. This system achieves the acquisition of skin temperature values in different areas of the knee joint through infrared imaging and remote transmission of the detection results to the diagnosis and treatment terminal, providing temperature data support for the remote diagnosis and treatment of knee osteoarthritis.
[0004] For example, the invention patent with announcement number CN103385734B discloses a comprehensive diagnostic instrument for dual examination using infrared thermal imaging-guided ultrasound and the detection method of the instrument. The diagnostic instrument includes a worktable, on which is equipped an infrared thermal imaging acquisition device for whole-body examination and an ultrasound detection device for precise examination of local areas, as well as a computer and an image display connected to both the infrared thermal imaging acquisition device and the ultrasound detection device. The computer has software that can simultaneously or separately control the infrared thermal imaging acquisition device and the ultrasound detection device. The software has the function of analyzing and processing infrared and ultrasound data image information and a database, realizing the combined use of infrared thermal imaging technology and ultrasound detection technology to obtain functional and structural imaging examination results, and providing comprehensive examination basis for disease diagnosis or early warning through image data analysis and processing.
[0005] Existing infrared thermography-based methods for analyzing body surface temperature primarily focus on extracting and displaying the temperature of fixed areas, local temperature differences, or thermal zone distribution from a single frame of thermal imaging, or combining infrared thermographic results with other examination results. These methods provide a certain data foundation for observing body surface temperature. However, in scenarios involving handheld continuous acquisition and peripheral circulation screening, skin temperature changes are not only affected by local circulatory status but are also easily interfered with by factors such as prolonged sitting, support contact, clothing pressure, changes in acquisition posture, and fluctuations in ambient temperature. Especially after pressure is relieved, the local skin often experiences a short-term warming process, gradually recovering from a low-temperature depression. This process may manifest as localized warming, temperature gradient changes, or enhanced linear temperature changes in the infrared thermographic timeline, easily confused with the temporal changes in skin temperature related to peripheral circulatory abnormalities. Existing methods are insufficient in distinguishing the temporal relationship between the residual low temperature zone after compression, the warming segment after compression is relieved, and the temperature conduction characteristics related to peripheral circulation. It is difficult to effectively remove transient warming interference before generating screening prompts, which makes the screening results susceptible to short-term recovery and warming, thus affecting the stability and accuracy of peripheral circulation abnormality screening prompts. Summary of the Invention
[0006] To address the technical problem that existing methods struggle to distinguish between the short-term warming process after pressure relief and the temporal changes in skin temperature related to peripheral circulatory abnormalities, and easily misidentify transient recovery warming as an abnormal trend, leading to insufficient stability of the abnormal screening temporal judgment results, this invention provides an intelligent screening method and system for peripheral circulatory abnormalities based on the temporal changes of the skin temperature field.
[0007] The technical solutions provided by the embodiments of the present invention are as follows: First aspect: This invention provides an intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field, comprising: S1: Collect multi-source skin temperature screening data and preprocess it to form a skin temperature field time series data chain; S2: Perform limb segmentation and neighborhood temperature benchmark analysis on the skin temperature field time-series data chain to identify candidate temperature depression areas; and generate compression residue markers and compression residue areas. S3: Perform inter-frame correlation and temperature change point detection on the compression residual area; and identify compression relief and temperature recovery segments to generate temperature recovery suppression markers and screen skin temperature field sequences; S4: Perform root-to-terminal temperature time-series analysis and temperature gradient reversal analysis on the screened skin temperature field sequence, and generate temperature gradient reversal markers; and construct a vascular course temperature change zone to form a peripheral circulation screening feature group; S5: Generate screening markers based on peripheral circulation screening feature groups, and output retest prompts, screening prompt reports, abnormal area heat maps, and abnormal basis labels.
[0008] The second aspect: This invention provides an intelligent screening system for peripheral circulation abnormalities based on temporal changes in skin temperature field, comprising: processor; A memory storing computer-readable instructions, which, when executed by the processor, implement the intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field as described in the first aspect.
[0009] Third aspect: The present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent screening method for peripheral circulation anomalies based on the temporal variation of skin temperature field as described in the first aspect.
[0010] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this embodiment of the invention, a multi-source skin temperature field time-series data chain is constructed, and limb partitioning, neighborhood temperature benchmark analysis, residual pressure identification, inter-frame correlation, temperature rebound point detection, and temperature gradient reversal analysis are performed sequentially to achieve hierarchical discrimination of the sources of abnormal skin temperature changes. By identifying the low-temperature zone of residual pressure and the short-term warming segment after pressure relief, transient temperature interference caused by prolonged sitting, support contact, clothing pressure, posture changes, and environmental fluctuations can be eliminated before peripheral circulation screening, avoiding misjudgment of local warming, gradient changes, or linear temperature changes as circulatory abnormalities. At the same time, by combining root-to-periphery temperature conduction characteristics, temperature gradient reversal characteristics, and vascular course temperature change zones, a screening feature group that can reflect the peripheral circulation status is formed, thereby improving the accuracy of abnormal area location and abnormal evidence interpretation. Finally, retest prompts, screening prompt reports, abnormal area heat maps, and abnormal evidence tags are output, which can improve the stability, reliability, and traceability of screening results in handheld continuous acquisition scenarios and reduce the risk of false alarms caused by short-term recovery temperature rise. Attached Figure Description
[0011] 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.
[0012] Figure 1 This is a flowchart illustrating an intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field, provided by an embodiment of the present invention.
[0013] Figure 2 A comparison chart of temperature changes related to the pressure relief and warming segment and peripheral circulation abnormalities provided in an embodiment of the present invention.
[0014] Figure 3 The flowchart for generating feature groups for extracting and screening linear temperature change regions provided in the embodiments of the present invention is shown.
[0015] Figure 4 A heatmap of the screening prompt area provided in an embodiment of the present invention.
[0016] Figure 5 This is a schematic diagram of the structure of an intelligent screening system for peripheral circulation abnormalities based on the temporal variation of skin temperature field, provided in an embodiment 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] Reference manual attached Figure 1 The diagram shows a flowchart of an intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field, provided by an embodiment of the present invention.
[0023] This invention provides an intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field. This method can be implemented by an intelligent screening device for peripheral circulation anomalies based on temporal changes in skin temperature field, which can be a terminal or a server. The processing flow of the intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field may include the following steps: S1: Collect multi-source skin temperature screening data and preprocess it to form a skin temperature field time series data chain.
[0024] Specifically, the process of collecting and preprocessing multi-source skin temperature screening data to form a skin temperature field time-series data chain is as follows: Collecting multi-source skin temperature screening data of the examined limb: Infrared thermal image sequences and frame timestamps are collected along the limb from root to tip using a handheld infrared thermometer. The infrared thermal image sequences are directly output by the area array temperature measurement unit of the infrared thermometer, and the frame timestamps are synchronously recorded by the acquisition clock of the handheld infrared thermometer. A visible light limb image sequence is synchronously acquired using a visible light imaging component. This sequence is obtained by synchronously imaging the surface of the same examined limb using the visible light imaging component. A ranging sequence is acquired using a ranging component. This sequence is obtained frame-by-frame by the ranging component recording the distance from the handheld infrared thermometer to the surface of the examined limb. A device attitude angle sequence is acquired using an attitude sensing component. This sequence is obtained by the attitude sensing component recording the pitch, yaw, and roll angles of the handheld infrared thermometer during the acquisition process. The ambient temperature sequence is collected by an ambient temperature sensing component, which continuously records the real-time temperature of the space being collected.
[0025] The infrared thermal image sequence and the visible light limb image sequence are frame-level correlated according to the thermal image frame timestamp, so that the pixel temperature value in the infrared thermal image sequence is associated with the limb contour position in the visible light limb image sequence at the same acquisition time. The infrared thermal image sequence is subjected to bad pixel replacement, temperature range verification, inter-frame smoothing and temperature compensation based on the ambient temperature sequence. Bad pixel replacement is completed based on the temperature value of neighboring pixels and the temperature value of pixels at the same position in adjacent infrared thermal image frames. Temperature range verification is used to remove pixel temperature values that are outside the effective temperature measurement range of the infrared temperature measurement component. Inter-frame smoothing is used to suppress random temperature measurement noise in a single frame. Temperature compensation is used to reduce the impact of temperature changes in the acquisition space on pixel temperature values. The visible light limb image sequence is subjected to background cropping and limb region preservation processing. Background cropping removes non-limb background areas, and limb region preservation processing preserves the boundary and surface texture information of the examined limb.
[0026] A semantic segmentation network based on dilated convolution is used to segment limb regions in a sequence of visible light limb images. The network takes the visible light limb image sequence as input and a limb contour mask as output. Training samples are visible light limb images with labeled limb regions. The training process constrains the predicted limb contour mask to maintain consistency with the labeled limb regions. The limb contour mask is then mapped onto an infrared thermogram sequence. The mapping process uses affine alignment based on a pre-calibrated spatial transformation matrix jointly defined by the visible light imaging component and the handheld infrared thermometer to obtain the effective skin temperature region. The equipment attitude angle difference is determined by the square root of the sum of the squares of the pitch angle difference, yaw angle difference, and roll angle difference corresponding to adjacent infrared thermal image frames. Based on the ranging sequence and the equipment attitude angle sequence, jump detection is performed on the ranging difference and equipment attitude angle difference between adjacent infrared thermal image frames. When the ranging difference is greater than the sum of the mean and standard deviation of the differences between adjacent frames in the ranging sequence, or when the equipment attitude angle difference is greater than the sum of the mean and standard deviation of the differences between adjacent frames in the equipment attitude angle sequence, the corresponding infrared thermal image frame is recorded as having a jump. Linear interpolation correction is performed on the jump point, and smooth correction is applied. The target is the jump point in the ranging sequence and the device attitude angle sequence. It does not change the pixel temperature value in the infrared thermal image sequence. The acquisition quality mark is generated according to the number of infrared thermal image frames with jumps. The acquisition quality mark participates in the infrared thermal image frame screening. The acquisition quality mark indicates that the infrared thermal image frames without jumps participate in the subsequent temperature depression identification. The acquisition quality mark indicates that the temperature depression candidate area formed in the infrared thermal image frame with jumps is not retained. The acquisition quality mark is used to distinguish the infrared thermal image frames that can participate in the analysis and the infrared thermal image frames that need to be downweighted before the subsequent temperature depression identification.
[0027] Z-score normalization is employed to perform dimensionless processing on the ranging sequence, device attitude angle sequence, and ambient temperature sequence. Z-score normalization centers and scales the data based on the global mean and global standard deviation of each sequence, allowing data of three different dimensions—distance, angle, and ambient temperature—to be included in the same subsequent calculation process. Maximum and minimum value normalization is used to perform interval normalization on the pixel temperature values in the infrared thermal image sequence and the pixel coordinates in the visible light limb image sequence. This maps pixel temperature values and pixel coordinates to a unified numerical range while maintaining inter-frame comparability. Pixel temperature values are linearly mapped to the 0-1 interval, and pixel coordinates are normalized to maximum and minimum values according to the width and height of a single frame in the visible light limb image sequence, forming a skin temperature field temporal data chain. This skin temperature field temporal data chain includes the infrared thermal image sequence after completion time correspondence, ambient temperature compensation, limb region constraints, and dimensionless numerical processing, as well as the limb contour mask, ranging sequence, device attitude angle sequence, and ambient temperature sequence.
[0028] This implementation scheme establishes a time-series data chain for skin temperature field that is consistent in time, regionally limited, and of controllable quality. This enables the pixel temperature values in the infrared thermogram sequence to be expressed within the effective skin temperature area corresponding to the examined limb, reducing the interference of background area, changes in acquisition posture, ranging fluctuations, and changes in ambient temperature on the skin temperature time-series analysis. It also improves the comparability and continuity of temperature changes between different infrared thermogram frames, providing a stable and reliable data foundation for subsequent identification of residual temperature indentations due to compression, stripping of regenerative segments after compression relief, and generation of time-series features of abnormal peripheral circulation.
[0029] S2: Perform limb segmentation and neighborhood temperature benchmark analysis on the skin temperature field time-series data chain to identify candidate areas for temperature depressions. Also generate residual compression markers and residual compression regions.
[0030] Specifically, the process of segmenting the limb measurement area and performing neighborhood temperature benchmark analysis on the skin temperature field time-series data chain to identify candidate areas for temperature depressions is as follows: The limb centerline is extracted based on the limb contour mask in the skin temperature field time-series data chain. The extraction of the limb centerline adopts the distance transform skeletonization algorithm. After performing Euclidean distance transformation on the binary region of the limb contour mask, the connected centerline is obtained by iterative search along the distance ridge. The effective skin temperature area is divided into root lateral measurement area, middle measurement area and peripheral measurement area along the direction of the limb centerline. The root lateral measurement area temperature matrix, middle measurement area temperature matrix and peripheral measurement area temperature matrix are extracted from the infrared thermogram sequence, respectively. The matrix elements of each temperature matrix are the pixel temperature values in the corresponding measurement area.
[0031] In the root-side temperature matrix, middle-section temperature matrix, and tip-side temperature matrix, a neighborhood temperature reference surface is established based on the median temperature of the neighborhood trimmed for the pixel to be determined. The median temperature is calculated by removing the highest and lowest quarters of the temperature values of all pixels within a window of a set size surrounding the pixel to be determined. The window corresponding to the median temperature is an odd-length pixel window surrounding the pixel to be determined and located within the same measurement area. Low-temperature pixels whose temperature values are lower than the neighborhood temperature reference surface and satisfy an eight-neighbor connectivity relationship are identified as candidate temperature depression areas. A pixel temperature value lower than the neighborhood temperature reference surface means that the difference between the neighborhood temperature reference surface and the pixel temperature value is greater than the standard deviation of the pixel temperature values within the odd-length pixel window. Combined with the acquisition quality marker, candidate temperature depression areas within infrared thermal image frames exhibiting ranging jumps or device attitude angle jumps are removed. The removal method is that if the acquisition quality marker of an infrared thermal image frame indicates a jump, all detected candidate temperature depression areas within that frame are marked as invalid and removed from the candidate set.
[0032] In this implementation scheme, a temperature matrix representation based on limb anatomical zoning is formed, enabling the effective skin temperature area to be systematically divided into root-lateral, mid-segment, and distal measurement zones, thus improving the spatial and temporal identifiability of low-temperature anomaly areas. The generation range of temperature depression candidate areas is constrained by the neighboring temperature reference surface and acquisition quality markers, reducing interference from false low-temperature areas caused by local noise, posture disturbances, and ranging fluctuations. This provides a foundation of clearly defined and reliable candidate areas for subsequent identification of compression-induced residual temperature depressions.
[0033] Specifically, the process of generating compression residue markers and compression residue regions is as follows: For temperature depression candidate regions, the infrared thermogram sequence is sorted according to the thermogram frame timestamps to obtain infrared thermogram frames. These frames are used to maintain the temporal order of the temperature depression candidate regions. The edge pixel temperature values and the pixel temperature values of the neighboring regions outside the outer contour of the temperature depression candidate regions are extracted. Edge pixels are pixels on the outer contour of the temperature depression candidate regions, and the outer neighboring regions are a set of eight neighboring pixels extending outwards from the outer contour of the temperature depression candidate regions without overlapping with them. The absolute value of the difference between the average edge pixel temperature and the average neighboring region pixel temperature is used as the depression boundary temperature difference, which reflects the temperature contrast between the temperature depression candidate regions and the surrounding normal tissue. The number of frames in which the temperature depression candidate regions maintain an eight-neighbor connectivity relationship in adjacent infrared thermogram frames is counted. An eight-neighbor connectivity relationship is defined as the existence of at least one pixel in the binary mask region of the temperature depression candidate regions in the current frame and the next frame that satisfies the eight-neighbor connectivity condition. The duration of the indentation is obtained by combining the timestamp of the thermal image frame. The duration is calculated by multiplying the number of frames maintaining an eight-neighbor connectivity relationship by the time interval between adjacent infrared thermal image frames. The number of pixels in each infrared thermal image frame is recorded, which is the total number of pixels contained in the candidate temperature indentation in the current frame. The change in indentation area is obtained by the average absolute value of the difference in the number of pixels between adjacent infrared thermal image frames. The change in indentation area is expressed as a change in pixel area and is calculated after removing infrared thermal image frames with ranging jumps or device attitude angle jumps.
[0034] The temperature difference at the concave boundary, the duration of the concave concavity, and the change in the concave area are input into the isolated forest anomaly detection process. This process uses these parameters from historical compression residue samples and non-compression low-temperature samples as training samples. The isolated forest anomaly detection process recursively constructs multiple isolated trees by randomly selecting segmentation features and values. Anomaly scores are calculated based on the average path length of the samples within the isolated trees, resulting in a compression residue anomaly score. A lower compression residue anomaly score indicates that the candidate temperature concave concavity is closer to the characteristics of a normal low-temperature region, while a higher score indicates that the candidate temperature concave concavity deviates from the characteristics of a normal low-temperature region. Compression residue is then identified in the candidate temperature concave concavity based on the compression residue anomaly score. When a candidate temperature concave concavity is identified as a compression residue region, a compression residue marker and a compression residue region are generated. The identification rules for compression residue regions are: the compression residue anomaly score reaches a compression residue score threshold, the concave concavity duration reaches a duration threshold, and the change in concave concavity area is within the allowable area change range. The compression residue score threshold is determined by the distribution quantiles of the compression residue anomaly scores from historical compression residue samples and non-compression low-temperature samples, and its value ranges from 0 to 1. The duration threshold is determined by the frame interval of the infrared thermogram sequence and the screening acquisition duration, ranging from one frame interval to the screening acquisition duration. The allowed area change range is determined by the pixel area change rate distribution of temperature indentation regions in historical compression residue samples across adjacent infrared thermogram frames, ranging from 0 to 1. When a temperature indentation candidate region is identified as a compression residue region, a compression residue marker and a compression residue region are generated.
[0035] This implementation plan establishes a temporal discrimination criterion for residual temperature depressions caused by compression. This expands the candidate region for temperature depressions from a single-frame low-temperature area to a temporal object that combines boundary temperature difference, continuous stability, and area change characteristics, thereby improving the ability to distinguish residual compression regions from ordinary low-temperature fluctuations and local temperature measurement noise. The residual compression anomaly score and the residual compression determination criteria together limit the generation range of residual compression markers, ensuring that subsequent compression relief and warming segment stripping can be based on stable, continuous regions with residual compression characteristics.
[0036] Reference manual attached Figure 2 The diagram shows a comparison of temperature changes related to the compression relief and warming segment and peripheral circulation abnormalities provided in the embodiments of the present invention.
[0037] Reference manual attached Figure 3 The flowchart of the linear temperature change region extraction and screening feature group generation provided by the embodiment of the present invention is shown.
[0038] S3: Perform inter-frame correlation and temperature change point detection on the residual compression area. Identify compression relief and temperature recovery segments, generate temperature recovery suppression markers, and screen skin temperature field sequences.
[0039] Specifically, the process of inter-frame correlation and temperature recovery change point detection based on the compression residual region is as follows: Based on the compression residual region and the timestamp of the thermal image frame, inter-frame correlation is performed on the compression residual region in the infrared thermal image frame. The time-series trajectory of the same region with compression residual mark, temperature depression state and subsequent temperature recovery state is defined as compression relief and temperature recovery candidate event. The time-series trajectory of the same region is established according to the region centroid distance and region area overlap relationship. Region pairs with a region area intersection-union ratio of zero do not enter the same region matching. Region pairs with a region area intersection-union ratio of non-zero are uniquely matched according to the correlation cost from small to large. Hungarian matching is used to minimize the correlation cost of all compression residual region pairs between adjacent frames. The correlation cost is composed of the normalized value of the region centroid distance and the complement of the region area intersection-union ratio with the same weight, forming the compression residual temperature trajectory and the compression residual area trajectory. Both the compression residual temperature trajectory and the compression residual area trajectory are time-series sequences arranged according to the timestamp of the thermal image frame.
[0040] The residual temperature trajectory is composed of the average regional temperature of the residual compression area within the infrared thermal image frame. The residual area trajectory is composed of the pixel area of the residual compression area within the infrared thermal image frame. Based on the residual temperature trajectory, change point detection is performed. The changes in the average regional temperature of adjacent infrared thermal image frames are compared sequentially along the timestamps of the thermal image frames. The infrared thermal image frame where the average regional temperature changes from decreasing or remaining flat to increasing is determined as the temperature recovery start frame. Change point detection adopts Bayesian online change point detection processing. By maintaining the posterior distribution of the travel length, the probability that the current point is a change point is calculated when each new data point arrives. The change point probability is used to characterize the degree of abrupt change in the temperature change location in the residual temperature trajectory.
[0041] In this implementation scheme, a continuous temporal trajectory representation of the compression residue region is formed, transforming the compression residue region in discrete infrared thermal image frames into a traceable compression residue temperature trajectory and compression residue area trajectory, thereby improving the stability of the same region at different acquisition times. The identification of the warming start frame clearly locates the transition node from the temperature depression state to the temperature recovery state, providing a reliable time reference for the subsequent extraction of the warming rate, warming duration, rebound temperature rise amplitude, and depression region shrinkage, thus enhancing the accuracy of compression relief and warming segment stripping.
[0042] Specifically, the process of identifying the compression relief and warming segment, generating warming inhibition markers, and screening skin temperature field sequences is as follows: Starting from the warming start frame, the first infrared thermogram frame where the average regional temperature changes from increasing to flat or decreasing is determined as the warming end frame, following the timestamp sequence of the thermogram frames. Based on the compression residual temperature trajectory and the thermogram frame timestamps, the warming rate, warming duration, and rebound temperature rise amplitude are generated. The warming rate is calculated as the ratio of the temperature change between the warming start frame and the warming end frame to the time span. The warming duration is determined by the absolute time length elapsed between the warming start frame and the warming end frame. The rebound temperature rise amplitude is the increment by which the peak temperature of the compression residual area exceeds the average temperature of the neighboring area during the warming process. Based on the pixel area difference between the warming start frame and the warming end frame in the compression residual area trajectory, the depression area shrinkage is generated. The depression area shrinkage is obtained by dividing the difference between the pixel area of the warming start frame and the pixel area of the warming end frame by the pixel area of the warming start frame, and is expressed as a relative shrinkage ratio. The residual temperature trajectory and residual area trajectory after the initial frame of the warming process are used to construct a warming morphology sequence. This sequence, combined with the warming rate, warming duration, rebound temperature rise amplitude, and depression area contraction, is then dynamically time-warped to calculate the distance to a pre-stored reference sequence for warming after compression relief. This reference sequence is formed from the warming morphology sequence, warming rate, warming duration, rebound temperature rise amplitude, and depression area contraction in historical warming samples. The dynamic time-warped distance calculation is performed by constructing a local cost matrix using the residual temperature trajectory and residual area trajectory from the tested warming morphology sequence and the reference sequence for warming after compression relief. As the input sequence, the local distance is obtained by adding the absolute value of the difference in the mean temperature of the alignment points and the absolute value of the difference in the pixel area. The alignment path with the minimum cumulative cost is found. The degree of nonlinear matching between the test sequence and the compression relief and warming reference sequence on the time axis is measured. The compression relief and warming segment is generated only when the warming morphology sequence and the compression relief and warming reference sequence have completed morphological alignment, the dynamic time warping distance is not higher than the corresponding distance distribution quantile of the historical compression relief and warming samples, and the warming rate, rebound temperature rise amplitude and depression area shrinkage amount show synchronous changes in temperature recovery and area shrinkage. This limitation is used to distinguish between simple temperature rise segments and transient warming segments after compression relief.
[0043] A warming suppression marker is assigned to the infrared thermogram temperature change within the compression relief warming segment. The warming suppression marker is bound to the thermogram frame timestamp and regional location of the compression relief warming segment. The warming suppression marker does not delete the infrared thermogram data, but only marks the compression relief warming segment as a data segment that does not participate in the calculation during the generation of the peripheral circulation screening feature group, thus obtaining the screening skin temperature field sequence. The screening skin temperature field sequence is the skin temperature field time series data after removing the interference of the compression relief warming segment.
[0044] In this embodiment, as Figure 2The figure shows a comparison of temperature changes related to the compression relief and warming segment and peripheral circulation anomalies. The horizontal axis represents time, indicating the acquisition time corresponding to the infrared thermogram frames arranged according to their timestamps. The vertical axis represents temperature, indicating the average regional temperature at the corresponding time. The compression relief and warming curve represents the temperature change process in the residual compression area after compression is relieved. The curve initially shows a low-temperature depression, then enters a temperature recovery process from the warming start frame, and stabilizes after the warming end frame. The peripheral circulation anomaly-related temperature change curve serves as a comparison curve, representing the temporal change of skin temperature that does not steadily recover according to the compression relief and warming pattern. The black dots in the figure indicate the positions of the warming start frame, warming end frame, and rebound temperature peak point on the curve. The vertical double arrows indicate the rebound temperature amplitude, and the horizontal double arrows indicate the warming duration. As can be seen from the two types of curve morphologies in the figure, the compression relief warming segment exhibits the temporal characteristics of a low-temperature depression followed by a rebound, a clearly defined warming duration, and a tendency to stabilize after warming. In contrast, the temperature change curve related to peripheral circulation anomalies shows continuous fluctuations or abnormal increases. Therefore, the compression relief warming segment can be identified by calculating the dynamic time-normalized distance between the warming morphology sequence and the compression relief warming reference sequence, providing a basis for the subsequent generation of warming suppression markers. The warming suppression marker is bound to the timestamp and regional location of the thermal image frame of the compression relief warming segment. The original infrared thermal image data and the temperature curves in the figure are still retained, but are not included in the calculation during the generation of the peripheral circulation screening feature group.
[0045] This implementation plan establishes an interference stripping mechanism for the compression relief and warming process. This mechanism enables the comprehensive identification of compression relief and warming segments based on temperature recovery patterns, area contraction patterns, and reference time-series patterns, improving the ability to distinguish between transient recovery warming and peripheral circulation-related temperature changes. The warming suppression marker preserves the integrity of the original infrared thermogram data while reducing the impact of compression relief and warming segments on subsequent time-series analysis. This results in a screening skin temperature field sequence that better reflects the temporal transmission relationship of peripheral circulation, enhancing the stability and reliability of anomaly screening results.
[0046] S4: Perform root-to-peripheral temperature time-series analysis and temperature gradient inversion analysis on the screening skin temperature field sequence, and generate temperature gradient inversion markers. Construct vascular course temperature variation zones to form a peripheral circulation screening feature group.
[0047] Specifically, the process of performing root-to-tip temperature time-series analysis and temperature gradient reversal analysis on the screening skin temperature field sequence and generating temperature gradient reversal markers is as follows: In the screening skin temperature field sequence, the mean regional temperature of the root-side measurement area, the middle section measurement area and the tip measurement area are extracted frame by frame, and after being arranged according to the timestamp of the heat map frame, the root-side temperature change sequence, the middle section temperature change sequence and the tip temperature change sequence are obtained. Cross-correlation time-series alignment processing was used to calculate the time delay between the root-side temperature change sequence and the tip-side temperature change sequence. The cross-correlation time-series alignment processing used the complete screening acquisition period as the calculation window. Before the calculation, the root-side temperature change sequence and the tip-side temperature change sequence were respectively subjected to mean removal and standard deviation normalization. The cross-correlation time-series alignment processing calculated the cross-correlation coefficient of the two sequences under different lag step sizes. The search range of the lag step size is the positive lag step size of the root-side temperature change sequence and the tip-side temperature change sequence with overlapping sampling points. The time delay was obtained by multiplying the lag step size corresponding to the largest cross-correlation coefficient by the frame interval. When the cross-correlation coefficients corresponding to multiple lag step sizes are the same, the lag step size with the smaller value is selected to obtain the root-side to tip-side temperature lag time. The root-side to tip-side temperature lag time represents the time required for the temperature signal to be transmitted from the root of the limb to the tip.
[0048] A cross-sectional temperature gradient is calculated along the limb centerline of the screening skin temperature field sequence. This cross-sectional temperature gradient is approximated by central difference of the temperature values at equally spaced sampling points along the limb centerline, yielding the rate of temperature change between adjacent sampling points and thus the root-to-terminal temperature gradient sequence. When the root-to-terminal temperature gradient sequence exhibits a direction reversal in the infrared thermogram frame, and the gradient direction sign remains consistent across n consecutive frames, a temperature gradient reversal marker is generated. The gradient direction sign change is defined as a gradient value changing from positive to negative or vice versa. The n-frame consistency requirement indicates that the reversal is not caused by inter-frame noise but is a genuine temperature gradient reversal phenomenon with temporal continuity. n is a positive integer, calculated from the frame interval of the infrared thermogram sequence and the duration quantile of historically valid reversal samples, ranging from 2 to the total number of screening frames.
[0049] This implementation scheme establishes a temporal characterization of skin temperature conduction from the root side to the extremity, enabling the screening skin temperature field sequence to be transformed into root side temperature change sequences, mid-section temperature change sequences, extremity temperature change sequences, and root side to extremity temperature gradient sequences. This enhances the analyzability of peripheral circulation-related temperature changes in terms of conduction velocity, spatial gradient, and duration. The root side to extremity temperature lag time and temperature gradient reversal markers jointly characterize the temporal delay and gradient anomalies of skin temperature changes along the limb direction, reducing the impact of single-frame temperature variations or localized warming on screening results. This provides temporally stable and spatially directional feature data for peripheral circulation screening feature groups.
[0050] Specifically, the process of constructing vascular temperature variation zones to form a peripheral circulation screening characteristic group is as follows: Figure 3 The flowchart shown is for the extraction and screening of linear temperature change regions and the generation of feature groups. The modulus of the temperature difference between adjacent pixels in the horizontal and vertical directions is calculated frame by frame for the screened skin temperature field sequence to obtain a temperature gradient map. The temperature gradient map is used to characterize the temperature change boundary and linear temperature change structure in the skin temperature field. Linear temperature change structure enhancement processing based on the Hessian matrix is performed. The Hessian matrix is used to extract the thermal linear temperature change structure from the temperature gradient map to obtain a thermal linear candidate response map. This processing calculates the Hessian matrix of the temperature gradient map, performs eigenvalue decomposition on the Hessian matrix, and uses the ratio and modulus of the two eigenvalues to enhance the linear structure and suppress the point and area structures. The candidate thermal response map is connected to eight neighborhoods along the limb centerline. The eight-neighbor connectivity is performed using a region growing algorithm, with pixels in the thermal response map whose response values are higher than the response threshold as seed points. Pixels that satisfy the eight-neighbor connectivity relationship and directional consistency are searched along the limb centerline and the extension direction of the thermal response map to form a continuous three-dimensional tubular region. The response threshold is determined by the distribution quantile of the response values of the effective linear temperature change region and the background temperature change region in the historical thermal response map, with a value range of 0 to 1. The resulting vascular course temperature change zone is a linear temperature change region that extends longitudinally along the limb and has temporal continuity in the infrared thermogram sequence. It is not used as a direct location result of the actual vascular anatomy.
[0051] Temperature change sequences of vascular course temperature-varying zones were extracted. These sequences were obtained by arranging the average regional temperature within the vascular course temperature-varying zones according to the timestamps of the thermal image frames. A continuous temperature change quantity for the vascular course temperature-varying zones was generated based on the regional overlap ratio and the number of frames with the same temperature change direction in adjacent infrared thermal image frames. This continuous quantity was obtained by multiplying the average regional overlap ratio of the vascular course temperature-varying zones in adjacent infrared thermal image frames by the proportion of frames with the same temperature change direction to the total number of adjacent frames, with a value ranging from 0 to 1. The continuous quantity of the vascular course temperature-varying zones characterizes the temporal continuity of the vascular course temperature-varying zones in terms of spatial location and temperature change direction. The root-to-terminal temperature lag time, temperature gradient reversal marker, continuous quantity of the vascular course temperature-varying zones, and back-temperature inhibition marker were combined to form a peripheral circulation screening feature set.
[0052] This implementation scheme establishes a temporally continuous feature representation for linear temperature-changing structures. This allows temperature abrupt change boundaries and longitudinally extending temperature-changing regions in the skin temperature field sequence to be transformed into vascular course temperature-changing bands and their temperature change sequences, enhancing the ability to identify linear temperature-changing regions in both spatial extension and temporal continuity. The continuity of temperature changes within these bands reflects the degree of positional continuity and temperature change consistency of the vascular course temperature-changing bands in adjacent infrared thermogram frames. Combined with root-to-terminal temperature lag time, temperature gradient reversal markers, and warming inhibition markers, this forms a peripheral circulation screening feature set, providing a more complete spatiotemporal feature basis for subsequent intelligent screening and result output.
[0053] Reference manual attached Figure 4 The image shows a heatmap of the screening prompt area provided in an embodiment of the present invention.
[0054] S5: Generate screening markers based on peripheral circulation screening feature groups, and output retest prompts, screening prompt reports, abnormal area heat maps, and abnormal basis labels.
[0055] Specifically, the process of generating screening labels based on peripheral circulation screening feature groups and outputting retest prompts, screening prompt reports, abnormal area heatmaps, and abnormality basis labels is as follows: The acquisition quality labels and peripheral circulation screening feature groups are input into a gated temporal convolutional network. These labels, arranged according to the heatmap frame timestamps, serve as the input to the network. The activation states of the corresponding retest prompt labels, peripheral circulation screening labels, and vascular course temperature change screening labels in historical samples are used as the training output. The gated temporal convolutional network consists of alternating layers of dilated causal convolutions and gated linear units. The receptive field range is controlled by the dilation factor to capture the multi-scale dependencies of temporal features, generating retest prompt labels, peripheral circulation screening labels, and vascular course temperature change screening labels. The vascular course temperature change screening label is generated using continuous temperature changes along the course of the vessel. Each screening label is a binary output, taking the value of "label activated" or "label inactive." When a retest prompt label is generated, a retest prompt is output. When peripheral circulation screening markers or vascular course temperature change screening markers are generated, a screening prompt report, an abnormal area heat map, and an abnormality basis label are output by combining the root-to-terminal temperature lag time, temperature gradient reversal marker, and continuous temperature change of the course zone. The abnormal area heat map marks the boundary of the abnormal area on the original infrared heat map in a pseudo-color overlay. The boundary of the abnormal area comes from the infrared heat map frame area corresponding to the vascular course temperature change zone that triggered the screening marker and the temperature gradient reversal marker. The abnormality basis label lists the key quantitative indicators involved in the screening prompt and their deviation from the historical sample baseline range in text form. The abnormality basis label includes at least the prompt content corresponding to the root-to-terminal temperature lag time, temperature gradient reversal marker, continuous temperature change of the course zone, and temperature rebound inhibition marker. The screening prompt report is a prompt output formed based on the screening marker and the corresponding key quantitative indicators, and is not a disease diagnosis conclusion.
[0056] In this embodiment, after the screening skin temperature field sequence is input into the intelligent screening and result output module, a screening prompt report is generated based on the peripheral circulation screening feature group, and the screening prompt report is superimposed on the corresponding infrared thermal image frame to form a... Figure 4 The heat map showing the screening prompt area is shown. Figure 4 The left side displays the original infrared thermogram, the middle side displays the abnormal area thermogram after overlaying the screening warning area boundaries, and the right side displays a magnified local area. The dashed white boundaries in the image represent the limb outline, and the lateral, mid-section, and distal measurement areas are segmented measurement areas obtained based on the limb's centerline direction. The relative temperature bands represent the relative temperature range after normalizing the pixel temperature values in the infrared thermogram sequence. The red, yellow, and blue dashed boundaries represent high-temperature, medium-temperature, and low-temperature warning areas, respectively, used to characterize the relative degree of temperature rise within the screening warning area. The gray dashed rectangles represent magnified local areas, and the gray dashed connecting lines indicate the correspondence between the magnified local areas and the magnified image on the right. Figure 4 The screening prompt area heat map is only used to display the location and relative temperature changes of the area prompt in the infrared heat map. It is not used as a conclusion to diagnose the disease, nor does it output treatment plans, medication plans or surgical treatment suggestions.
[0057] In this embodiment, the same batch of infrared thermogram sequences of the examined limbs were used to run a control process without a warming suppression marker and a screening process with a warming suppression marker. The control process directly used the temperature rise segment to generate a peripheral circulation screening feature group. The screening process first generated the compression residual area through the compression residual temperature indentation identification module, and then generated the warming suppression marker through the compression relief warming segment stripping module. In the peripheral circulation abnormality time sequence feature generation module, the participation weight of the compression relief warming segment was reduced. The number of false alarms corresponding to the transient recovery warming segment, the consistency rate of screening results under repeated acquisition of the same examined limb, the area under the receiver operating characteristic curve, sensitivity, and specificity were statistically analyzed for both processes. When the screening process showed a reduction in the number of false alarms, an increase in the consistency rate of screening results, an increase in the area under the receiver operating characteristic curve, and sensitivity and specificity within the preset statistical evaluation range compared to the control process, it was determined that the warming suppression marker had a stripping effect on the short-term warming interference of compression relief.
[0058] This implementation plan establishes an intelligent judgment and visualization output mechanism for peripheral circulation anomaly screening. Data acquisition quality markers can trigger retest prompts for low-quality data. Peripheral circulation screening feature groups support the classification and generation of retest prompt markers, peripheral circulation screening markers, and vascular course temperature change screening markers, improving the screening prompt report's ability to identify multi-scale skin temperature temporal changes. Abnormal area heatmaps and abnormal basis labels simultaneously present the screening prompt location and key quantitative evidence, enhancing the interpretability and review convenience of screening results. At the same time, it limits the auxiliary prompt attributes of the screening prompt report, avoiding the output of disease diagnosis conclusions, treatment plans, medication plans, or surgical intervention suggestions.
[0059] Reference manual attached Figure 5 The diagram shows a schematic of the structure of an intelligent screening system for peripheral circulation abnormalities based on the temporal variation of skin temperature field provided by the present invention.
[0060] The present invention also provides an intelligent screening system 20 for peripheral circulation abnormalities based on temporal changes in skin temperature field, applied to the above-mentioned intelligent screening method for peripheral circulation abnormalities based on temporal changes in skin temperature field, comprising: Processor 201.
[0061] The memory 202 stores computer-readable instructions. When the computer-readable instructions are executed by the processor 201, they implement the intelligent screening method for peripheral circulation anomalies based on the temporal variation of skin temperature field as described in the method embodiment.
[0062] The intelligent screening system 20 for peripheral circulation abnormalities based on the temporal variation of skin temperature field provided by the present invention can perform the above-mentioned intelligent screening method for peripheral circulation abnormalities based on the temporal variation of skin temperature field and achieve the same or similar technical effects. To avoid repetition, the present invention will not elaborate further.
[0063] It should be understood that the processor in the embodiments of the present invention can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.
[0064] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).
[0065] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.
[0066] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.
[0067] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.
[0068] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0069] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0070] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0071] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.
[0072] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0073] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0074] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0075] This invention provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the intelligent screening method for peripheral circulation anomalies based on the temporal variation of skin temperature field as described in the method embodiment.
[0076] The present invention provides a computer-readable storage medium that can implement the steps and effects of the intelligent screening method for peripheral circulation anomalies based on the temporal variation of skin temperature field in the above-described method embodiments. To avoid repetition, the present invention will not elaborate further.
[0077] 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.
[0078] The following points need to be explained: (1) The accompanying drawings of the embodiments of the present invention only involve the structures involved in the embodiments of the present invention. Other structures can refer to the general design.
[0079] (2) For clarity, the thickness of layers or regions is enlarged or reduced in the drawings used to describe embodiments of the invention, i.e., these drawings are not drawn to scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element or there may be intermediate elements.
[0080] (3) Where there is no conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.
[0081] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for intelligent screening of peripheral circulation abnormalities based on temporal changes in skin temperature field, characterized in that, include: S1: Collect multi-source skin temperature screening data and preprocess it to form a skin temperature field time series data chain; S2: Perform limb segmentation and neighborhood temperature benchmark analysis on the skin temperature field time-series data chain to identify candidate temperature depression areas; and generate compression residue markers and compression residue areas. S3: Perform inter-frame correlation and temperature change detection on the compressed residual area; It also identifies compression relief and warming fragments, generates warming inhibition markers, and screens skin temperature field sequences; S4: Perform root-to-periphery temperature time-series analysis and temperature gradient reversal analysis on the screening skin temperature field sequence to generate temperature gradient reversal markers; and construct a vascular course temperature change zone to form a peripheral circulation screening feature group. S5: Generate screening markers based on peripheral circulation screening feature groups, and output retest prompts, screening prompt reports, abnormal area heat maps, and abnormal basis labels.
2. The intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field according to claim 1, characterized in that, S1 specifically includes: S101: The handheld infrared thermometer collects infrared thermal image sequences and thermal image frame timestamps along the direction from the root to the tip of the limb being examined using a handheld infrared thermometer; the visible light imaging component simultaneously collects visible light limb image sequences; the ranging component collects ranging sequences; the attitude sensing component collects device attitude angle sequences; and the ambient temperature sensing component collects ambient temperature sequences. S102: Perform frame-level correspondence between the infrared thermal image sequence and the visible light limb image sequence according to the thermal image frame timestamp, and perform bad pixel replacement, temperature range verification, inter-frame smoothing and temperature compensation based on the ambient temperature sequence on the infrared thermal image sequence, and perform background cropping and limb region preservation processing on the visible light limb image sequence. S103: A semantic segmentation network based on dilated convolution is used to segment the limb regions of the processed visible light limb image sequence to generate a limb contour mask; and the limb contour mask is mapped onto the infrared thermogram sequence to obtain the effective skin temperature region. S104: Determine the equipment attitude angle difference based on the square root of the sum of the squares of the pitch angle difference, yaw angle difference, and roll angle difference corresponding to adjacent infrared thermal image frames. Based on the ranging sequence and the equipment attitude angle sequence, identify jumps in the ranging difference and equipment attitude angle difference between adjacent infrared thermal image frames, record the infrared thermal image frames with jumps, perform linear interpolation correction on the jump points, and generate acquisition quality markers according to the number of infrared thermal image frames with jumps. The acquisition quality markers participate in the infrared thermal image frame screening. S105: Z-score normalization is used to perform dimensionless processing on the corrected ranging sequence, device attitude angle sequence, and ambient temperature sequence; maximum and minimum value normalization is used to perform interval normalization processing on the pixel temperature values in the effective skin temperature region and the pixel coordinates in the visible light limb image sequence; according to the timestamp of the thermal image frame, the processed pixel temperature values, pixel coordinates in the visible light limb image sequence, ranging sequence, device attitude angle sequence, ambient temperature sequence, and acquisition quality marker are temporally correlated to form the skin temperature field temporal data chain.
3. The intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field according to claim 2, characterized in that, In step S2, the skin temperature field time-series data chain is divided into limb segmentation measurement areas and neighborhood temperature benchmark analysis is performed to identify candidate areas for temperature depressions. Specifically, this includes: S201: Extract the limb centerline based on the limb contour mask in the skin temperature field time-series data chain, divide the effective skin temperature area into root lateral measurement area, middle measurement area and peripheral measurement area along the direction of the limb centerline, and extract the root lateral measurement area temperature matrix, middle measurement area temperature matrix and peripheral measurement area temperature matrix from the infrared thermogram sequence respectively, and the matrix elements of each temperature matrix are the pixel temperature values in the corresponding measurement area; S202: In the root side temperature matrix, the middle section temperature matrix and the tip temperature matrix, establish a neighborhood temperature reference surface based on the neighborhood trimming median temperature of the pixel to be determined. S203: Determine the set of low-temperature pixels whose pixel temperature values are lower than the neighborhood temperature reference plane and satisfy the eight-neighbor interconnection relationship as the initial temperature depression candidate area; and combine the acquisition quality mark to remove abnormal acquisition frames in the initial temperature depression candidate area that have ranging jumps or device attitude angle jumps, so as to obtain the temperature depression candidate area.
4. The intelligent screening method for peripheral circulation anomalies based on the temporal variation of skin temperature field according to claim 1, characterized in that, The generation of compression residue markers and compression residue areas in S2 specifically includes: S204: For the temperature depression candidate area, the infrared thermal image sequence is sorted according to the thermal image frame timestamp to obtain an infrared thermal image frame, and the edge pixel temperature value of the temperature depression candidate area and the pixel temperature value of the neighborhood area outside the outer contour of the temperature depression candidate area are extracted. S205: The absolute value of the difference between the average temperature of the edge pixels and the average temperature of the pixels in the neighboring region is taken as the temperature difference of the depression boundary; the number of frames in which the temperature depression candidate area maintains an eight-neighbor connection relationship in adjacent infrared thermal image frames is counted, and the depression duration is obtained by combining the thermal image frame timestamp; the number of pixels of the temperature depression candidate area in each infrared thermal image frame is recorded, and the change in depression area is obtained based on the average absolute value of the difference in the number of pixels between adjacent infrared thermal image frames. S206: Input the temperature difference at the concave boundary, the duration of the concave, and the change in the area of the concave into the isolated forest anomaly detection processing to obtain the compression residue anomaly score; and identify compression residue in the temperature concave candidate area based on the compression residue anomaly score; when the temperature concave candidate area is identified as a compression residue area, generate a compression residue marker and a compression residue area.
5. The intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field according to claim 1, characterized in that, The S3 step involves inter-frame correlation and temperature change point detection of the compression residue area, specifically including: S301: Based on the compression residue area and the timestamp of the thermal image frame, perform inter-frame correlation on the compression residue area in the infrared thermal image frame, and establish a time-series trajectory of the same area according to the centroid distance and area overlap relationship of the area, forming the compression residue temperature trajectory and the compression residue area trajectory. S302: Based on the residual temperature trajectory of the compression, change point detection is performed, and the changes in the average regional temperature of adjacent infrared thermal images are compared in the order of the timestamp of the thermal image frame. The infrared thermal image frame in which the average regional temperature changes from decreasing or remaining flat to increasing is determined as the warming start frame.
6. The intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field according to claim 5, characterized in that, The process in S3, which identifies compression relief and warming fragments, generates warming inhibition markers, and screens skin temperature field sequences, specifically includes: S303: Starting from the warming start frame, the first infrared thermal image frame in which the average regional temperature changes from rising to flat or falling according to the timestamp sequence of the thermal image frames is determined as the warming end frame. S304: Generate the warming rate, warming duration, and rebound temperature rise amplitude based on the residual temperature trajectory of the compression and the timestamp of the thermal image frame; generate the shrinkage amount of the concave region based on the pixel area difference between the warming start frame and the warming end frame in the residual area trajectory of the compression. S305: Based on the residual temperature trajectory and residual area trajectory after the recovery start frame, a recovery morphology sequence is constructed. Combined with the recovery rate, the recovery duration, the rebound temperature rise amplitude, and the shrinkage amount of the concave area, dynamic time-normalized distance calculation is performed with the pre-stored compression relief recovery reference sequence to obtain the compression relief recovery segment. S306: Assign a warming inhibition marker to the infrared thermogram temperature change within the compression relief warming segment to obtain a screening skin temperature field sequence.
7. The intelligent screening method for peripheral circulation anomalies based on temporal changes in skin temperature field according to claim 3, characterized in that, In step S4, root-to-tip temperature time-series analysis and temperature gradient inversion analysis are performed on the screened skin temperature field sequence to generate temperature gradient inversion markers, specifically including: S401: In the screening skin temperature field sequence, the mean regional temperature of the root lateral measurement area, the middle section measurement area and the terminal measurement area are extracted frame by frame, and then arranged according to the timestamp of the heat map frame to obtain the root lateral temperature change sequence and the terminal temperature change sequence. S402: The time delay between the root-side temperature change sequence and the tip temperature change sequence is calculated by using cross-correlation time alignment processing to obtain the root-side to tip temperature lag time. S403: Calculate the cross-sectional temperature gradient of the screening skin temperature field sequence along the limb centerline to obtain the root-to-terminal temperature gradient sequence; when the root-to-terminal temperature gradient sequence shows a direction reversal in the infrared thermogram frame and the gradient direction sign changes consistently in multiple consecutive frames, generate a temperature gradient reversal marker.
8. The intelligent screening method for peripheral circulation anomalies based on the temporal variation of skin temperature field according to claim 7, characterized in that, In S4, a temperature-varying vascular pathway is constructed to form a peripheral circulation screening feature group, specifically including: S404: Calculate the magnitude of the temperature difference between adjacent pixels in the horizontal and vertical directions frame by frame for the screened skin temperature field sequence to obtain a temperature gradient map. Then, through the linear temperature change structure enhancement processing based on the Hessian matrix, extract the thermal linear temperature change structure from the temperature gradient map to obtain a thermal linear candidate response map. S405: Connect the hot linear candidate response map along the limb centerline direction with eight neighboring regions to obtain the vascular course temperature change zone as a linear temperature change region. S406: Extract the temperature change sequence of the vascular course temperature change zone from the vascular course temperature change zone, generate a continuous temperature change quantity of the course zone based on the regional overlap ratio of the vascular course temperature change zone in adjacent infrared thermogram frames and the number of frames with the same temperature change direction; and combine the root-to-terminal temperature lag time, temperature gradient reversal marker, continuous temperature change quantity of the course zone and back-temperature inhibition marker to form a peripheral circulation screening feature group.
9. The intelligent screening method for peripheral circulation anomalies based on the temporal variation of skin temperature field according to claim 8, characterized in that, S5 specifically includes: S501: Input the acquisition quality label and peripheral circulation screening feature group into the gated temporal convolutional network to generate retest prompt label, peripheral circulation screening label and vascular course temperature change screening label; S502: When the retest prompt mark is generated, a retest prompt is output; when the peripheral circulation screening mark or the vascular course temperature change screening mark is generated, in combination with the root-to-terminal temperature lag time, the temperature gradient reversal mark and the course temperature change continuity, a screening prompt report, an abnormal area heat map and an abnormal basis label are output.
10. A smart screening system for peripheral circulation abnormalities based on temporal changes in skin temperature field, characterized in that, include: processor; A memory storing computer-readable instructions, which, when executed by the processor, implement the intelligent screening method for peripheral circulation anomalies based on the temporal variation of skin temperature field as described in any one of claims 1 to 9.
Citation Information
Patent Citations
Dual-examination comprehensive diagnostic instrument using infrared thermal image-guided ultrasound and detection method of the diagnostic instrument
CN103385734B
A remote diagnosis and treatment system for knee osteoarthritis based on infrared imaging
CN109394180B
