Shoulder joint semi-dislocation image data processing method and device and medium

Abstract

The invention provides a shoulder joint semi-dislocation image data processing method and device and a medium. According to the method, medical image data of a first area of a shoulder joint of a to-be-detected object is obtained, then motion artifact detection processing is performed on the medical image data of the first area, motion artifact level information of the medical image data is obtained, and under the condition that the motion artifact level information indicates that the medical image data meets a preset quality condition, motion artifact detection is performed on the shoulder joint of the to-be-detected object. And performing shoulder joint anatomical mark detection and distance measurement processing on the medical image data to obtain a first distance parameter between the acromion and the humeral head and / or a second distance parameter between the acromion and the humeral greater tubercle, and finally, determining a shoulder joint subluxation evaluation result of the to-be-detected object according to the first distance parameter and / or the second distance parameter. Therefore, accurate and repeatable quantitative measurement of the distance between the acromion and the humeral head and the distance between the acromion and the humeral greater tubercle can still be realized in a high-exercise-risk population, namely a stroke patient.

Description

Technical Field

[0001] This application relates to data processing technology, and more particularly to a method, device and medium for processing image data of shoulder subluxation. Background Technology

[0002] In the assessment of shoulder subluxation, clinicians mainly rely on physicians to manually identify bony anatomical landmarks such as the acromion, humeral head, and greater tuberosity of the humerus on shoulder X-ray and planar ultrasound images, and to manually measure the acromion. Distance between humeral heads and acromion The distance between the greater tuberosity of the humerus is measured, and then the degree of shoulder subluxation is graded based on the ratio of the distance between the affected and healthy sides.

[0003] However, stroke patients commonly exhibit abnormal upper limb muscle tone, poor trunk control, and difficulty maintaining a fixed posture for extended periods. During X-ray exposure and ultrasound scanning, involuntary movements, respiratory movements, and relative probe slippage are highly likely to occur, resulting in significant motion artifacts in medical image data. This leads to blurred, trailing, or double contours of bony edges, making it difficult to locate anatomical landmarks, causing large fluctuations in distance measurement results, and resulting in poor repeatability and objectivity. Consequently, it is difficult to meet the clinical needs for accurate and quantitative assessment of shoulder subluxation. Summary of the Invention

[0004] This application provides a method, device, and medium for processing image data of shoulder subluxation, which solves the technical problems of unstable ranging and unreliable assessment caused by motion artifacts in stroke patients, a high-risk group for movement.

[0005] In a first aspect, this application provides a method for processing image data of shoulder subluxation, including: Acquire medical image data of a first region of the shoulder joint of the subject under test, the first region including the acromion bony structure and the humeral head and / or the greater tubercle bony structure; Motion artifact detection processing is performed on the medical image data of the first region to obtain motion artifact level information of the medical image data. The motion artifact level information is used to indicate the intensity of motion artifacts in the medical image data. When the motion artifact level information indicates that the medical image data meets the preset quality conditions, shoulder joint anatomical landmark detection and distance measurement are performed on the medical image data to obtain a first distance parameter between the acromion and the humeral head and / or a second distance parameter between the acromion and the greater tubercle of the humerus. The shoulder subluxation assessment result of the subject is determined based on the first distance parameter and / or the second distance parameter.

[0006] Secondly, this application provides an electronic device, comprising: Processor; and, Memory for storing the executable instructions of the processor; The processor is configured to perform any of the possible methods described in the first aspect by executing the executable instructions.

[0007] Thirdly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement any of the possible methods described in the first aspect.

[0008] The shoulder subluxation image data processing method, device, and medium provided in this application acquire medical image data of a first region of the shoulder joint of the subject under test, then perform motion artifact detection processing on the medical image data of the first region to obtain motion artifact level information of the medical image data. When the motion artifact level information indicates that the medical image data meets preset quality conditions, shoulder joint anatomical landmark detection and distance measurement processing are performed on the medical image data to obtain a first distance parameter between the acromion and the humeral head and / or a second distance parameter between the acromion and the greater tubercle of the humerus. Finally, the shoulder joint subluxation assessment result of the subject under test is determined based on the first distance parameter and / or the second distance parameter, thereby improving the acromion... Humeral head distance and acromion The greater tuberosity distance of the humerus can still be accurately and reproducibly measured in stroke patients, a high-risk population for movement, thus solving the technical problems of unstable distance measurement and unreliable assessment caused by motion artifacts in the prior art. Attached Figure Description

[0009] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0010] Figure 1 This is a schematic flowchart illustrating a method for processing image data of shoulder subluxation according to an exemplary embodiment of this application; Figure 2 This application illustrates a flowchart of a specific implementation of S120 based on an example embodiment; Figure 3 This application illustrates a flowchart of a specific implementation of S120 according to another exemplary embodiment; Figure 4 This application illustrates a flowchart of a specific implementation of S130 according to another exemplary embodiment; Figure 5 This is a schematic diagram of the structure of an electronic device according to an example embodiment of this application.

[0011] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0012] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0013] Shoulder joint X-ray imaging is an imaging method that integrates transmitted rays during the exposure time. When the acromion or humeral head is displaced during the exposure, its bony contour is equivalent to convolving with a one-dimensional blur kernel along the direction of movement. This results in an increase in the grayscale transition width of the bone boundary, a decrease in the gradient peak value, and even the appearance of double or multiple lines of the same anatomical boundary. This causes key anatomical contours such as the lower edge of the acromion and the upper edge of the humeral head to degenerate from a clear single sharp edge to a wide edge or multiple edge bands.

[0014] Furthermore, shoulder ultrasound images are high-frame-rate, frame-by-frame slice images. Patient trunk translation, shoulder girdle micromovement, and relative probe slippage can cause significant translation and deformation of the hyperechoic lines of the cortical bone in adjacent frames. Within a single frame, this manifests as blurred or trailing edges of the cortical bone. When multiple frames are superimposed or simply averaged, the contrast and edge sharpness between the cortical bone and surrounding tissues are further reduced, making the acromion appear less prominent. In the measurement of the acromion–greater tuberosity distance (AGT), the true cortical positions of the acromion and greater tuberosity of the humerus are difficult to extract stably, resulting in millimeter-level systematic errors and random fluctuations, which directly weakens the acromion-based measurement method. The reliability of assessments based on the Acromiohumeral Distance (AHD) and the AGT shoulder subluxation grading system (GHS).

[0015] To address the aforementioned issues, the embodiments provided in this application, for shoulder joint X-ray images, perform edge detection within the region of interest, including the acromion and humeral head, extract grayscale profiles along the normal direction of the bony edges, and calculate the amplitude of the first derivative of the grayscale values. This allows for the estimation of the local edge diffusion width at various locations. The diffusion widths of the acromion and humeral head edges are statistically analyzed and compared with preset artifact-free baseline statistical parameters to form a first motion artifact index reflecting the degree of edge blurring. Simultaneously, a second motion artifact index reflecting the degree of dual contours is obtained through gradient multi-peak detection and calculation of the length ratio of dual-line candidate regions. Based on this, the X-ray image is classified into motion artifact levels. Furthermore, during the ranging stage, robust statistical methods such as least-squares fitting of the bone edge pixel set, removal of outliers with significantly weak gradients, and use of the median of multi-point geometric distances are employed to reduce the influence of local artifacts on the first distance parameter.

[0016] For shoulder joint ultrasound images, two-dimensional rigid body registration is performed on adjacent frames within a third region of interest containing the acromion and the greater tubercle of the humerus to obtain an inter-frame translation vector sequence. The cumulative motion is calculated within a sliding time window to automatically identify stationary windows where the cumulative motion is below a threshold. Then, within the stationary window, the image quality of each frame is quantitatively scored based on the local contrast of the cortical bone and the edge gradient intensity. The target frame with the best clarity is selected, and the cortical bone position is detected along a preset measurement line and its adjacent measurement lines. Statistical fusion is performed using the median or extreme value averaging of distance results from multiple measurement lines and multiple frames. Finally, without relying on neural networks or large models, the first and second distance parameters on which the assessment of shoulder subluxation depends still have good accuracy and reliability in actual acquisition environments with motion interference.

[0017] Figure 1 This is a schematic flowchart illustrating a method for processing image data of shoulder subluxation according to an example embodiment of this application. Figure 1 As shown, the method provided in this embodiment includes: S110. Obtain medical image data of the first region of the shoulder joint of the subject under test.

[0018] In this step, medical image data of the first region of the shoulder joint of the subject is acquired, wherein the first region includes the acromion bony structure and the humeral head and / or the greater tubercle bony structure.

[0019] Specifically, when acquiring shoulder joint X-ray images, based on preset standard projection parameters for the shoulder joint, the X-ray acquisition device is controlled to acquire original shoulder joint X-ray images containing the acromion bony structure and the humeral head bony structure at the shoulder joint of the subject. In the original shoulder joint X-ray images, at least one first positioning point is determined based on the spatial positional relationship and gray-scale distribution characteristics of the acromion bony structure, and at least one second positioning point is determined based on the spatial positional relationship and gray-scale distribution characteristics of the humeral head bony structure and / or the greater tuberosity bony structure. Based on the relative positional relationship between the at least one first positioning point and the at least one second positioning point, a first region shoulder joint X-ray sub-image containing both the acromion bony structure and the humeral head and / or the greater tuberosity bony structure is extracted. The first region shoulder joint X-ray sub-image is used as part of the medical image data.

[0020] In addition, when acquiring shoulder joint ultrasound images, the ultrasound probe can be guided to perform real-time scanning at the shoulder joint of the subject based on a preset standard section scanning scheme from the acromion to the greater tubercle of the humerus. In the real-time ultrasound display image, the hyperechoic bands of the acromion cortex and the greater tubercle cortex are identified to determine the target image frame containing the bony structures of the acromion and the greater tubercle of the humerus. A first region shoulder joint ultrasound sub-image covering the bony structures of the acromion and the humeral head and / or the greater tubercle of the humerus is then extracted from the target image frame. The first region shoulder joint ultrasound sub-image serves as another part of the medical image data.

[0021] S120. Perform motion artifact detection processing on the medical image data of the first region to obtain motion artifact level information of the medical image data.

[0022] In this step, motion artifact detection processing can be performed on the medical image data of the first region to obtain motion artifact level information of the medical image data. The motion artifact level information is used to indicate the intensity of motion artifacts in the medical image data.

[0023] In the first possible implementation, the medical image data includes at least an X-ray image of the shoulder joint. Correspondingly, Figure 2 This application illustrates a flowchart of a specific implementation of S120 based on an example embodiment. For example... Figure 2 As shown, the above-mentioned S120 specifically includes: S210. Based on the high grayscale features of bone and connected component analysis, a first region of interest including the acromion bony structure and a second region of interest including the humeral head bony structure are determined.

[0024] In this step, based on the high grayscale features of bone and connected component analysis, a first region of interest including the acromion bony structure and a second region of interest including the humeral head bony structure are determined in the shoulder joint X-ray image.

[0025] Specifically, grayscale normalization and / or contrast enhancement processing can be performed on the shoulder joint X-ray image to obtain a preprocessed shoulder joint X-ray image for bone extraction. Then, in the preprocessed shoulder joint X-ray image, binarization processing is performed based on a preset bone grayscale threshold and / or an adaptive grayscale threshold to obtain a candidate binary image of bone.

[0026] Then, connected component labeling is performed on the binary bone candidate images to obtain multiple bone candidate connected components, each corresponding to a candidate bony structure region in the shoulder joint X-ray image. For each bone candidate connected component, at least one of the following region feature parameters is calculated: area, major axis direction, minor axis direction, aspect ratio, gray mean, and / or gray variance.

[0027] Next, based on the matching results between the regional feature parameters and the preset anatomical location range, morphological feature range and / or size feature range of the acromial bony structure, in the image area above the acromioclavicular joint and the lateral edge of the acromion, a target connected region that meets the constraint conditions of the acromial bony structure features is selected from multiple bone candidate connected regions. Then, a first region of interest covering the acromial bony structure is generated based on the bounding rectangle of the target connected region according to the preset extended pixel boundary.

[0028] Finally, based on the matching results between the regional feature parameters and the preset anatomical location range, morphological feature range, and / or size feature range of the humeral head bony structure, in the image region near the proximal end of the humerus and located below or medial to the acromion bony structure, a target connected region that meets the constraint conditions of the humeral head bony structure features is selected from multiple bone candidate connected regions. Based on the outer circular region and / or outer rectangular region of the target connected region, a second region of interest covering the humeral head bony structure is generated according to the preset extended pixel boundary.

[0029] S220. Perform edge detection processing on the first region of interest and the second region of interest respectively to obtain the acromion edge pixel set and the humeral head edge pixel set.

[0030] Specifically, within the first and / or second region of interest, local grayscale profiles are extracted along the predicted bony edge normal direction. Gradient multi-peak detection is then performed on these local grayscale profiles to identify bilinear candidate regions with multiple significant gradient peaks within a preset neighborhood. Next, based on the ratio of the total length of the bilinear edges in the bilinear candidate regions to the total length of the corresponding bony edges, a second motion artifact index for the shoulder joint X-ray image is determined. Finally, when the first and second motion artifact indices satisfy a preset combination judgment condition, the shoulder joint X-ray image is determined to be a high-level motion artifact image.

[0031] It is worth noting that during anteroposterior or oblique views of the shoulder joint, multiple three-dimensional bony structures, such as the costal arch, scapular spine, and lateral acromion margin, spatially overlap along the X-ray beam direction. Their projected contours on the imaging plane may be approximately parallel to the lower edge of the acromion or the upper edge of the humeral head, and their local positions may be close. This causes the grayscale distribution on a single scan line to successively cross different cortical and soft tissues, forming multiple obvious gradient peaks. True motion artifacts originate from the displacement of the same bony structure within the exposure time. The contours of the same bone block at two adjacent moments form nearly translated paired edges on the two-dimensional image. The corresponding one-dimensional grayscale profile also shows bimodal or even multimodal patterns. When detection is based solely on the gradient amplitude and peak number of a single grayscale profile, these two completely different types of double lines have a high degree of similarity at the local one-dimensional signal level. They cannot be distinguished by simple threshold or peak number discrimination, thus mistaking a large number of double lines caused by anatomical overlap for motion double images, directly interfering with the accuracy of motion artifact assessment.

[0032] To address this, after performing gradient multi-peak detection on the local grayscale profile, further binarization and connected component analysis can be performed on the first and second regions of interest in the shoulder joint X-ray image based on the high grayscale features of bone. Connected component labeling can be performed on the binary bone regions to obtain multiple bone connected components. Within the double-line candidate region, the edge pixels on each double-line edge are associated with the corresponding bone connected component to determine the bone connected component labels to which the double-line edge pixels constituting the double-line candidate region belong.

[0033] Specifically, when the bone connectivity domain labels corresponding to the edge pixels of the double-line candidate region are the same, and the edge direction difference and / or curvature difference in the local neighborhood are less than a preset threshold, the double-line candidate region is determined as a motion double image region caused by the displacement of the same bone structure during exposure.

[0034] When the bone connectivity labels corresponding to the edge pixels of the double-line candidate region are different, and / or when there is an edge direction difference and / or curvature difference greater than a preset threshold in the local neighborhood, the double-line candidate region is determined as an anatomically overlapping double-line region caused by the anatomical overlap of different bone structures in the projection direction.

[0035] Furthermore, when determining the second motion artifact index of shoulder joint X-ray images based on the ratio of the total length of the double-line edges to the total length of the corresponding bony edges in the double-line candidate region, only the length of the double-line edges that are identified as motion double-image regions is included in the total length of the double-line edges, while the length of the double-line edges that are identified as anatomically overlapping double-line regions is removed from the total length of the double-line edges.

[0036] Optionally, the step of comparing the edge direction difference and / or curvature difference of the double-line edge pixels constituting the same double-line candidate region with a preset threshold in the local neighborhood is as follows: Using each double-line edge pixel within the double-line candidate region as the center, least-squares linear fitting and / or curve fitting are performed on the local bony edges within the preset pixel neighborhood to obtain the corresponding local edge tangential direction vector and curvature parameters. For pairs of double-line edge pixels belonging to the same double-line candidate region, the edge direction angle is calculated based on the corresponding local edge tangential direction vector, and the curvature difference is calculated based on the corresponding curvature parameters. The edge direction angle is then compared with a preset direction difference threshold, and the curvature difference is compared with a preset curvature difference threshold to determine the geometric consistency of the local bony edges corresponding to the pairs of double-line edge pixels. Where the edge direction angle is less than the preset direction difference threshold and the curvature difference is less than the preset curvature difference threshold, the geometric features of the corresponding double-line candidate region are used as one of the criteria for determining a motion double image region generated by the same bony structure.

[0037] It is worth understanding that, firstly, by utilizing the high grayscale features of the bone region relative to soft tissue in the shoulder joint X-ray image, threshold segmentation and connected component analysis are performed on the first and second regions of interest. Different bony structures such as ribs, scapula, acromion, and humeral head are divided into several binary bone connected components, and each connected component is assigned a unique label.

[0038] Based on this, each edge pixel of the double-line candidate region is mapped to the corresponding bone connected component label, thereby distinguishing the two contour projections of the same bone block and the two contour projections of different bone blocks in two-dimensional space. Then, by performing least-squares straight line and / or curve fitting on the edge pixel set in the local neighborhood, the local edge tangential direction vector and curvature parameters are extracted. Threshold comparison is then performed on the edge direction angle and curvature difference of paired edges in the same double-line candidate region. If the corresponding connected component labels are the same and the difference in edge direction and curvature is less than the preset threshold, it can be regarded as an approximate translational double image caused by a small displacement of the same bony structure during exposure. If the corresponding connected component labels are different, and / or the difference in edge direction or curvature is greater than a preset threshold, it can be regarded as an anatomically overlapping double line caused by the spatial overlap of different bony structures in the projection direction.

[0039] This reduces the interference of double lines on motion artifact indices caused by anatomical overlap, avoids misjudging normal anatomical overlap as motion artifacts, and obtains a second motion artifact index that more accurately reflects the true degree of bone movement. Thus, without sacrificing usable images, it improves the accuracy of shoulder joint X-ray image quality assessment and the reliability of subsequent first distance parameter measurement.

[0040] S230. Based on the edge diffusion width features of the acromion edge pixel set and the humeral head edge pixel set, the first motion artifact index of the shoulder joint X-ray image is determined.

[0041] Specifically, the first grayscale profile is extracted along the edge normal direction of each edge pixel in the acromion edge pixel set, and the second grayscale profile is extracted along the edge normal direction of each edge pixel in the humeral head edge pixel set.

[0042] Then, the amplitude of the first gray-level derivative is calculated for each first gray-level profile and each second gray-level profile, and the corresponding local edge diffusion width is determined based on the amplitude of the first gray-level derivative. Statistical processing is then performed on all local edge diffusion widths to obtain the average diffusion width of the acromion edge and the average diffusion width of the humeral head edge.

[0043] Finally, the first motion artifact index is determined based on the comparison between the average diffusion width of the acromion edge and / or the average diffusion width of the humeral head edge and the average diffusion width and deviation range of the preset baseline.

[0044] S240. Determine the first motion artifact level of the shoulder joint X-ray image based on the first motion artifact index.

[0045] In this step, the first motion artifact level of the shoulder joint X-ray image can be determined based on the relationship between the first motion artifact index and the preset reference statistical parameters.

[0046] Specifically, it can be based on multiple shoulder joint X-ray training image samples that meet the preset acquisition specifications and have negligible motion artifacts. The average diffusion width of the acromion edge and the average diffusion width of the humeral head edge are statistically analyzed to obtain the corresponding first baseline mean, first baseline standard deviation, second baseline mean, and second baseline standard deviation. The first baseline mean and first baseline standard deviation are used as reference statistical parameters to characterize the diffusion width of the acromion edge, and the second baseline mean and second baseline standard deviation are used as reference statistical parameters to characterize the diffusion width of the humeral head edge.

[0047] Then, after calculating the average diffusion width of the acromion edge and / or the average diffusion width of the humeral head edge from the X-ray image of the shoulder joint to be tested, the corresponding first standardized deviation value and / or second standardized deviation value are determined based on the degree of deviation between the average diffusion width of the acromion edge and the first baseline mean and the first baseline standard deviation, and / or based on the degree of deviation between the average diffusion width of the humeral head edge and the second baseline mean and the second baseline standard deviation.

[0048] Next, based on the first standardized deviation value and / or the second standardized deviation value, a comprehensive first motion artifact index score is constructed to characterize the overall edge diffusion degree of the shoulder joint X-ray image. The comprehensive first motion artifact index score is compared with the preset mild motion artifact threshold and severe motion artifact threshold.

[0049] When the overall first motion artifact index score is less than the mild motion artifact threshold, the first motion artifact level of the shoulder joint X-ray image is determined as no motion artifact or negligible motion artifact level.

[0050] When the score of the comprehensive first motion artifact index is within a preset range between the mild motion artifact threshold and the severe motion artifact threshold, the first motion artifact level of the shoulder joint X-ray image is determined to be the mild motion artifact level.

[0051] When the score of the comprehensive first motion artifact index is greater than or equal to the severe motion artifact threshold, the first motion artifact level of the shoulder joint X-ray image is determined as the severe motion artifact level.

[0052] In the second possible implementation, the medical image data includes at least shoulder joint ultrasound images. Correspondingly, such as Figure 3 As shown, Figure 3 This application illustrates a flowchart of a specific implementation of S120 according to another exemplary embodiment. For example... Figure 3 As shown, the above-mentioned S120 specifically includes: S310. Acquire multiple frames of shoulder joint ultrasound images within a preset time window.

[0053] In this step, after obtaining the shoulder joint ultrasound image sequence, multiple frames of shoulder joint ultrasound images are acquired within a preset time window, and the multiple frames of shoulder joint ultrasound images correspond to the same standard section from the acromion to the greater tubercle of the humerus.

[0054] Specifically, the operator can place the ultrasound probe in the lateral region of the acromion of the subject according to the shoulder joint ultrasound examination specifications, aligning the long axis of the probe roughly along the coronal plane, and observe the real-time ultrasound image to simultaneously bring the echo structure of the acromion cortex and the echo structure of the greater tubercle of the humerus into the same scanning screen to form a standard section from the acromion to the greater tubercle of the humerus.

[0055] Secondly, when the standard section from the acromion to the greater tubercle of the humerus is stably displayed, the length of the preset time window for image acquisition is determined based on the preset acquisition frame rate and / or acquisition duration, and ultrasound image acquisition commands are continuously triggered at fixed or approximately fixed time intervals within the preset time window.

[0056] Furthermore, within the preset time window, whenever an ultrasound image acquisition command is triggered, the current frame of shoulder joint ultrasound image is extracted from the real-time ultrasound playback stream as a candidate frame, and each candidate frame is cached and stored in the order of acquisition time to form the initial sequence of shoulder joint ultrasound images.

[0057] Then, during the image acquisition process, by limiting the amplitude of probe posture changes and / or monitoring the positional relationship between the acromial cortex and the greater tubercle cortex in the real-time image, when it is detected that both the acromial cortex and the greater tubercle cortex are within the preset spatial position tolerance range, the candidate frames cached in the corresponding time period are marked as the target frame set that meets the standard section condition from the same acromion to the greater tubercle.

[0058] Finally, multiple shoulder ultrasound images arranged chronologically in the target frame set are used as a shoulder ultrasound image sequence corresponding to the same standard section from the acromion to the greater tuberosity of the humerus.

[0059] S320. In each frame of shoulder ultrasound image, identify a third region of interest including the acromion and the greater tubercle of the humerus.

[0060] Specifically, based on the imaging parameters and probe type of the ultrasound equipment, the image resolution information and physical spatial scale information of the corresponding shoulder joint ultrasound image can be obtained. Based on the typical display range of the acromion cortex and the greater tubercle cortex of the humerus under the standard section from the acromion to the greater tubercle of the humerus, the approximate spatial range parameters used to define the third region of interest can be preset.

[0061] In shoulder ultrasound images, candidate band regions containing the main distribution of high echogenicity of the cortical bone are determined by using grayscale projection distribution along the horizontal and / or vertical directions. These candidate band regions are used to include the echogenic structures of the acromion cortex and the greater tubercle cortex of the humerus.

[0062] Within the candidate strip region, continuous hyperechoic connected regions are extracted based on grayscale threshold segmentation and connected component analysis. Then, based on morphological features, area range, and long axis direction, the first bony candidate region that meets the morphological features of the acromion cortex and the second bony candidate region that meets the morphological features of the greater tubercle cortex of the humerus are selected.

[0063] When both the first and second bony candidate regions exist, the bounding rectangle boundaries of the first and second bony candidate regions are calculated, and the bounding rectangle boundaries are expanded around the bounding rectangle boundaries according to a preset pixel expansion boundary to generate a third region of interest covering the acromion and the greater tubercle of the humerus.

[0064] Finally, when only a unilateral bony candidate region is detected or the detection results show unstable fluctuations, the third region of interest in the current frame is predicted and corrected based on the location of the third region of interest already determined in the previous frame or several previous frames of shoulder joint ultrasound images by limiting the maximum displacement threshold and the centroid drift range, so as to ensure the spatial continuity and stability of the third region of interest in the time series.

[0065] S330. Based on the third region of interest, perform two-dimensional rigid body registration on two adjacent frames of shoulder joint ultrasound images to obtain the inter-frame translation vector sequence.

[0066] First, in the shoulder joint ultrasound image sequence, any two adjacent shoulder joint ultrasound images in chronological order are used as the reference frame and the frame to be registered, respectively. The corresponding third region of interest is extracted from the reference frame and the frame to be registered. The third region of interest is used as the reference image block and the image block to be registered, respectively.

[0067] Secondly, grayscale normalization and / or bandpass filtering are performed on the reference image patch of interest and the image patch of interest to be registered to reduce the influence of overall brightness variation and random noise on rigid body registration, and to preserve the main hyperechoic structures of the acromion cortex and the greater tubercle cortex of the humerus.

[0068] Furthermore, under the premise that the translation satisfies the two-dimensional rigid body transformation constraint, the horizontal and vertical translation amounts of the image block to be registered relative to the reference image block are used as the translation parameters to be determined. Within the preset translation search range, similarity calculations are performed on the registration results under different combinations of translation parameters using similarity measurement functions based on normalized cross-correlation coefficients and / or correlation coefficients.

[0069] Then, under each set of candidate translation parameters, the image block of interest to be registered is transformed into a pixel-level translation according to the corresponding candidate translation parameters, aligned with the reference image block of interest, the value of the corresponding similarity metric function is calculated, and the target translation parameter combination that makes the value of the similarity metric function reach the maximum value is selected from all candidate translation parameter combinations as the optimal rigid body registration translation parameter between the two adjacent frames of shoulder joint ultrasound images.

[0070] After obtaining the optimal rigid body registration translation parameters, the corresponding horizontal and vertical translation components are combined to form the translation vector of the current frame to be registered relative to the reference frame. The translation vectors of all adjacent frame pairs are calculated and stored sequentially according to the time order of the shoulder joint ultrasound image sequence to form the translation vector sequence between frames.

[0071] Finally, when the maximum value of the similarity metric function is lower than the preset reliability threshold, the translation vector of the current frame is interpolated and thresholded based on the translation vectors already determined at adjacent time points and / or on the time smoothing constraints of translation vectors in multiple frames, so as to improve the continuity of the translation vector sequence in the time dimension.

[0072] S340. Determine the cumulative amount of exercise within each time window, and select a stationary window from each time window where the cumulative amount of exercise meets the preset threshold condition.

[0073] In this step, the translation vector sequence is statistically analyzed based on the sliding time window to determine the cumulative motion within each time window, and a stationary window is selected from each time window whose cumulative motion meets the preset threshold condition.

[0074] Specifically, in the translation vector sequence obtained between each frame of shoulder joint ultrasound images arranged in chronological order, the translation vector between two adjacent frames of shoulder joint ultrasound images is used as the basic motion vector of the corresponding time interval, and a translation vector time sequence corresponding to the inter-frame time sequence is constructed based on the frame order index of the translation vector sequence.

[0075] On the translation vector time series, a sliding time window is set according to the preset window length and window step size. The sliding time window covers several consecutive translation vectors, and each sliding time window corresponds to a time window index.

[0076] Then, for each sliding time window, based on the horizontal and vertical components of each translation vector contained in the time window, the magnitude of each translation vector is calculated, and the magnitudes are accumulated within the time window and / or statistically analyzed according to the maximum value and average value to obtain the first cumulative exercise index and / or the second cumulative exercise index corresponding to the time window.

[0077] Next, the first cumulative exercise index and / or the second cumulative exercise index corresponding to each time window are compared with the preset rest threshold range. When the first cumulative exercise index and / or the second cumulative exercise index of a certain time window are less than or equal to the upper limit of rest threshold and preferably greater than or equal to the lower limit of rest threshold, the time window is determined as a candidate rest window.

[0078] Furthermore, in the presence of multiple candidate static windows, based on the magnitude relationship between the first cumulative motion index and / or the second cumulative motion index corresponding to the candidate static windows, one or more candidate static windows with the smallest cumulative motion are selected as the final set of static windows, and the corresponding time window index and time range are output to indicate the time interval within which the shoulder joint ultrasound image sequence meets the motion artifact judgment condition within the static window.

[0079] S350. If there is a stationary window and the cumulative motion volume corresponding to the stationary window is lower than a preset threshold, it is determined that the shoulder joint ultrasound image sequence meets the motion artifact judgment condition within the stationary window.

[0080] Specifically, after statistically analyzing the translation vector sequence based on the sliding time window and obtaining the cumulative motion amount corresponding to each time window, the cumulative motion amount of each time window is compared with the preset first motion threshold and / or second motion threshold. When the cumulative motion amount of a certain time window is less than or equal to the first motion threshold and / or within the preset static threshold range, the time window is marked as a static window, and the corresponding time window index and the frame number range covered by the time window are recorded.

[0081] If at least one stationary window exists, the cumulative motion volume corresponding to each stationary window is further compared and / or sorted to determine the stationary window with the lowest cumulative motion volume or several stationary windows with cumulative motion volume below a preset preferred threshold, which are then used as the target stationary window set.

[0082] Furthermore, for each target still window in the target still window set, based on the frame number range covered by the target still window, multiple corresponding shoulder ultrasound images are selected from the shoulder ultrasound image sequence as a candidate image set that meets the motion artifact judgment criteria. Optionally, sharpness evaluation and / or artifact residual evaluation are performed on the candidate image set to remove frame images with significant artifacts.

[0083] When the cumulative motion amount corresponding to at least one target static window in the target static window set is lower than a preset threshold and the corresponding candidate image set passes the sharpness evaluation and / or artifact residual amount evaluation, the time interval corresponding to the target static window is determined as the time interval in which the shoulder joint ultrasound image sequence meets the motion artifact judgment condition within the static window, and the corresponding multiple frames of shoulder joint ultrasound images are determined as valid ultrasound image data for subsequent shoulder joint anatomical landmark detection and second distance parameter ranging processing.

[0084] It is worth noting that in X-ray imaging, slight translation or rotation of the patient during exposure creates an integral projection of the bony contour onto the detector over the exposure time. This causes an ideal, single, sharp edge to appear as an increased edge spread or even a double contour in the image, resulting in a systematic shift in the edge detection results of key bony structures such as the acromion and humeral head. In contrast, ultrasound imaging is a real-time sequential imaging process. Slight probe jitter, operator hand stability, and patient movements can cause rigid body translation and local deformation between frames, resulting in continuous changes in the spatial relationship from the acromion to the greater tuberosity of the humerus in adjacent frames. If the translation vector sequence is not quantitatively analyzed and a low-motion static window is not found based on a sliding time window, distance is often measured directly on frames where significant inter-frame displacement still exists. This causes the statistical results of the second distance parameter to be dominated by motion error rather than the actual anatomical distance difference.

[0085] To address this, the motion artifact level information can be further configured to include a first-level motion artifact level from shoulder joint X-ray images and / or a second-level motion artifact level from shoulder joint ultrasound image sequences.

[0086] When the first-level motion artifact level indicates that the edge diffusion width and / or double image ratio of the shoulder joint X-ray image exceeds the first preset threshold, the shoulder joint X-ray image is determined as medical image data that does not meet the preset quality conditions, and the distance measurement processing of the first distance parameter is not performed on the shoulder joint X-ray image.

[0087] Specifically, within the first region of interest (ROI) of the shoulder joint X-ray image, grayscale profiles are extracted along the normal direction of the acromial bony edge and / or the humeral head bony edge, and the corresponding grayscale first derivative curves are calculated based on the grayscale profiles. The local edge diffusion widths corresponding to the acromial bony edge and / or the humeral head bony edge are determined according to the half-peak width of the derivative peak in the grayscale first derivative curve, and statistical calculations are performed on multiple local edge diffusion widths to obtain the global edge diffusion width index of the shoulder joint X-ray image. Within the first ROI of the shoulder joint X-ray image, multiple grayscale gradient multi-peak structures are detected along the normal direction of the acromial bony contour and / or the humeral head bony contour. The double image ratio index of the shoulder joint X-ray image is determined based on the ratio of the total length of edge pixels exhibiting double-line contours within a preset neighborhood to the total length of the bony contours. Then, the global edge diffusion width index and / or double image ratio index are compared with a first preset threshold. When the global edge diffusion width index is greater than the first preset threshold and / or the double image ratio index is greater than the first preset threshold, a first quality control signal is generated indicating that the first-level motion artifact level corresponding to the current shoulder joint X-ray image exceeds the upper quality limit. Finally, based on the first quality control signal, the current shoulder joint X-ray image is marked as medical image data that does not meet the preset quality conditions, and in subsequent processing, the call to the first distance parameter ranging module corresponding to the shoulder joint X-ray image is blocked, and no ranging calculations related to the first distance parameter are performed on the shoulder joint X-ray image.

[0088] When the first level of motion artifact level indicates that the edge diffusion width of the shoulder joint X-ray image is within the acceptable range of mild motion artifacts, a ranging strategy that includes fitting and outlier removal is used to calculate the first distance parameter.

[0089] Specifically, within the first region of interest (ROI) of the shoulder joint X-ray image, multiple local gray-level profiles are extracted along the normal direction of the acromion bony edge. Based on the amplitude distribution of the first derivative of gray-level values ​​for each local gray-level profile, the corresponding set of candidate edge pixels for the acromion is determined. Within the second ROI of the shoulder joint X-ray image, multiple local gray-level profiles are extracted along the normal direction of the humeral head bony edge. Based on the amplitude distribution of the first derivative of gray-level values ​​for each local gray-level profile, the corresponding set of candidate edge pixels for the humeral head is determined. Then, a linear fitting based on the least squares criterion is performed on the candidate edge pixel set for the acromion to obtain an initial fitted linear model of the acromion; and an arc fitting and / or quadratic curve fitting based on the least squares criterion is performed on the candidate edge pixel set for the humeral head to obtain an initial fitted contour model of the humeral head. Then, the vertical residual of each candidate acromion edge pixel relative to the initial fitted straight line model of the acromion, and the radial residual of each candidate humeral head edge pixel relative to the initial fitted contour model of the humeral head are calculated. Candidate edge pixels with an absolute residual value greater than the corresponding preset residual threshold are marked as outliers, and candidate edge pixels with an absolute residual value less than or equal to the preset residual threshold are marked as inliers. After removing outliers, the set of candidate acromion edge pixels is refitted with least squares straight lines based on the remaining inliers to obtain the target fitted straight line of the acromion; and the set of candidate humeral head edge pixels is refitted with circular arc fitting and / or quadratic curve fitting based on the remaining inliers to obtain the target fitted contour of the humeral head. Finally, the minimum Euclidean distance between the target fitted straight line of the acromion and the target fitted contour of the humeral head is calculated along the preset ranging direction, and the minimum Euclidean distance is determined as the first distance parameter corresponding to the shoulder joint X-ray image. This is to reduce the interference of local edge blurring and artifacts on the ranging results in the presence of mild motion artifacts by using fitting and outlier removal strategies.

[0090] When the second-level motion artifact level indicates that there is no stationary window that meets the motion threshold and sharpness threshold within the preset time window, the corresponding shoulder joint ultrasound image sequence is determined as medical image data that does not meet the preset quality conditions, and the second distance parameter ranging processing is not performed on the shoulder joint ultrasound image sequence.

[0091] Specifically, based on the statistical results of the sliding time window of the translation vector sequence, the cumulative motion amount is calculated for each time window. Time windows with cumulative motion amounts greater than a preset motion threshold are marked as motion windows, and time windows with cumulative motion amounts less than or equal to the preset motion threshold are marked as candidate stationary windows. Then, for multiple frames of shoulder ultrasound images within each candidate stationary window, a corresponding sharpness evaluation index is calculated based on sharpness features such as intra-frame image noise level, cortical bone boundary gradient intensity, and / or local contrast. Candidate stationary windows with sharpness evaluation indices lower than a preset sharpness threshold are removed. Finally, after performing sharpness threshold filtering on all candidate stationary windows, if no stationary window simultaneously meets both the preset motion threshold and the sharpness threshold, the shoulder ultrasound image sequence is marked as a sequence type that does not meet the preset quality conditions. Finally, when a shoulder ultrasound image sequence is marked as a sequence type that does not meet the preset quality conditions, the corresponding second distance parameter ranging module is prohibited from being called. The detection and ranging calculation of the acromion cortex position and the greater tubercle cortex position of the humerus are not performed in the shoulder ultrasound image sequence. The quality failure indication information associated with the shoulder ultrasound image sequence is output to prompt the re-acquisition of shoulder ultrasound image data.

[0092] S130. Perform shoulder joint anatomical landmark detection and distance measurement on medical image data.

[0093] In this step, if the motion artifact level information indicates that the medical image data meets the preset quality conditions, shoulder joint anatomical landmark detection and distance measurement are performed on the medical image data to obtain the first distance parameter between the acromion and the humeral head and / or the second distance parameter between the acromion and the greater tubercle of the humerus.

[0094] In one possible implementation, the medical image data includes X-ray images and ultrasound images of the shoulder joint. Correspondingly, such as... Figure 4 As shown, Figure 4 This application illustrates a flowchart of a specific implementation of S130 according to another exemplary embodiment. For example... Figure 4 As shown, the above-mentioned S130 specifically includes: S410. In the X-ray image of the shoulder joint, perform least-squares fitting on the set of edge pixels of the bony edge region of the acromion to obtain the fitted straight line of the lower edge of the acromion.

[0095] First, in the shoulder joint X-ray image, based on a pre-determined first region of interest including the acromial bony structure, an edge pixel set corresponding to the acromial bony edge region is extracted from the first region of interest, and each edge pixel in the edge pixel set has a corresponding image coordinate value.

[0096] Secondly, the image coordinate values ​​of each edge pixel in the edge pixel set are represented as a two-dimensional coordinate point set, and a linear least squares fitting model is constructed based on the two-dimensional coordinate point set.

[0097] Next, in the linear least squares fitting model, with the horizontal pixel coordinates of the image as the independent variable and the vertical pixel coordinates of the image as the dependent variable, the sum of squares of the vertical distances from all edge pixels to the candidate line is minimized to solve for the slope and intercept parameters of the fitted line at the lower edge of the shoulder peak.

[0098] After obtaining the slope and intercept parameters, the straight line characterized by the slope and intercept parameters is used as the fitting straight line of the lower edge of the acromion corresponding to the bony edge region of the acromion, and is used for the subsequent calculation of the first distance parameter between the acromion and the humeral head.

[0099] S420: Perform curve or line fitting on the pixel set of the upper edge of the humeral head to obtain the fitted contour of the upper edge of the humeral head.

[0100] First, in the shoulder joint X-ray image, based on the second region of interest including the bony structure of the humeral head, edge detection processing is performed on the second region of interest to extract the set of edge pixels of the upper edge of the humeral head. Each edge pixel in the set of edge pixels of the upper edge of the humeral head has a corresponding image coordinate value.

[0101] Secondly, the image coordinate values ​​of each edge pixel in the set of edge pixels of the upper edge of the humeral head are represented as a two-dimensional coordinate point set, and at least one of the curve fitting model and the straight line fitting model is constructed based on the two-dimensional coordinate point set.

[0102] In the straight line fitting model, the horizontal pixel coordinates of the image are used as independent variables and the vertical pixel coordinates of the image are used as dependent variables. The slope and intercept parameters of the straight line fitting profile of the upper edge of the humeral head are solved by minimizing the sum of squares of the vertical distances from all edge pixels to the candidate straight line.

[0103] In the curve fitting model, the two-dimensional coordinate point set is substituted into a pre-defined polynomial curve model and / or circular arc model. The sum of squared fitting residuals is minimized by the least squares method to solve for the polynomial coefficient parameters and / or center and radius parameters of the curve fitting profile of the upper edge of the humeral head.

[0104] After obtaining the straight-line fitted profile and the curve fitted profile, the residual statistics of the pixel set of the upper edge of the humeral head to the straight-line fitted profile and the curve fitted profile are calculated respectively. Based on the comparison of the root mean square of the residuals and / or the median of the residuals, the one with smaller residuals is selected from the straight-line fitted profile and the curve fitted profile as the upper edge fitted profile of the humeral head, which is used for the subsequent distance calculation of the first distance parameter between the acromion and the humeral head.

[0105] S430. Determine the first distance parameter between the acromion and the humeral head based on the minimum distance between the fitted straight line at the lower edge of the acromion and the fitted contour at the upper edge of the humeral head.

[0106] First, after obtaining the linear parameters of the fitted line at the lower edge of the acromion and the contour parameters of the fitted contour at the upper edge of the humeral head, based on the image coordinate values ​​of each edge pixel in the set of edge pixels at the upper edge of the humeral head, several sampling points are selected from the fitted contour at the upper edge of the humeral head according to a preset sampling step size. The image coordinate values ​​of these sampling points constitute the set of sampling points at the upper edge of the humeral head.

[0107] Secondly, for each sampling point in the set of sampling points at the upper edge of the humeral head, the vertical distance from the sampling point to the fitted line at the lower edge of the acromion is calculated based on the linear parameters of the fitted line at the lower edge of the acromion, and the corresponding distance value sequence is obtained.

[0108] Then, statistical processing is performed on the distance value sequence to obtain the minimum distance value in the distance value sequence and its corresponding sampling point position. The minimum distance value is determined as the first candidate distance parameter between the acromion and the humeral head.

[0109] Furthermore, when multiple fitting contour models exist, the corresponding first candidate distance parameters are calculated based on the straight-line fitting contour and the curve fitting contour, respectively. The first candidate distance parameter of the fitting contour model with smaller residual is selected according to the corresponding residual statistics and output as the first distance parameter, which is used to indicate the minimum interosseous distance between the acromion and the humeral head.

[0110] S440. In the shoulder joint ultrasound image, along the preset measurement line and its adjacent measurement lines, determine the position of the acromion cortex and the position of the greater tubercle cortex of the humerus, respectively, and obtain the second distance parameter between the acromion and the greater tubercle of the humerus through statistical fusion based on the distance results of multiple measurement lines.

[0111] Specifically, in the selected target shoulder joint ultrasound image, a central measurement line connecting the acromial cortex region and the greater tuberosity cortex region of the humerus is determined. Within a preset pixel offset range to the left and right of the central measurement line, several parallel neighboring measurement lines are generated. On each measurement line, the positions of the acromial cortex and the greater tuberosity cortex of the humerus are detected along the depth direction based on grayscale and gradient changes, and the corresponding measurement distances are calculated. Then, median processing and / or extreme value removal averaging are performed on the measurement distances corresponding to all measurement lines to obtain the intra-frame second distance parameter of the shoulder joint ultrasound image. Furthermore, in the case of multiple target shoulder joint ultrasound images, median fusion is performed on the intra-frame second distance parameters of each frame to obtain the second distance parameter.

[0112] S140. Determine the assessment result of shoulder subluxation of the subject based on the first distance parameter and / or the second distance parameter.

[0113] Specifically, a preset reference threshold range for a first distance parameter between the acromion and the humeral head can be obtained, including an upper limit threshold of the normal range and / or a lower limit threshold of the abnormal range. Furthermore, a preset reference threshold range for a second distance parameter between the acromion and the greater tubercle of the humerus can be obtained, including an upper limit threshold of the normal range and / or a lower limit threshold of the abnormal range.

[0114] Given the first distance parameter, the first distance parameter is compared with the corresponding first distance parameter reference threshold range. When the first distance parameter is less than or equal to the upper limit threshold of the normal range and greater than or equal to the lower limit threshold of the abnormal range, the shoulder subluxation risk level corresponding to the first distance parameter is determined as the first candidate assessment level.

[0115] Once the second distance parameter is obtained, it is compared with the corresponding reference threshold range for the second distance parameter. When the second distance parameter is less than or equal to the upper limit threshold of the normal range and greater than or equal to the lower limit threshold of the abnormal range, the shoulder subluxation risk level corresponding to the second distance parameter is determined as the second candidate assessment level.

[0116] When the first distance parameter and the second distance parameter are obtained simultaneously, the first candidate assessment level and the second candidate assessment level are fused based on the preset decision fusion rule to determine the target shoulder subluxation assessment result. The decision fusion rule includes at least one of the following: taking the higher risk level as the final shoulder subluxation assessment result, or making a comprehensive judgment on the first candidate assessment level and the second candidate assessment level based on weighted voting.

[0117] If only the first distance parameter or only the second distance parameter is obtained, the corresponding first candidate assessment level or second candidate assessment level is directly used as the shoulder subluxation assessment result, and the shoulder subluxation assessment result is associated with and stored with the corresponding distance parameter value and its usage.

[0118] In this embodiment, medical image data of a first region of the shoulder joint of the subject is acquired, and motion artifact detection processing is performed on the medical image data of the first region to obtain motion artifact level information of the medical image data. When the motion artifact level information indicates that the medical image data meets preset quality conditions, shoulder joint anatomical landmark detection and distance measurement processing are performed on the medical image data to obtain a first distance parameter between the acromion and the humeral head and / or a second distance parameter between the acromion and the greater tubercle of the humerus. Finally, the shoulder joint subluxation assessment result of the subject is determined based on the first distance parameter and / or the second distance parameter, thereby ensuring that the acromion... Humeral head distance and acromion The greater tuberosity distance of the humerus can still be accurately and reproducibly measured in stroke patients, a high-risk population for movement, thus solving the technical problems of unstable distance measurement and unreliable assessment caused by motion artifacts in the prior art.

[0119] In other words, in the above embodiment, a medical image of the first region of the patient's shoulder joint is first acquired. This first region includes at least the acromion, the humeral head, and / or the greater tubercle of the humerus. The purpose is to ensure that the distances from the acromion to the humeral head and from the acromion to the greater tubercle of the humerus can be directly measured within the same image.

[0120] When the patient or the probe moves, edge blurring or ghosting can occur in the acquired medical images, resulting in double lines or ghosting of bone contours. To address this issue, the above embodiment calculates the level of blurring using indicators such as the degree of edge blurring, the proportion of double contours, and inter-frame displacement.

[0121] Next, if the quality level indicates that motion artifacts are too severe and do not meet the preset quality standard, such images will not be further measured; they are automatically discarded. This avoids obtaining false data from a severely blurry or ghosted image. Only if the quality level indicates that the image quality is within an acceptable range will the next step be taken to automatically identify bony structures and perform distance measurement.

[0122] Figure 5 This is a schematic diagram of the structure of an electronic device according to an example embodiment of this application. For example... Figure 5 As shown, the electronic device 500 provided in this embodiment includes: a processor 501 and a memory 502; wherein: Memory 502 is used to store computer programs, and the memory may also be flash memory.

[0123] Processor 501 is used to execute the execution instructions stored in the memory to implement the various steps in the above method. For details, please refer to the relevant descriptions in the preceding method embodiments.

[0124] Alternatively, the memory 502 can be either standalone or integrated with the processor 501.

[0125] When the memory 502 is a device independent of the processor 501, the electronic device 500 may further include: Bus 503 is used to connect the memory 502 and the processor 501.

[0126] This embodiment also provides a readable storage medium storing a computer program, which, when executed by at least one processor of an electronic device, enables the electronic device to perform the methods provided in the various embodiments described above.

[0127] This embodiment also provides a program product including a computer program stored in a readable storage medium. At least one processor of an electronic device can read the computer program from the readable storage medium, and the at least one processor executes the computer program to cause the electronic device to perform the methods provided in the various embodiments described above.

[0128] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the claims.

[0129] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for processing image data of shoulder subluxation, characterized in that, include: Acquire medical image data of a first region of the shoulder joint of the subject under test, the first region including the acromion bony structure and the humeral head and / or the greater tubercle bony structure; Motion artifact detection processing is performed on the medical image data of the first region to obtain motion artifact level information of the medical image data. The motion artifact level information is used to indicate the intensity of motion artifacts in the medical image data. When the motion artifact level information indicates that the medical image data meets the preset quality conditions, shoulder joint anatomical landmark detection and distance measurement are performed on the medical image data to obtain a first distance parameter between the acromion and the humeral head and / or a second distance parameter between the acromion and the greater tubercle of the humerus. The shoulder subluxation assessment result of the subject is determined based on the first distance parameter and / or the second distance parameter.

2. The method for processing shoulder joint subluxation image data according to claim 1, characterized in that, The medical image data includes at least a shoulder joint X-ray image; correspondingly, the motion artifact detection processing on the medical image data of the first region includes: In the X-ray image of the shoulder joint, a first region of interest including the bony structures of the acromion and a second region of interest including the bony structures of the humeral head are determined; Edge detection processing is performed on the first region of interest and the second region of interest respectively to obtain the acromion edge pixel set and the humeral head edge pixel set; Based on the edge diffusion width characteristics of the acromion edge pixel set and the humeral head edge pixel set, a first motion artifact index for the shoulder joint X-ray image is determined. The first motion artifact level of the shoulder joint X-ray image is determined based on the first motion artifact index.

3. The method for processing shoulder joint subluxation image data according to claim 2, characterized in that, The first motion artifact index for determining shoulder joint X-ray images includes: Extract the corresponding first grayscale profile along the edge normal direction of each edge pixel in the acromion edge pixel set, and extract the corresponding second grayscale profile along the edge normal direction of each edge pixel in the humeral head edge pixel set; The corresponding local edge diffusion width is determined based on each of the first grayscale profiles and each of the second grayscale profiles; Statistical processing was performed on all the local edge diffusion widths to obtain the average diffusion width of the acromion edge and the average diffusion width of the humeral head edge, respectively. The first motion artifact index is determined based on the comparison between the average diffusion width of the acromion edge and / or the average diffusion width of the humeral head edge and the average diffusion width and deviation range of the preset baseline.

4. The method for processing shoulder joint subluxation image data according to claim 2, characterized in that, The step of performing edge detection processing on the first region of interest and the second region of interest respectively to obtain the acromion edge pixel set and the humeral head edge pixel set includes: In the first region of interest and / or the second region of interest, a local grayscale profile is extracted along the estimated bony edge normal direction, and gradient multi-peak detection is performed on the local grayscale profile to identify multiple bilinear candidate regions within a preset neighborhood. The second motion artifact index of the shoulder joint X-ray image is determined based on the ratio of the total length of the double-line edge to the total length of the corresponding bony edge in the double-line candidate region. When the first motion artifact index and the second motion artifact index meet the preset combination judgment conditions, the shoulder joint X-ray image is determined to be a motion artifact image.

5. The method for processing shoulder joint subluxation image data according to claim 1, characterized in that, The medical image data includes at least shoulder joint ultrasound images; correspondingly, the motion artifact detection processing of the medical image data in the first region includes: Multiple frames of shoulder joint ultrasound images are acquired within a preset time window, and the multiple frames of shoulder joint ultrasound images correspond to the same standard section from the acromion to the greater tubercle of the humerus. In each frame of shoulder ultrasound image, a third region of interest was identified, including the acromion and the greater tubercle of the humerus. Based on the third region of interest, two-dimensional rigid body registration is performed on two adjacent frames of shoulder joint ultrasound images to obtain the inter-frame translation vector sequence. The translation vector sequence is statistically analyzed based on a sliding time window to determine the cumulative motion within each time window, and a stationary window whose cumulative motion satisfies a preset threshold condition is selected from each time window. If a static window exists and the cumulative motion corresponding to the static window is lower than the preset threshold, the shoulder joint ultrasound image sequence is determined to meet the motion artifact judgment condition within the static window.

6. The method for processing shoulder joint subluxation image data according to claim 1, characterized in that, The medical image data includes shoulder joint X-ray images and shoulder joint ultrasound images; correspondingly, shoulder joint anatomical landmark detection and distance measurement processing are performed on the medical image data, including: In shoulder joint X-ray images, least squares fitting is performed on the set of edge pixels of the acromion bony border region to obtain a fitted straight line at the lower edge of the acromion; Perform curve or line fitting on the pixel set of the upper edge of the humeral head to obtain the fitted contour of the upper edge of the humeral head. Based on the minimum distance between the fitted straight line at the lower edge of the acromion and the fitted contour at the upper edge of the humeral head, a first distance parameter between the acromion and the humeral head is determined. In shoulder ultrasound images, the positions of the acromion cortex and the greater tubercle cortex of the humerus are determined along a preset measurement line and its adjacent measurement lines, respectively. Based on the distance results of multiple measurement lines, a second distance parameter between the acromion and the greater tubercle of the humerus is obtained through statistical fusion.

7. The method for processing shoulder joint subluxation image data according to claim 6, characterized in that, The determination of the location of the acromial cortex and the greater tuberosity cortex of the humerus includes: In the selected target shoulder joint ultrasound image, determine a central measurement line connecting the acromial cortical region and the greater tuberosity cortical region of the humerus; Within a preset pixel offset range to the left and right of the central measurement line, several parallel adjacent measurement lines are generated; On each of the aforementioned measurement lines, the positions of the acromion cortex and the greater tuberosity cortex of the humerus are detected along the depth direction based on grayscale and gradient changes, and the corresponding measurement distances are calculated. Median processing and / or extreme value removal averaging are performed on the measurement distances corresponding to all measurement lines to obtain the intra-frame second distance parameter of the corresponding frame of shoulder joint ultrasound image; In the presence of multiple frames of ultrasound images of the target shoulder joint, median fusion is performed on the intra-frame second distance parameter of each frame to obtain the second distance parameter.

8. The method for processing shoulder subluxation image data according to claim 1, characterized in that, The motion artifact level information includes the first-level motion artifact level of the shoulder joint X-ray image and / or the second-level motion artifact level of the shoulder joint ultrasound image sequence; the method further includes: When the first level of motion artifact level indicates that the edge diffusion width and / or double shadow ratio of the shoulder joint X-ray image exceeds the first preset threshold, the shoulder joint X-ray image is determined as medical image data that does not meet the preset quality conditions, and the distance measurement processing of the first distance parameter is not performed on the shoulder joint X-ray image. When the first level of motion artifact level indicates that the edge diffusion width of the shoulder joint X-ray image is within the acceptable range of motion artifact intervals, the first distance parameter is calculated using a ranging strategy that includes fitting and outlier removal. When the second-level motion artifact level indicates that there is no stationary window that meets the motion threshold and sharpness threshold within the preset time window, the corresponding shoulder joint ultrasound image sequence is determined as medical image data that does not meet the preset quality conditions, and the second distance parameter ranging processing is not performed on the shoulder joint ultrasound image sequence.

9. An electronic device, characterized in that, include: processor; as well as, Memory for storing the executable instructions of the processor; The processor is configured to execute the method of any one of claims 1 to 7 by executing the executable instructions.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 7.

Images