A spinal column activity analysis method and system based on multi-sensor fusion

Abstract

The present application relates to the technical field of modeling, in particular to a kind of spinal column activity analysis method and system based on multi-sensor fusion, comprising the following steps: obtaining angular velocity acceleration and time sequence and accumulating difference, identifying same-direction change, converting angle to direction vector and extracting turning node by dot product, updating space coordinates according to change direction and combining distance calibration, dividing interval to extract displacement direction and amplitude.The present application, by constructing the posture evolution mechanism of angular velocity accumulation and time difference, combined with continuous same-direction change identification, and introducing direction vector dot product to determine motion turning node to depict fine-grained change trend, simultaneously update space coordinates according to angle change direction and fuse segment distance constraint to correct trajectory, extract displacement direction and amplitude difference at time interval level to complete distribution division, strengthen the stage characteristics of motion process, improve turning recognition sensitivity and enhance trajectory consistency, so that activity result presents spatial correlation.

Description

Technical Field

[0001] This invention relates to the field of modeling technology, and in particular to a method and system for analyzing spinal mobility based on multi-sensor fusion. Background Technology

[0002] The field of modeling technology mainly involves using mathematical expressions, geometric structures, and data mapping relationships to abstractly represent and calculate the morphology, motion state, and change patterns of objective objects. Its core aspects include constructing spatial positional relationships through coordinate systems, describing dynamic changes using time series, establishing parametric models by combining multi-source data, and achieving state assessment and prediction through numerical calculations. It is widely used in human motion analysis, engineering structure simulation, and scientific computing. Among them, the traditional spinal mobility analysis method based on multi-sensor fusion refers to collecting posture data by arranging inertial measurement units and angle sensors at different spinal segments on the human back. After synchronizing the acceleration and angular velocity values ​​output by each sensor in time, the bending angle of each spinal segment is obtained by calculating the relative rotation angle between adjacent sensors. The overall range of mobility is obtained by calculating the sum or difference of the angles of each segment. At the same time, the angle changes are spatially converted by combining the preset spinal segment length parameters to achieve a quantitative description of spinal activity.

[0003] Existing processing methods rely on direct derivation of the overall range based on angle changes. The process lacks detailed expression of motion stage division and directional continuity, resulting in a discretized dynamic process. In complex motion or rhythmic change scenarios, it is difficult to accurately reflect subtle stage differences. At the same time, spatial relationships mostly rely on fixed parameter mapping, and the changes in distance between segments in actual motion cannot be fully reflected. When there are non-uniform changes or short-term reverse fluctuations in the action, the results are prone to ambiguity. For example, local changes in continuous bending are difficult to distinguish, which affects the accuracy of activity range division and reduces the stability of the analysis results. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a method for spinal mobility analysis based on multi-sensor fusion; To achieve the above objectives, the present invention adopts the following technical solution: a spinal mobility analysis method based on multi-sensor fusion, comprising the following steps: S1: Obtain the angular velocity signal, acceleration signal and timestamp sequence of the inertial measurement device deployed at the location of the spine segment on the back of the human body, accumulate the angular velocity and differentiate adjacent time points, corresponding to the attitude angle change sequence and time axis, to obtain the segment attitude time sequence record set; S2: Based on the segment attitude time sequence record set, scan the continuous time points of the angle change sequence, identify the time positions of the continuous time points changing in the same direction, and obtain the segment response time sequence arrangement result; S3: Based on the segment attitude time sequence record set, the attitude angle change sequence is converted into a direction vector sequence. The dot product of the direction vectors at adjacent time points is performed and the included angle value is extracted. The included angle value range is compared and the turning node is selected to obtain the segment direction trajectory sequence. S4: Based on the segment attitude time sequence record set and the segment orientation trajectory sequence, update the segment spatial coordinates by projecting the angle change direction at the time point, extract the spatial coordinate distance between adjacent segments and compare it with the distance calibration value between segments to obtain the segment spatial pose change trajectory. S5: Based on the segment response time sequence arrangement results and the segment spatial pose change trajectory, divide the time interval and extract the segment displacement direction and change amplitude. By comparing the range of variation of the differential interval, obtain the distribution results of the spinal segment mobility interval.

[0005] As a further aspect of the present invention, the segmental attitude time sequence record set includes attitude angle change values, unified time axis identifiers, time index sequences, angular velocity integral results, and adjacent difference features; the segmental response time sequence arrangement results include response start point sequences, time sorting indexes, adjacent interval value sets, response rhythm markers, and continuity judgment labels; the segmental direction trajectory sequence includes direction vector sets, vector angle values, turning node identifiers, unitized direction vectors, and trajectory segmentation markers; the segmental spatial pose change trajectory includes three-dimensional spatial coordinate sequences, inter-segment distance deviations, scaling adjustment coefficients, pose direction consistency indicators, and spatial path accumulation; and the spinal segment mobility interval distribution results include time interval division sets, displacement direction classifications, change amplitude intervals, interval connection relationships, and spatial correspondence mappings.

[0006] As a further embodiment of the present invention, the corresponding attitude angle change sequence refers to the time attitude angle change data sequence that corresponds one-to-one with the time axis after integrating the angular velocity and differentiating adjacent time points. The continuous time points changing in the same direction refer to the time interval and starting position corresponding to multiple consecutive time points changing in the same direction in the angle change sequence.

[0007] As a further aspect of the present invention, the turning point refers to the point in time when the angle between the direction vectors of adjacent time points falls into a set turning range, indicating a change in the direction of motion. The updated segment spatial coordinates refer to the projection of the angle change direction at each time point and the accumulation of displacement, updating the position of the segment in three-dimensional space time by time.

[0008] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Obtain the angular velocity signal, acceleration signal and timestamp sequence output by the inertial measurement device deployed at the position of the spine segment on the back of the human body, perform equal interval rearrangement and alignment of the timestamp sequence to unify the time axis, perform accumulation operation on the angular velocity signal in time order and perform differential processing on the adjacent time points of the accumulation sequence to obtain the attitude angle change sequence. S102: Based on the attitude angle change sequence, associate it with the unified time axis index, match the attitude angle change sequence with the unified time axis point by point in chronological order, and bind the attitude angle change value at each time point to obtain the attitude time corresponding sequence. S103: Based on the attitude time correspondence sequence, the attitude angle change values ​​at time points are continuously arranged in chronological order, the attitude angle change values ​​are associated with the corresponding time points, and the time axis is correlated with the attitude angle change sequence to obtain the segment attitude time sequence record set.

[0009] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Based on the segmental attitude time sequence record set, scan the angle change sequence point by point, determine the sign of the angle change direction at each time point, segment and filter the continuous time points according to the sign consistency, extract the continuous same-direction change segments and record the corresponding time positions to obtain the same-direction change time segments. S202: Based on the same-direction change time segments, count the number of time points in each segment, compare the number of time points with the three-point consecutive judgment benchmark, filter the segment start time position corresponding to the length of the three consecutive points, sort the time positions according to the time sequence, and extract the time difference between adjacent time positions to obtain the segment response start time sequence. S203: Based on the segment response start time sequence, arrange the time sequence in chronological order, associate the time difference of adjacent time positions, and map the time positions to the time intervals to obtain the segment response time sequence arrangement result.

[0010] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Based on the attitude angle change sequence in the segment attitude time sequence record, perform three-dimensional coordinate mapping on the attitude angle change sequence, convert the attitude angle value at each time point into spatial direction components, perform vector normalization processing on the spatial direction components and arrange them in time order to obtain a direction vector sequence. S302: Based on the direction vector sequence, perform a dot product operation on the direction vectors of adjacent time points and extract the included angle value. By comparing the included angle value with the range of direction turning angles, filter the time points within the turning angle range and arrange the time order to obtain the turning node sequence. S303: Based on the sequence of turning points, the direction vectors between adjacent turning points are accumulated and the accumulation result is normalized. The processed direction vectors are mapped in time order to obtain the sequence of segment direction trajectories.

[0011] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Based on the attitude angle change sequence and the segment direction trajectory sequence in the segment attitude time-series record set, project the attitude angle change direction at each time point, superimpose the direction projection result with the initial segment spatial coordinates, and update the spatial position in time order to obtain the segment spatial coordinate sequence. S402: Based on the segment spatial coordinate sequence, extract the spatial coordinate distance between adjacent segments, compare the distance with the inter-segment distance calibration value, and proportionally scale the spatial coordinates of the time point that deviates along the direction of the line connecting adjacent segments to obtain the spatial position correction sequence. S403: Based on the spatial position correction sequence, determine the consistency of the scaling direction at consecutive time points, perform cumulative processing on the scaling with consistent direction and associate and map them in time order to obtain the segment spatial pose change trajectory.

[0012] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Based on the segment response timing arrangement result and the segment spatial pose change trajectory, perform time sequence division on the segment spatial pose change under a unified time axis, divide the continuous time point spatial pose change into multiple time intervals and index them according to the segment number to obtain the segment time interval sequence. S502: Based on the segment time interval sequence, extract the displacement direction and change amplitude of segment spatial pose changes within the time interval, compare the change amplitude of the same segment in different time intervals, and obtain the segment change interval sequence. S503: Based on the segment change interval sequence, connect the segment change intervals in chronological order, and map the connected intervals to the segment spatial pose change trajectory to obtain the distribution result of spinal segment mobility intervals.

[0013] A spinal mobility analysis system based on multi-sensor fusion includes: The attitude acquisition module acquires the angular velocity signal, acceleration signal and timestamp sequence output by the inertial measurement device deployed at the location of the spine segment on the back of the human body, rearranges the timestamps to unify the time axis, accumulates the angular velocity and performs differential processing on adjacent time points, and obtains the segment attitude time sequence record set corresponding to the attitude angle change sequence and time axis. Based on the segment attitude time sequence record set, the response recognition module scans the continuous time points of the angle change sequence, filters the positions of three consecutive time points that change in the same direction, sets the time position as the segment response start position, sorts the response start positions by time and extracts the time interval to obtain the segment response time sequence arrangement result. Based on the segment attitude time sequence record set, the orientation processing module converts the attitude angle change sequence into an orientation vector sequence, performs a dot product on the orientation vectors at adjacent time points and extracts the included angle value, compares the included angle value to determine the range and filters the turning nodes, and performs accumulation and normalization processing on the orientation vectors between nodes to obtain the segment orientation trajectory sequence. Based on the segment posture time sequence record set and the segment orientation trajectory sequence, the pose constraint module projects the angle change direction at time points and updates the segment spatial coordinates, extracts the spatial coordinate distance between adjacent segments and compares it with the distance calibration value between segments, and scales and adjusts the deviation position along the connecting line direction to obtain the segment spatial pose change trajectory. Based on the segment response time sequence arrangement and the segment spatial pose change trajectory, the activity analysis module divides the time intervals and extracts the segment displacement direction and change amplitude. It compares the range of changes in the differentiated time intervals and selects the change intervals. It connects and matches the spatial relationships in chronological order to obtain the distribution results of the spinal segment mobility intervals.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, an attitude evolution mechanism that combines angular velocity accumulation and time difference is constructed. This mechanism is combined with continuous unidirectional change recognition and the dot product of direction vectors is introduced to determine motion turning points and characterize fine-grained change trends. At the same time, the spatial coordinates are updated based on the direction of angle change and the trajectory is corrected by integrating segment distance constraints. The displacement direction and amplitude differences are extracted at the time interval level to complete the distribution division, strengthen the stage characteristics of the motion process, improve the sensitivity of turning point recognition and enhance the consistency of the trajectory, so that the activity results show spatial correlation. Attached Figure Description

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

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

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

[0018] 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.

[0019] Please see Figure 1 This invention provides a method for analyzing spinal mobility based on multi-sensor fusion, comprising the following steps: S1: Obtain the angular velocity signal, acceleration signal and timestamp sequence of the inertial measurement device deployed at the location of the spine segment on the back of the human body, rearrange the timestamp sequence at equal intervals to unify the time axis, accumulate the angular velocity signal in time order and perform differential processing on the accumulation result between adjacent time points, and match the attitude angle change sequence with the time axis to obtain the segment attitude time sequence record set; S2: Based on the segment attitude time sequence record set, scan the continuous time points of the angle change sequence, identify the time position where the angle changes in three consecutive time points remain in the same direction, set the time position as the segment response start position, sort the segment response start positions in chronological order, and extract the adjacent time intervals to obtain the segment response time sequence arrangement result; S3: Based on the segmental attitude time sequence record set, the attitude angle change sequence is converted into a direction vector sequence. The dot product operation is performed on the direction vectors of adjacent time points and the included angle value is extracted. The included angle value is compared with the direction turning angle range and the turning node is selected. The direction vectors between the turning nodes are accumulated and normalized to obtain the segmental direction trajectory sequence. S4: Based on the segmental attitude time sequence record set and segmental orientation trajectory sequence, project the angle change direction at each time point and update the segmental spatial coordinates. Extract the distance between the updated segmental spatial coordinates and the spatial coordinates of adjacent segments and compare it with the distance calibration value between segments. For spatial positions with deviations, scale and adjust them along the connecting line direction. Identify the consistency of the scaling amount in the direction of continuous time points and accumulate it to obtain the segmental spatial pose change trajectory. S5: Call the segment response time sequence arrangement results and segment spatial pose change trajectory, divide the segment spatial pose change under a unified time axis into time intervals, extract the displacement direction and change amplitude of each segment in the differentiated time intervals, compare the change range of the same segment in the differentiated time intervals and select the change intervals, connect the segment change intervals in time order and match the spatial correspondence to obtain the distribution results of spinal segment mobility intervals.

[0020] The segmental attitude time sequence record set includes attitude angle change values, unified time axis identifier, time index sequence, angular velocity integral results, and adjacent difference features. The segmental response time sequence arrangement results include response start point sequence, time sorting index, adjacent interval value set, response rhythm label, and continuity judgment label. The segmental direction trajectory sequence includes direction vector set, vector angle value, turning node identifier, unitized direction vector, and trajectory segment label. The segmental spatial pose change trajectory includes three-dimensional spatial coordinate sequence, inter-segment distance deviation, scaling adjustment coefficient, pose direction consistency index, and spatial path accumulation. The spinal segmental mobility interval distribution results include time interval division set, displacement direction classification, change amplitude interval, interval connection relationship, and spatial correspondence mapping.

[0021] Please see Figure 2 The specific steps of S1 are as follows: S101: Obtain the angular velocity signal, acceleration signal and timestamp sequence output by the inertial measurement device deployed at the position of the spine segment on the back of the human body, perform equal interval rearrangement and alignment of the timestamp sequence to unify the time axis, perform accumulation operation on the angular velocity signal in time order and perform differential processing on the adjacent time points of the accumulation sequence to obtain the attitude angle change sequence. First, raw data signals are acquired and transmitted in real time by inertial measurement units (IMUs) deployed at different segments of the spine on the human back. These raw data signals include triaxial angular velocity signals, triaxial acceleration signals, and a raw timestamp sequence corresponding to each set of physical quantity data. The sampling frequency of the raw timestamp sequence is set to 100 Hz, meaning the time interval between adjacent data points is 10 milliseconds. By performing an equal-interval rearrangement and alignment operation on the raw timestamp sequence, the inconsistency of the time axis caused by hardware clock offset or wireless transmission jitter is eliminated. A unified time axis with a starting time of 0 milliseconds and a constant step size of 10 milliseconds is established. The angular velocity and acceleration signals are mapped onto this unified time axis. During the mapping process, a unified sampling period of 10 milliseconds is used to perform a multiplication operation on the raw angular velocity value of each sampling point to calculate the values ​​of each sampled point. The axial minute radian increment is then accumulated sequentially along the time axis. This involves summing the current increment with the accumulated values ​​from all previous times to obtain an integral accumulation sequence of angular velocity reflecting the rotation trend. Next, a difference processing is performed on this accumulation sequence between adjacent time points to calculate the numerical deviation between the current and previous accumulated values. This eliminates constant interference caused by sensor zero-point offset, thereby deriving the attitude angle change value at each time point relative to the initial state. For example, when acquiring the 10th sampling point, its original angular velocity is 0.5 degrees per second. Multiplying this by a 0.01-second sampling interval yields an angular increment of 0.005 degrees. If the accumulated value of the first 9 points is 0.045 degrees, then the accumulated value at the current point is 0.05 degrees. After differential processing and removal of drift terms, the final attitude angle change sequence is obtained.

[0022] S102: Based on the attitude angle change sequence, associate it with the unified time axis index, match the attitude angle change sequence with the unified time axis point by point in chronological order, and bind the attitude angle change value at each time point to obtain the attitude time corresponding sequence. First, based on the attitude angle change sequence generated in the previous steps, it is associated with the established unified timeline index with a start time of 0 milliseconds and an interval of 10 milliseconds. A point-by-point matching operation is then performed, logically binding each discrete value in the attitude angle change sequence to its corresponding time point on the unified timeline. During execution, the index sequence of the unified timeline is called, traversing all elements in the attitude angle change sequence. The first attitude angle change value is bound to the 10-millisecond scale, the 500th attitude angle change value to the 5000-millisecond scale, and so on, establishing a one-to-one mapping table. If during the matching process... If a missing attitude angle data point is detected due to signal transmission fluctuations, the attitude angle values ​​of the two adjacent valid points before and after the missing point are retrieved, and their arithmetic mean is calculated. This arithmetic mean is then used as the compensation value for the missing point and forcibly bound. For example, if the data at 100 milliseconds is 0.12 degrees, the data at 120 milliseconds is 0.14 degrees, and the data at 110 milliseconds is missing, the average value is calculated to be 0.13 degrees and filled into the 110 milliseconds index. Subsequently, the attitude angle change value after compensation and alignment is used as an attribute label and bound to each time point, ultimately obtaining the attitude time sequence.

[0023] S103: Based on the attitude time correspondence sequence, the attitude angle change values ​​at time points are continuously arranged in chronological order, the attitude angle change values ​​are associated with the corresponding time points, and the time axis is matched with the attitude angle change sequence to obtain the segment attitude time sequence record set. First, based on the attitude time-correspondence sequence, the attitude angle change values ​​at all time points are continuously arranged in ascending order of the time axis. Through a logical association mechanism, each discrete attitude angle change value is strongly coupled with its corresponding precise time point. The one-dimensional linear time axis is combined with the one-dimensional attitude angle change value sequence to construct a dynamic change trajectory with time as the axial coordinate and attitude angle change as the attribute component. During execution, all aligned attitude angle values ​​from the previous steps are retrieved and stored in structured storage units in ascending order of timestamp. Each record is defined as a composite containing the segment number, precise timestamp, and corresponding three-axis attitude angle change. For example, for the 12th thoracic segment, at 1500 milliseconds, the attitude angle change components of 0.12 degrees, 0.05 degrees, and -0.03 degrees are stored as a vector in the corresponding slot of the time-series record set. By traversing the storage units of all monitored segments, a multi-dimensional data matrix that can completely reconstruct the entire process of spinal movement is formed, and finally, the segment attitude time-series record set is obtained.

[0024] Please see Figure 3 The specific steps of S2 are as follows: S201: Based on the segmental attitude time sequence record set, scan the angle change sequence point by point, determine the sign of the angle change direction at each time point, and segment and filter continuous time points according to the consistency of the sign. Extract and record the corresponding time positions of continuous unidirectional change segments to obtain unidirectional change time segments. First, based on the segmental attitude time-series record set, a point-by-point scanning procedure is executed to perform feature detection on the angle change sequence. For each value in the sequence, the sign representing the direction of motion is determined. By calculating the difference between the current attitude angle change value and the value at the previous moment, the sign of the difference is determined. If the difference is greater than 0, the direction sign is determined to be positive, representing an increase in angle; if the difference is less than 0, the direction sign is determined to be negative, representing a decrease in angle. According to the principle of sign consistency, a segmented filtering operation is performed on the continuously distributed time points. When multiple consecutive time points are detected to have unchanged signs, i.e., consecutive positive or consecutive negative sequences appear, the result is considered. If a sequence of signs is obtained, it is determined that the sequence belongs to the same unidirectional spinal bending or straightening process. These segments with continuous identical sign characteristics are extracted, and the start and end index positions of each segment on a unified time axis are recorded simultaneously. For example, the attitude angle change values ​​of 5 consecutive sampling points are retrieved, with values ​​of 0.02, 0.05, 0.08, 0.11, and 0.13, respectively. The sign sequence obtained by calculating the difference between adjacent points is positive, and it is determined that it is increasing in the same direction. If the change value of the 6th point decreases, resulting in a negative difference, it is determined that the segment in the same direction ends at the 5th point, and finally the segment of change in the same direction is obtained.

[0025] S202: Based on the same direction change time segments, count the number of time points in each segment, compare the number of time points with the three consecutive point judgment criteria, filter the segment start time position corresponding to the length of the three consecutive points, sort the time positions according to the time sequence, and extract the time difference between adjacent time positions to obtain the segment response start time sequence. First, based on the same-direction change time segments, the total number of time points contained in each extracted segment is precisely counted. The calculated number of time points is compared with a preset three-point judgment benchmark. This benchmark is set to 3 sampling points, corresponding to a duration of 30 milliseconds, to filter out noise generated by non-autonomous motion. During execution, the statistical value of the time point of each same-direction segment is called. If the statistical value is greater than or equal to 3, the segment is determined to be a valid motion response, and its starting time position is retained. If the statistical value is less than 3, it is determined to be high-frequency noise interference and is discarded. After filtering out all valid segments, these starting time positions are sorted in ascending order according to the order of their occurrence. Then, two adjacent starting time positions are retrieved, and the difference operation is performed to calculate the time interval between adjacent responses. For example, if the first valid segment starts at 500 milliseconds and the second valid segment starts at 850 milliseconds, the difference between the two is 350 milliseconds. This series of time difference values ​​are arranged in order to finally obtain the segment response starting time sequence.

[0026] S203: Based on the segment response start time series, arrange the time series in chronological order, associate the time difference of adjacent time positions, and map the time positions to the time intervals to obtain the segment response time sequence arrangement result; First, based on the segment response start time series, all start time positions are arranged in a linear order of time. A sequential mapping is established between the start time position and the time difference of the corresponding adjacent time position. The absolute time point of each moment is mapped one-to-one with the subsequent time interval. During execution, a two-column mapping table is constructed. The first column stores the absolute start time of each segment response, and the second column stores the duration from that moment to the next response trigger. The first response start time position of 1200 milliseconds is retrieved, and its corresponding time interval is 250 milliseconds. The second response start time position of 1450 milliseconds is retrieved, and its corresponding time interval is 300 milliseconds. These data items are mapped to the segment number label. By traversing all monitored segments, a data matrix that can reflect the sequence and pace of response of different spinal segments is formed, and the segment response time sequence arrangement result is finally obtained.

[0027] Please see Figure 4 The specific steps of S3 are as follows: S301: Based on the attitude angle change sequence in the segment attitude time sequence record, perform three-dimensional coordinate mapping on the attitude angle change sequence, convert the attitude angle value at each time point into spatial direction components, perform vector normalization processing on the spatial direction components and arrange them in time order to obtain a direction vector sequence. First, based on the attitude angle change sequence stored in the segment attitude time-series record, a three-dimensional spatial coordinate mapping operation is performed on each set of three-axis angle data in the sequence. The pitch, roll, and yaw angle changes at each moment are converted into spatial direction components in a rectangular coordinate system. During the execution, cosine and sine logic functions are called to convert the angle values ​​into projection values ​​in a unit spherical coordinate system. Then, vector normalization processing is performed on the derived spatial direction components. The sum of squares of the three axial components is calculated, and the square root of the sum of squares is then performed to obtain the vector magnitude. Finally, each original component is divided by this magnitude to make the length of the processed direction vector constant to 1. The normalized unit direction vectors are arranged in chronological order along the time axis. For example, if the attitude angle components at a certain moment are retrieved, the initial spatial direction components after conversion are 0.49, 0.17, and 0.08, and their magnitude is calculated to be 0.525, the normalized components after division are 0.933, 0.324, and 0.152, which constitute the unit direction vector at that moment, and finally, the direction vector sequence is obtained.

[0028] S302: Based on the direction vector sequence, perform a dot product operation on the direction vectors of adjacent time points and extract the included angle value. By comparing the included angle value with the range of direction turning angles, filter the time points within the turning angle range and arrange the time order to obtain the turning node sequence. First, based on the direction vector sequence, a dot product operation is performed on the direction vectors corresponding to two adjacent time points in the sequence. By calculating the sum of the products of the corresponding components of the two three-dimensional vectors, the angle between the two adjacent time vectors is extracted. The extracted angle value is then compared with a preset direction turning angle range, which is set between 5 and 15 degrees. During the process, the angle sequence is scanned point by point. If the angle value at a certain time falls within this range, the point is determined to be a critical point where the direction of motion has significantly shifted. For example, the dot product calculation is performed on the direction vectors at 100 milliseconds and 110 milliseconds, and the angle value is 8.5 degrees. Since it is within the range of 5 to 15 degrees, this time is determined to be a valid turning point. Subsequently, all turning points that meet the screening criteria are arranged in chronological order to form a set of skeletal nodes of the spinal motion path, and finally, a turning point sequence is obtained.

[0029] S303: Based on the sequence of turning nodes, the direction vectors between adjacent turning nodes are accumulated and the accumulation result is normalized. The processed direction vectors are mapped in time order to obtain the sequence of segment direction trajectories. First, based on the sequence of turning points, all direction vectors between two adjacent turning points are accumulated to obtain a composite vector reflecting the overall motion trend within that time interval. This accumulated result is then normalized by recalculating the magnitude of the accumulated vector and dividing each component by the magnitude to restore its length to 1. During this process, all direction vectors between the first and second turning points are retrieved, and their horizontal, vertical, and longitudinal components are summed to obtain the composite displacement direction. For example, if the accumulated composite vector components are 2.5, 1.2, and -0.8, and its magnitude is 2.886, the normalized components after division are approximately 0.866, 0.416, and -0.277. These processed unit vectors are then mapped to the corresponding motion intervals in chronological order, constructing a motion chain composed of feature directions, ultimately resulting in a sequence of segmental directional trajectories.

[0030] Please see Figure 5 The specific steps of S4 are as follows: S401: Based on the attitude angle change sequence and the segment direction trajectory sequence in the segment attitude time-series record set, project the attitude angle change direction at each time point, superimpose the direction projection result with the initial segment spatial coordinates, and update the spatial position in time order to obtain the segment spatial coordinate sequence. First, based on the attitude angle change sequence and the generated segment direction trajectory sequence from the segment attitude time-series record set, a projection calculation operation is performed. The attitude angle change amplitude at each time point is projected onto the corresponding motion direction vector, and the spatial displacement increment within the sampling period is calculated. During the execution, the segment spatial coordinate values ​​at the initial time are retrieved as the starting reference, usually set as an anatomical reference point. The calculated displacement increment is superimposed with the spatial coordinates at the previous time, that is, the position values ​​of the segment in the three-dimensional coordinate system are updated in real time through summation. For example, if the initial spatial coordinates are retrieved as 0, 0, 0, and the displacement increments calculated based on the angle and direction at the current time are retrieved as 2.5 mm, 1.1 mm, 0.4 mm, the spatial coordinates at the new time are obtained after the addition operation. By updating all sampling points one by one in chronological order, a coordinate set reflecting the actual displacement path of the segment is formed, and finally the segment spatial coordinate sequence is obtained.

[0031] S402: Based on the segment spatial coordinate sequence, extract the spatial coordinate distance between adjacent segments, compare the distance with the inter-segment distance calibration value, and proportionally scale the spatial coordinates of the time point that deviates along the direction of the line connecting adjacent segments to obtain the spatial position correction sequence. First, based on the segmental spatial coordinate sequence, the real-time spatial coordinate distance is extracted by calculating the Euclidean distance between the spatial coordinate points of two adjacent spinal segments. This real-time distance is then compared with a pre-stored inter-segment distance calibration value, which is set according to the physiological spacing of the subject's back segments and is usually between 20 mm and 30 mm. During the process, a deviation tolerance threshold of 10% is set. When the ratio deviation between the real-time distance and the calibration value at a certain moment exceeds this threshold, it is determined that there is a drift error in the coordinates of that point. At this time, the scaling logic is invoked. Taking the coordinates of the adjacent previous segment as the origin, the coordinates of the current deviation point are linearly scaled along the direction of the line connecting the two points, forcing the distance between them to revert to the calibration value. For example, if the real-time distance is retrieved as 35.0 mm and compared with the calibration value of 25.0 mm, the scaling ratio is calculated to be 0.714. All components of the current point coordinates are multiplied by 0.714, and finally, the spatial position correction sequence is obtained.

[0032] S403: Based on the spatial position correction sequence, determine the consistency of the scaling direction at consecutive time points, perform cumulative processing on the scaling with consistent direction and associate and map them in time order to obtain the segment spatial pose change trajectory. First, based on the spatial position correction sequence, the consistency of the scaling direction of consecutive time points is determined by analyzing the changing trend of displacement vectors at adjacent time points. When scaling vectors with a directional deviation of less than 5 degrees are detected consecutively, a systematic error is identified. These scaling vectors with consistent directions are accumulated to generate a global compensation vector. During the execution process, the accumulated compensation value is mapped to each coordinate point according to the chronological order of the time axis to smooth the original spatial pose. For example, if the correction vectors of 10 consecutive points are found to be consistent, the total compensation value is obtained by summing these 10 correction components and distributing them to each time point according to their weights. The final output trajectory data excludes non-physiological coordinate jumps and completely restores the motion posture of the spinal segments, ultimately obtaining the segmental spatial pose change trajectory.

[0033] Please see Figure 6 The specific steps of S5 are as follows: S501: Based on the segment response timing arrangement result and the segment spatial pose change trajectory, perform time sequence division on the segment spatial pose change under a unified time axis, divide the continuous time point spatial pose change into multiple time intervals and index them according to the segment number to obtain the segment time interval sequence. First, based on the segmental response time sequence and segmental spatial pose change trajectory, the dynamic evolution of spinal segment pose is divided into time dimensions within a unified time axis framework. According to the motion response start point identified in the previous steps, the continuous time stream is divided into multiple time intervals with independent motion characteristics. Each interval is assigned a unique index label and associated with the corresponding spinal segment number. During the execution process, node values ​​in the response start time sequence, such as 1200 milliseconds and 1450 milliseconds, are retrieved. All pose trajectory data within this interval are extracted and encapsulated into a motion stage. In this way, the originally complex, long-cycle spinal motion is decomposed into a series of quantifiable basic action units, ultimately yielding the segmental time interval sequence.

[0034] S502: Based on the segment time interval sequence, extract the displacement direction and change amplitude of segment spatial pose changes within the time interval, compare the change amplitude of the same segment in different time intervals, and obtain the segment change interval sequence. First, based on the segment time interval sequence, for each divided time interval, the segment spatial coordinates at the start and end times of the interval are extracted, and the geometric displacement vector between them is calculated. By calculating the magnitude of the displacement vector in three-dimensional space, the amplitude of the segment's motion change within a specific interval is obtained. The amplitude of the same segment's change in different time intervals is compared and analyzed, or the amplitudes of different segments in the same interval are compared laterally. During the execution process, the start and end coordinates of a segment in interval 1 are retrieved, the displacement components are calculated, and the motion amplitude is derived. For example, if the start points are 2.5, 1.1, and 0.4, and the end points are 10.2, 5.5, and 2.8, the calculated amplitude of change is 9.2 mm. These values ​​are arranged in segment order to reflect the activity level of each region, and finally, the segment change interval sequence is obtained.

[0035] S503: Based on the segment change interval sequence, connect the segment change intervals in chronological order, and map the connected intervals to the segment spatial pose change trajectory to obtain the distribution result of the spinal segment mobility interval; First, based on the segmental change interval sequence, the change intervals of each segment are spatially logically connected according to the chronological order of occurrence, constructing a kinematic chain that runs through the entire spine. The connected interval sequence is then spatially mapped to the segmental spatial pose change trajectory, and the projection coverage of each interval on the longitudinal axis of the entire spine is calculated. During the process, the calculated change amplitude values ​​of each segment are mapped to the corresponding physiological segments in the spinal anatomy atlas, and the mobility of each region is reflected by color depth or numerical labeling. For example, comparing the projection amplitude of different segments, the projection amplitude of the cervical spine is 15.0 mm, which is evaluated as normal; the projection amplitude of the thoracic spine is 6.0 mm, which is judged as limited mobility; and the projection amplitude of the lumbar spine is 12.5 mm, which is evaluated as normal, ultimately yielding the distribution results of spinal segmental mobility intervals.

[0036] Please see Figure 7 A spinal mobility analysis system based on multi-sensor fusion includes: The attitude acquisition module acquires the angular velocity signal, acceleration signal and timestamp sequence output by the inertial measurement device deployed at the location of the spine segment on the back of the human body, rearranges the timestamps to unify the time axis, accumulates the angular velocity and performs differential processing on adjacent time points, and obtains the segment attitude time sequence record set corresponding to the attitude angle change sequence and time axis. The response recognition module scans continuous time points of the angle change sequence based on the segment attitude time series, filters the positions of three consecutive time points that change in the same direction, sets the time position as the segment response start position, sorts the response start positions by time and extracts the time interval to obtain the segment response time series arrangement result. The orientation processing module converts the attitude angle change sequence into an orientation vector sequence based on the segment attitude time sequence record set. It performs dot product on the orientation vectors at adjacent time points and extracts the included angle value. It compares the included angle value to determine the range and filters the turning nodes. It performs accumulation and normalization processing on the orientation vectors between nodes to obtain the segment orientation trajectory sequence. The pose constraint module projects the angle change direction at time points and updates the segment spatial coordinates based on the segment attitude time sequence record set and segment orientation trajectory sequence. It extracts the spatial coordinate distance between adjacent segments and compares it with the distance calibration value between segments. It scales and adjusts the deviation position along the connecting line direction to obtain the segment spatial pose change trajectory. The activity analysis module divides time intervals based on the segment response time sequence and the segment spatial pose change trajectory, extracts the segment displacement direction and change amplitude, compares the range of variation of the differentiated time intervals and selects the change intervals, connects them in time sequence and matches the spatial relationship to obtain the distribution results of the spinal segment mobility intervals.

[0037] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for analyzing spinal mobility based on multi-sensor fusion, characterized in that, Includes the following steps: S1: Obtain the angular velocity signal, acceleration signal and timestamp sequence of the inertial measurement device deployed at the location of the spine segment on the back of the human body, accumulate the angular velocity and differentiate adjacent time points, corresponding to the attitude angle change sequence and time axis, to obtain the segment attitude time sequence record set; S2: Based on the segment attitude time sequence record set, scan the continuous time points of the angle change sequence, identify the time positions of the continuous time points changing in the same direction, and obtain the segment response time sequence arrangement result; S3: Based on the segment attitude time sequence record set, the attitude angle change sequence is converted into a direction vector sequence. The dot product of the direction vectors at adjacent time points is performed and the included angle value is extracted. The included angle value range is compared and the turning node is selected to obtain the segment direction trajectory sequence. S4: Based on the segment attitude time sequence record set and the segment orientation trajectory sequence, update the segment spatial coordinates by projecting the angle change direction at the time point, extract the spatial coordinate distance between adjacent segments and compare it with the distance calibration value between segments to obtain the segment spatial pose change trajectory. S5: Based on the segment response time sequence arrangement results and the segment spatial pose change trajectory, divide the time interval and extract the segment displacement direction and change amplitude. By comparing the range of variation of the differential interval, obtain the distribution results of the spinal segment mobility interval.

2. The spinal mobility analysis method based on multi-sensor fusion according to claim 1, characterized in that, The segmental attitude time sequence record set includes attitude angle change values, unified time axis identifiers, time index sequences, angular velocity integral results, and adjacent difference features. The segmental response time sequence arrangement results include response start point sequences, time sorting indexes, adjacent interval value sets, response rhythm markers, and continuity judgment labels. The segmental direction trajectory sequence includes direction vector sets, vector angle values, turning node identifiers, unitized direction vectors, and trajectory segmentation markers. The segmental spatial pose change trajectory includes three-dimensional spatial coordinate sequences, inter-segment distance deviations, scaling adjustment coefficients, pose direction consistency indicators, and spatial path accumulation. The spinal segmental mobility interval distribution results include time interval division sets, displacement direction classifications, change amplitude intervals, interval connection relationships, and spatial correspondence mappings.

3. The spinal mobility analysis method based on multi-sensor fusion according to claim 1, characterized in that, The corresponding attitude angle change sequence refers to the data sequence of attitude angle changes at each time point that is obtained by integrating the angular velocity and differentiating adjacent time points, and that corresponds one-to-one with the time axis. The continuous time points changing in the same direction refer to the time interval and starting position corresponding to multiple consecutive time points changing in the same direction in the angle change sequence.

4. The spinal mobility analysis method based on multi-sensor fusion according to claim 1, characterized in that, The turning point refers to the point in time when the angle between the direction vectors of adjacent time points falls into a set turning range, indicating a change in the direction of motion. The updated segment spatial coordinates refer to the projection of the angle change direction at each time point and the accumulation of displacement, updating the position of the segment in three-dimensional space time by time.

5. The spinal mobility analysis method based on multi-sensor fusion according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Obtain the angular velocity signal, acceleration signal and timestamp sequence output by the inertial measurement device deployed at the position of the spine segment on the back of the human body, perform equal interval rearrangement and alignment of the timestamp sequence to unify the time axis, perform accumulation operation on the angular velocity signal in time order and perform differential processing on the adjacent time points of the accumulation sequence to obtain the attitude angle change sequence. S102: Based on the attitude angle change sequence, associate it with the unified time axis index, match the attitude angle change sequence with the unified time axis point by point in chronological order, and bind the attitude angle change value at each time point to obtain the attitude time corresponding sequence. S103: Based on the attitude time correspondence sequence, the attitude angle change values ​​at time points are continuously arranged in chronological order, the attitude angle change values ​​are associated with the corresponding time points, and the time axis is correlated with the attitude angle change sequence to obtain the segment attitude time sequence record set.

6. The spinal mobility analysis method based on multi-sensor fusion according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Based on the segmental attitude time sequence record set, scan the angle change sequence point by point, determine the sign of the angle change direction at each time point, segment and filter the continuous time points according to the sign consistency, extract the continuous same-direction change segments and record the corresponding time positions to obtain the same-direction change time segments. S202: Based on the same-direction change time segments, count the number of time points in each segment, compare the number of time points with the three-point consecutive judgment benchmark, filter the segment start time position corresponding to the length of the three consecutive points, sort the time positions according to the time sequence, and extract the time difference between adjacent time positions to obtain the segment response start time sequence. S203: Based on the segment response start time sequence, arrange the time sequence in chronological order, associate the time difference of adjacent time positions, and map the time positions to the time intervals to obtain the segment response time sequence arrangement result.

7. The spinal mobility analysis method based on multi-sensor fusion according to claim 1, characterized in that, The specific steps for S3 are as follows: S301: Based on the attitude angle change sequence in the segment attitude time sequence record, perform three-dimensional coordinate mapping on the attitude angle change sequence, convert the attitude angle value at each time point into spatial direction components, perform vector normalization processing on the spatial direction components and arrange them in time order to obtain a direction vector sequence. S302: Based on the direction vector sequence, perform a dot product operation on the direction vectors of adjacent time points and extract the included angle value. By comparing the included angle value with the range of direction turning angles, filter the time points within the turning angle range and arrange the time order to obtain the turning node sequence. S303: Based on the sequence of turning points, the direction vectors between adjacent turning points are accumulated and the accumulation result is normalized. The processed direction vectors are mapped in time order to obtain the sequence of segment direction trajectories.

8. The spinal mobility analysis method based on multi-sensor fusion according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Based on the attitude angle change sequence and the segment direction trajectory sequence in the segment attitude time-series record set, project the attitude angle change direction at each time point, superimpose the direction projection result with the initial segment spatial coordinates, and update the spatial position in time order to obtain the segment spatial coordinate sequence. S402: Based on the segment spatial coordinate sequence, extract the spatial coordinate distance between adjacent segments, compare the distance with the inter-segment distance calibration value, and proportionally scale the spatial coordinates of the time point that deviates along the direction of the line connecting adjacent segments to obtain the spatial position correction sequence. S403: Based on the spatial position correction sequence, determine the consistency of the scaling direction at consecutive time points, perform cumulative processing on the scaling with consistent direction and associate and map them in time order to obtain the segment spatial pose change trajectory.

9. The spinal mobility analysis method based on multi-sensor fusion according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Based on the segment response timing arrangement result and the segment spatial pose change trajectory, perform time sequence division on the segment spatial pose change under a unified time axis, divide the continuous time point spatial pose change into multiple time intervals and index them according to the segment number to obtain the segment time interval sequence. S502: Based on the segment time interval sequence, extract the displacement direction and change amplitude of segment spatial pose changes within the time interval, compare the change amplitude of the same segment in different time intervals, and obtain the segment change interval sequence. S503: Based on the segment change interval sequence, connect the segment change intervals in chronological order, and map the connected intervals to the segment spatial pose change trajectory to obtain the distribution result of spinal segment mobility intervals.

10. A spinal mobility analysis system based on multi-sensor fusion, characterized in that, The system is used to implement the spinal mobility analysis method based on multi-sensor fusion as described in any one of claims 1-9, the system comprising: The attitude acquisition module acquires the angular velocity signal, acceleration signal and timestamp sequence output by the inertial measurement device deployed at the location of the spine segment on the back of the human body, rearranges the timestamps to unify the time axis, accumulates the angular velocity and performs differential processing on adjacent time points, and obtains the segment attitude time sequence record set corresponding to the attitude angle change sequence and time axis. Based on the segment attitude time sequence record set, the response recognition module scans the continuous time points of the angle change sequence, filters the positions of three consecutive time points that change in the same direction, sets the time position as the segment response start position, sorts the response start positions by time and extracts the time interval to obtain the segment response time sequence arrangement result. Based on the segment attitude time sequence record set, the orientation processing module converts the attitude angle change sequence into an orientation vector sequence, performs a dot product on the orientation vectors at adjacent time points and extracts the included angle value, compares the included angle value to determine the range and filters the turning nodes, and performs accumulation and normalization processing on the orientation vectors between nodes to obtain the segment orientation trajectory sequence. Based on the segment posture time sequence record set and the segment orientation trajectory sequence, the pose constraint module projects the angle change direction at time points and updates the segment spatial coordinates, extracts the spatial coordinate distance between adjacent segments and compares it with the distance calibration value between segments, and scales and adjusts the deviation position along the connecting line direction to obtain the segment spatial pose change trajectory. Based on the segment response time sequence arrangement and the segment spatial pose change trajectory, the activity analysis module divides the time intervals and extracts the segment displacement direction and change amplitude. It compares the range of changes in the differentiated time intervals and selects the change intervals. It connects and matches the spatial relationships in chronological order to obtain the distribution results of the spinal segment mobility intervals.

Images