Data-driven based bone density scanner measurement mode switching system and method

Abstract

The present application relates to the technical field of medical detection, and particularly relates to a data-driven bone density scanner measurement mode switching system and method, which comprises a measurement platform, a body state monitoring module, a data processing module and a control module, the measurement platform integrates pressure distribution data of each part and body weight data, and real-time captures spine curvature data, limb bending angle data and body position offset data. A body state deviation index is generated based on the data, the deviation degree of the body state from the standard body state is quantified, and corresponding measurement strategies are dynamically triggered according to a fuzzy judgment mechanism, and the measurement mode of the bone density scanner is switched and controlled according to the judgment result. The present application accurately quantifies the body state deviation degree and adaptively selects the best measurement strategy, improves the accuracy of height and weight measurement and the reliability of bone density data under the bedridden state. And according to the real-time body state condition, the resource allocation and compensation calibration effect are optimized, and the overall detection efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of medical testing technology, and in particular to a data-driven bone densitometer measurement mode switching system and method. Background Technology

[0002] Bone mineral density testing (usually using dual-energy X-ray absorptiometry, DEXA) is a core diagnostic tool for assessing bone health. By quantifying bone mineral content, it provides crucial diagnostic information for osteoporosis, fracture risk, and bone metabolic diseases. In clinical analysis, patient age, sex, height, and weight are essential reference indicators for interpreting bone mineral density results. For example, height and weight are used to calculate the bone mineral density T-score (the standard deviation from the mean of young adults of the same sex), which directly affects disease classification and treatment planning.

[0003] However, the examination requires patients to complete the test in two steps: first, obtaining height and weight data on a standing height and weight measuring device, and then transferring to a supine bone density testing bed for a DEXA scan. This process is not only fragmented, but also requires patients to position themselves multiple times between different devices. This is particularly difficult and inconvenient for patients with limited mobility, the elderly, or those who are frail, significantly reducing their measurement experience and making the entire testing process time-consuming and inefficient. Furthermore, multiple measurement platforms not only increase the cost of repeatedly purchasing, installing, and maintaining multiple sets of equipment, but also require multiple power supplies, chassis, and complex communication interfaces. Routine calibration and maintenance require multiple technical teams. This necessitates different maintenance techniques and spare parts for different devices, increasing operating costs. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a data-driven bone densitometer measurement mode switching system and method.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The data-driven bone density scanner measurement mode switching system includes a measurement platform, a posture monitoring module, a data processing module, and a control module. The measurement platform integrates a pressure sensor array for pressure distribution data and weight data of various body parts. Laser ranging modules are set on both sides of the measurement platform to acquire spatial coordinate data from top to bottom and calculate height based on the spatial coordinate data. The posture monitoring module includes an IMU sensor group to capture spinal curvature data, limb bending angle data, and body position offset data in real time. The data processing module receives data from the pressure sensor array, the laser ranging module, and the IMU sensor group, and connects to the control module; The inertial measurement unit data is used to generate the posture deviation index SI based on the spinal curvature data, limb bending angle data, and body position offset data obtained by the IMU sensor group. The data processing module calculates the fuzzy matching degree based on the body posture deviation index SI, and obtains the membership functions with high matching degree, medium matching degree, and low matching degree respectively. The control module determines the measurement mode based on the calculated membership functions with high matching degree, medium matching degree, and low matching degree, combined with a threshold. Based on the determination result, it switches the measurement mode of the bone density scanner. The measurement modes include hip region measurement, hip and lumbar spine dual region measurement, and multidimensional data measurement. The multidimensional data measurement includes multi-angle re-image, high-resolution small ROI scanning, and combined measurement with body posture compensation algorithm.

[0006] Furthermore, the membership function for high matching degree is: ; in, express The membership value of the body posture deviation index at a high degree of match. The postural deviation index is calculated from IMU data. This represents the lower threshold for determining the matching degree. This indicates the upper limit threshold for determining the matching degree. express The slope parameter of the function.

[0007] Furthermore, the membership function of the matching degree is: ; in, express The membership value of the body posture deviation index in the degree of matching. Represents the central value, i.e. This indicates the center point of a slight deviation. This indicates the width of the Gaussian function. This indicates a deviation from the body posture index. This represents the lower threshold for determining the matching degree. This indicates the upper limit threshold for determining the matching degree.

[0008] Furthermore, the membership function for low-matching degree is: ; in, express The membership value of the body deviation index at low matching degree. This indicates a deviation from the body posture index. This represents the lower threshold for determining the matching degree. This indicates the upper limit threshold for determining the matching degree. express The slope parameter of the function.

[0009] The operation method of a data-driven bone densitometer measurement mode switching system includes the following steps: S1: Data acquisition and standard attitude reference modeling. The three-axis angular velocity and acceleration of key parts are acquired through the IMU sensor group. The corresponding attitude data vector is established according to the system's built-in reference attitude model. S2: Calculate the posture deviation index based on the posture data vector. First, the degree of posture deviation of each part is calculated. Then, the deviations of all key parts are aggregated to form a posture deviation index. ; S3: Based on Fuzzy matching degree calculation is performed to obtain the membership function values ​​of high matching, medium matching and low matching respectively. Based on the calculated membership function values ​​of high matching, medium matching and low matching, the measurement mode is judged in combination with the threshold. The measurement mode of the bone density scanner is switched and controlled according to the judgment result. S4: Output three-dimensional measurement results, including height data, weight data, and bone density data. Finally, upload the data to the database and generate a measurement reliability and posture adjustment suggestion report for reference.

[0010] Furthermore, in step S1 above, the key areas include the cervical spine, thoracic spine, lumbar spine, knee joint, and ankle joint, and the system has a built-in reference posture model: Based on the vectors of various parts of the human body in a standard posture, and combined with a reference posture model, posture data vectors are established. ; where represents the reference pose dataset, Indicates the first Reference acceleration vectors at key points Indicates the first Reference gravity direction vectors at key points This represents the currently collected pose dataset. Indicates the first The current acceleration vector of each joint. Indicates the first The current gravity direction vector of each joint.

[0011] Furthermore, in step S2 above, the formula for calculating the degree of posture deviation of each joint is as follows: Posture Deviation Index The formula is ;in, Indicates the first The angle of deviation of the posture of each joint. This represents the dot product of two gravitational direction vectors. Representing vectors The length of the mold, The posture deviation index is used to quantify the overall degree of posture deviation. Indicates the first The weighting coefficient of each joint, This indicates the total number of key components involved in the calculation.

[0012] Furthermore, in step S3 above, the specific process for calculating the fuzzy matching degree includes: Calculate the corresponding membership value by combining the membership functions with high matching degree, medium matching degree, and low matching degree; The membership function with high matching degree is expressed as: The system determines that the subject's posture meets the normal upright standard and triggers a high-matching measurement strategy; The membership function of the matching degree is expressed as follows: The system determines that the subject's posture is slightly off from the normal state and triggers the medium matching degree measurement strategy; The membership function for low matching degree is expressed as: The system determines that the subject's posture is severely deviated from the normal state, triggering a low-match measurement strategy.

[0013] Furthermore, in step S3 above, the measurement mode is determined by combining the calculated membership function values ​​of high matching, medium matching, and low matching with a threshold. like Switch to standard mode for bone density testing, i.e., switch the bone density scanner to hip region measurement mode, hip and lumbar spine dual region measurement, and multidimensional data measurement. For example, scanning the hip is sufficient; if... If the value is the maximum, then switch to enhanced mode detection, that is, the bone density scanner switches to a dual-region measurement mode for the hip and lumbar spine; for example, scanning both the hip and lumbar spine regions. If the value is the maximum, then switch to refined detection, that is, the bone density scanner switches to multi-dimensional data mode and records body posture data.

[0014] Furthermore, in step S4 above, the height data is obtained by acquiring spatial coordinate data from top to bottom through a laser ranging module, and calculated based on the spatial coordinate data; the weight data is obtained by integrating a pressure sensor array.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention achieves the effect of accurately quantifying the degree of deviation of body posture and adaptively selecting the best measurement strategy by setting a dynamic grading mechanism for body posture and measurement matching degree, which significantly improves the accuracy of height and weight measurement and the reliability of bone density data in bedridden state. 2. This invention, by setting a three-level threshold differential measurement and bone density scanning strategy triggered by the maximum membership degree of the body posture deviation index, can intelligently optimize resource allocation (scanning area) and perform targeted compensation calibration (such as dynamic calibration of height laser path and weight pressure weight allocation) according to the real-time body posture of the test subject, thereby improving the overall detection efficiency while ensuring data validity. 3. This invention integrates pressure sensing, laser ranging, body posture monitoring, and bone density scanning into a unified measurement process. It simultaneously measures height and weight and detects bone density while the subject remains in the same position, and outputs the results in a correlated manner. This reduces multiple steps such as subject movement, equipment switching, and manual data entry, significantly shortening the detection time, reducing operational complexity and data correlation error rate. It also provides a measurement report with body posture reliability annotations for medical personnel to review and refer to. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the process of this invention; Figure 2 This is a graph of the membership functions corresponding to the three posture deviation indices; Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0018] Example 1: The data-driven bone density scanner measurement mode switching system disclosed in this invention includes a measurement platform, a posture monitoring module, a data processing module, and a control module. The measurement platform integrates a pressure sensor array for pressure distribution data and weight data of various parts of the body. Laser ranging modules are set on both sides of the measurement platform to acquire spatial coordinate data from top to bottom and calculate height based on the spatial coordinate data. The posture monitoring module includes an IMU sensor group to capture spinal curvature data, limb bending angle data, and body position offset data in real time. The data processing module receives data from the pressure sensor array, the laser ranging module, and the IMU sensor group, and connects to the control module; The inertial measurement unit data is used to generate the posture deviation index SI based on the spinal curvature data, limb bending angle data, and body position offset data obtained by the IMU sensor group. The data processing module calculates the fuzzy matching degree based on the body posture deviation index SI, and obtains the membership function values ​​for high matching, medium matching and low matching respectively; The control module determines the measurement mode based on the calculated membership function values ​​of high matching, medium matching, and low matching, combined with a threshold. Based on the determination result, it switches the measurement mode of the bone density scanner. The measurement modes include hip region measurement, hip and lumbar spine dual region measurement, and multidimensional data measurement. The multidimensional data measurement includes multi-angle re-image, high-resolution small ROI scanning, and combined measurement with body posture compensation algorithm.

[0019] Specifically, abnormal posture will lead to distortion in bone mineral density (BMD) measurements; a 15° lumbar lateral tilt can cause a T-score deviation of up to 12%, and femoral neck BMD errors exceeding 10% during hip internal rotation. Based on this, the proposed solution requires calibration triggered by the SI index, with direct scanning for high-match scores and laser path compensation initiated for medium / low-match scores (correcting the hip scanning area based on IMU data). Fuzzy logic classification is commonly used in human posture assessment. By pre-collecting the SI distribution of a healthy control group, Gaussian and S-shaped functions are fitted as membership degrees for three intervals: "normal – mild – severe," a method frequently found in various human posture and ergonomics literature. Based on the principle of fuzzy logic assessment, this application uses posture numerical references to determine the reliability or degree of deviation of measurement results, thereby triggering corresponding correction strategies.

[0020] The process of determining body posture, from data acquisition to index generation, includes data collection, data analysis, quantitative judgment, and comprehensive evaluation. Data acquisition involves using IMU sensor arrays installed in key body parts (such as the spine, hip joints, and knee joints) to continuously capture raw motion data, including acceleration and angular velocity, in real time. Data analysis involves the system's internal processor processing this raw data, filtering out interference signals caused by breathing, slight shaking, etc., and ultimately extracting the core data that best represents the spatial orientation of each part, namely the gravity direction vector. Quantitative judgment involves comparing the real-time spatial orientation of each part with a pre-stored reference orientation under the standard ideal lying posture. Through mathematical calculation, the specific angle value of each part's deviation from the standard posture is obtained. Comprehensive evaluation considers that the deviation of different body parts has different degrees of influence on the final bone mineral density measurement results; for example, hip joint deviation has a greater impact than ankle joint deviation. The system assigns different weights to the deviation angle of each part, and finally summarizes all weighted deviation values ​​to obtain a comprehensive and unique numerical index, namely the posture deviation index. The SI value quantitatively reflects the degree to which the test subject's current overall posture deviates from the standard lying posture.

[0021] Body posture data indirectly ensures the reliability of the final formula input data by determining the measurement strategy and whether to enable the compensation algorithm. The impact process is as follows.

[0022] Triggering the decision-making mechanism: The calculated SI value is immediately fed into a fuzzy logic decision-maker. This decision-maker has three preset evaluation curves (high, medium, and low matching degree) to assess the degree to which the current SI value belongs to the three states of ideal, acceptable, or severely biased.

[0023] Adaptive measurement mode selection: Based on the above evaluation results, the system will automatically select three different measurement modes, which will directly affect the subsequent scanning and data processing workflow.

[0024] If the body posture is determined to be ideal (high matching degree), the system assumes that the current posture has no significant impact on the measurement results. Therefore, the standard scanning procedure is executed, and the subsequent bone density calculation will directly use the standard formula without any correction.

[0025] If the deviation is determined to be slight (medium match), the system assumes the current posture may introduce error and will activate an enhanced scanning mode. This may include expanding the scanning area (e.g., scanning the hip and lumbar spine simultaneously) to obtain more data. During calculation, the system may invoke a simple compensation algorithm to fine-tune the scanning area based on the offset angle, and then substitute it into the standard formula to calculate, thereby offsetting some of the deviation.

[0026] If a significant deviation (low match) is detected, the system considers the direct scan results to have low reliability and will activate a refined compensation mode. At this point, specific body posture data (the offset angles and directions of each body part) will be used as key parameters and input into an advanced compensation algorithm. This algorithm may dynamically adjust the operation of the measurement device (such as calibrating the laser path, reallocating the weights of the body weight pressure sensors), or guide the device to scan from multiple angles. Ultimately, the bone density result is calculated from the raw scan data after processing by this compensation algorithm with body posture data as input.

[0027] Some literature indicates that the T-score represents the difference between a patient's bone mineral density (BMD) and the peak bone mass of a normal young adult of the same sex, while the Z-score represents the difference between a patient's BMD and the average BMD of individuals of the same sex, age, and ethnicity. The Z-score can indicate the patient's position relative to their age-matched BMD. By changing the standard internal rotation position to a natural position, BMD values ​​increased. Semi-quantitative studies further found that the WHO classification score for the DXA (Decentrifugal Angiography) was higher in the natural position than in the internal rotation position, while the corresponding fracture risk was lower in the natural position. Therefore, body posture can affect the results of bone mineral density measurements.

[0028] References: [1] Xu Zhengyang, Yang Zhen, Zhou Shiqing, et al. Effect of hip DXA examination position on bone mineral density measurement [J]. Chinese Journal of Osteoporosis and Bone Mineral Diseases, 2022, 15(01): 19-23. The pressure sensors can reflect the position of the center of force and the shift of the body's center of gravity in real time, helping to determine whether the measurement error is caused by the subject's weight being too heavy on one side, and providing a basis for posture correction (e.g., leaning to the left indicates possible tilt). Weight data is based on the output force values ​​of each unit on the pressure sensor array. The total weight is Because different parts of the human body experience different forces, any unit in the array can simultaneously report the pressure distribution. At the same time, dynamic oscillations are removed through filtering to ensure the stability of weight measurement.

[0029] To overcome the abrupt changes and fluctuations caused by traditional "hard threshold" judgments in the critical range of body posture, this application introduces a fuzzy membership function mechanism. This mechanism maps the precisely quantified posture deviation index SI to the 0,1 interval to represent its membership degree to three fuzzy sets: "high matching degree," "medium matching degree," and "low matching degree." The membership function is characterized by: 1. Smooth transition, capable of continuous S-shaped or Gaussian transition between two states, avoiding abrupt boundary changes; 2. Adjustable parameters: The response sensitivity can be flexibly adjusted for different groups of people by using threshold and slope bandwidth parameters. 3. The principle of maximum membership degree: When the interval where SI is located falls into the intersection area of ​​multiple membership functions, the measurement strategy corresponding to the largest membership degree is selected by comparing the three membership degree values ​​to ensure the consistency and uniqueness of the decision.

[0030] The significance of introducing membership functions lies in addressing the instability of the measurement strategy. Traditional threshold-based methods, such as setting SI > 40 as a mismatch and SI ≤ 40 as a match, can lead to critical transitions. For example, when SI changes from 39 to 41, the matching result suddenly changes from a perfect match to a complete mismatch, resulting in measurement strategy instability. Furthermore, even slight measurement errors can trigger different branches, affecting repeatability and reliability. With membership functions, the SI response near the threshold changes from a discrete switch to a continuous transition, allowing for fine-tuning of the measurement strategy in the critical region and avoiding one-time switching. It also exhibits a degree of robustness to input noise or individual differences (slight jitter, differences in shooting angle, etc.).

[0031] The membership function with high matching degree is: ; in, express The membership value of the body posture deviation index at a high degree of match. The postural deviation index is calculated from IMU data. This represents the lower threshold for determining the matching degree. This indicates the upper limit threshold for determining the matching degree. express The slope parameter of the function.

[0032] Specifically, when the subject is in an upright position with a balanced center of gravity, the postural deviation index is... Much smaller than the threshold That is, when hour, When the value approaches 1, the posture is close to the ideal upright state. This is achieved by setting a preset threshold, such as... The system switches to a "high-matching measurement strategy," meaning that the bone density testing process does not require additional posture correction and directly enters the data acquisition stage.

[0033] when hour, When the value approaches 0, it means that the reference standard is significantly deviated from the body posture. At this time, it is necessary to guide the test subject to adjust the body posture, such as straightening the torso and limbs, and symmetrically distributing the center of gravity. This process can be assisted by the posture correction module.

[0034] exist ~ Between these points, the function exhibits a continuous and smooth decrease, displaying an S-shaped transition characteristic.

[0035] The transition slope of the control function is adjusted. The transition slope adjustment factor ranges from 2 to 8 and is used to adjust the steepness of the transition from 1 to 0. By controlling the slope, the response sensitivity can be adjusted according to age group, posture standard, etc. in practical applications.

[0036] The membership function of the matching degree is: ; in, express The membership value of the body posture deviation index in the degree of matching. Represents the central value, i.e. This indicates the center point of a slight deviation. This indicates the width of the Gaussian function. This indicates a deviation from the body posture index. This represents the lower threshold for determining the matching degree. This indicates the upper limit threshold for determining the matching degree.

[0037] Specifically, , Used for identifying "postural borderline intervals," meaning the subject may not yet have reached an abnormal state, but has deviated from the upright standard. When the value exceeds the preset threshold of 0.5, the system triggers the "medium matching strategy", which prompts the test subject to check their posture and correct it before collecting bone density data.

[0038] The membership function for low matching degree is: ; in, express The membership value of the body deviation index at low matching degree. This indicates a deviation from the body posture index. This represents the lower threshold for determining the matching degree. This indicates the upper limit threshold for determining the matching degree. express The slope parameter of the function.

[0039] Preferably, if any When a value simultaneously meets the conditions for triggering high, medium, and low matching degree strategies, or meets the conditions for any two of these strategies, the membership degree is selected using the maximum membership degree principle to eliminate the tie mechanism. If a rare case occurs where two membership degree values ​​are exactly the same, a priority can be set (high > medium > low) or a small parameter perturbation can be introduced to break the tie, ensuring that the decision is unique and controllable.

[0040] Specifically, when the subject is in an upright position with a balanced center of gravity, the postural deviation index is... Much smaller than the threshold That is, when hour, When the value approaches 0, the posture is close to the ideal upright state. hour, A value approaching 1 indicates a significant deviation in posture, requiring guidance for the subject to adjust their posture, such as straightening the torso and limbs, and symmetrically distributing the center of gravity. This process can be assisted by a posture correction module. By setting a preset threshold, such as... The system triggers a "low-match measurement strategy," which guides the subject to adjust their posture, such as reminding them to stand straight and distribute their weight symmetrically, or activates a postural correction reference module to ensure the reliability of bone mineral density measurements. ~ Between these two intervals, the function exhibits a continuous and smooth increase. Within this interval, the function's continuous and smooth increase indicates that its response to the input parameter (SI) is monotonic and asymptotic, without abrupt changes. This characteristic helps the system achieve a smooth transition in critical states, enhancing the control strategy's tolerance to individual differences and minor measurement errors.

[0041] It should be noted that low-matching functions Specifically designed for severely off-target scenarios, when hour, ,at this time This triggers a high-matching strategy. When hour, Each event independently triggers a low-matching strategy, which is mutually exclusive with the high-matching strategy. The two constitute a complete event group, covering... Global and with no defined overlap.

[0042] Regarding the establishment of membership functions, membership functions with high matching degrees smoothly decrease near the threshold, with the upper limit approaching 1. High membership degrees are achieved when the standard is met. T1 and T2 can intuitively represent the take-off and landing point intervals. Let T1 = P20, T2 = P80, and let k such that... (SI=P20)≈0.8 (through fitting or expert annotation). Membership functions with medium matching degree have symmetrical peaks, suitable for describing the optimal range and gradually weakening at both sides, with a peak value of 1. Membership functions with low matching degree smoothly rise near the threshold, with the lower limit approaching 0 and the upper limit approaching 1, indicating a higher membership degree as the fit deviates further. Similarly, taking T1=P20 and T2=P80, fitting individually results in... (SI=P80)≈0.8.

[0043] Preferably, the values ​​of thresholds T1 and T2 are obtained based on statistical analysis of postural data from a large group of healthy subjects in standard postures. Specifically, a sufficient number of IMU data points are collected from healthy subjects in an ideal supine position, the distribution of their Postural Deviation Index (SI) is calculated, and a specific percentile of this distribution is selected as the threshold benchmark. For example, the 20th percentile of the SI value in a healthy population is often set as T1, and the 80th percentile as T2. That is, T1 represents that 80% of individuals in the healthy population have an SI value below this level, while T2 means that only 20% of individuals have an SI value above this level, thus mathematically defining the objective intervals for normal, mild deviation, and significant deviation.

[0044] The system calculates a precise SI value for the current subject based on real-time acquired IMU data. This value is then substituted into three membership functions in parallel to calculate the membership values ​​for high-match, medium-match, and low-match fuzzy sets. Each membership value represents the degree to which the current body posture belongs to the corresponding category. The final matching level decision follows the maximum membership principle: the three calculated membership values ​​are compared, and the category corresponding to the largest value is selected as the final matching level for the current body posture, triggering the corresponding measurement mode. This mechanism ensures the clarity and automation of the decision-making process.

[0045] For example, if If the value is the largest, it is considered a high match, and the system switches to standard mode (e.g., scanning only the hip); if If the value is the highest, it is determined to be a medium match, and the system switches to enhanced mode (e.g., scanning both the hip and lumbar spine regions). If If the value is the largest, it is determined to be a low match, and the system switches to a refined mode (such as starting multi-angle reshoot and compensation algorithms).

[0046] Taking SI data from 100 healthy control groups as an example, by determining the threshold range, fitting slope parameter k, calculating bandwidth, and plotting and validating the functions, three function curves were plotted, and their distribution on SI=[0,100] was checked to see if it was reasonable (high matching is almost 1 in [0,23], medium matching is symmetrical in the [23,50] interval, and low matching is close to 1 in [50,100]). On the independent validation group, the maximum membership degree was used for discrimination, and compared with expert annotations, a consistency rate ≥90% was acceptable; otherwise, k was fine-tuned until the requirements were met.

[0047] Example 2, as Figure 1 As shown, based on Embodiment 1, this embodiment proposes an operation method for a data-driven bone densitometer measurement mode switching system, including the following steps: S1: Data acquisition and standard attitude reference modeling. The three-axis angular velocity and acceleration of key parts are acquired through the IMU sensor group. The corresponding attitude data vector is established according to the system's built-in reference attitude model. S2: Calculate the posture deviation index based on the posture data vector. First, the degree of posture deviation of each part is calculated. Then, the deviations of all key parts are aggregated to form a posture deviation index. ; S3: Based on Fuzzy matching degree calculation is performed to obtain the membership function values ​​of high matching, medium matching and low matching respectively. Based on the calculated membership function values ​​of high matching, medium matching and low matching, the measurement mode is judged in combination with the threshold. The measurement mode of the bone density scanner is switched and controlled according to the judgment result. S4: Output three-dimensional measurement results, including height data, weight data, and bone density data. Finally, upload the data to the database and generate a measurement reliability and posture adjustment suggestion report for reference.

[0048] In step S1 above, the key areas include the cervical spine, thoracic spine, lumbar spine, knee joint, and ankle joint. The system has a built-in reference posture model. Based on the vectors of various parts of the human body in a standard posture, and combined with a reference posture model, posture data vectors are established. ; where represents the reference pose dataset, Indicates the first Reference acceleration vectors at key points Indicates the first Reference gravity direction vectors at key points This represents the currently collected pose dataset. Indicates the first The current acceleration vector of each joint. Indicates the first The current gravity direction vector of each joint.

[0049] Specifically, IMU sensor arrays are installed at key locations in the cervical spine, thoracic spine, lumbar spine, knee joint, and ankle joint. After system startup, the IMUs acquire instantaneous triaxial acceleration vectors of each key joint at preset frequencies. and the three-axis angular velocity vector High-frequency noise is removed using a filtering algorithm.

[0050] The system pre-loads the gravity direction vector and static acceleration vector of each key point under the standard upright posture to form a reference posture dataset.

[0051] , in, For standard upright position The direction of gravity at key points This represents the static acceleration component.

[0052] Subject posture data The system constructs the acceleration vectors acquired in real time. with respect to the direction vector of gravity The current subject pose dataset is composed of... , By using online calibration or human body model mapping, the IMU coordinate system data is transformed into a unified body coordinate system, ensuring that the data of each key point can be compared with each other.

[0053] In step S2 above, the formula for calculating the degree of posture deviation of each joint is: Posture Deviation Index The formula is ;in, Indicates the first The angle of deviation of the posture of each joint. This represents the dot product of two gravitational direction vectors. Representing vectors The modulus, representing the posture deviation index, is used to quantify the overall degree of posture deviation. Indicates the first The weighting coefficient of each joint, This indicates the total number of key parts involved in the calculation, namely the cervical spine, thoracic spine, lumbar spine, knee joint, and ankle joint.

[0054] Among them, the direction vector of gravity The method for obtaining this data involves real-time acquisition of raw acceleration data by IMU sensors installed at various key locations, in order to extract the pure gravity direction vector. It needs to be processed through an embedded processor. Real-time attitude calculation is performed to separate the dynamic acceleration and static gravitational acceleration components. The reference gravity direction vector... This refers to the baseline data collected and stored in each node by the IMU sensor when the subject is in a standard lying position before the system leaves the factory or during the calibration process.

[0055] The weighting coefficient The determination is based on the weighting coefficient. This index, used to represent the contribution of different joints to the overall postural deviation index, is determined based on the importance of each joint in bone mineral density measurement and its sensitivity to postural errors, through clinical data analysis. For example, in hip bone mineral density measurement, the hip joint is typically assigned the highest weight. In lumbar spine measurements, the lumbar vertebral nodes have the highest weight. The weighting coefficients satisfy the normalization condition, i.e. Example as follows: Lumbar spine weight =0.3, Hip joint weight =0.4, knee joint weight =0.2, Ankle joint weight =0.1.

[0056] The postural deviation index is a weighted sum of the deviation angles of each key joint. The smaller the value, the closer the current posture is to the standard posture, and the more ideal the measurement conditions. The larger the value, the more serious the postural deviation, the greater the potential interference with bone mineral density measurement results, and the more necessary it is to trigger subsequent fuzzy decision-making and compensation mechanisms.

[0057] In an IMU, the direction of gravity is from acceleration. The IMU triaxial acceleration readings, separated from the components, include gravitational acceleration and non-gravitational acceleration of the body. Through low-pass filtering or attitude calculation, such as Kalman filtering, the sensor output can be... Decomposed into: ; Static acceleration components From the original acceleration Extracted from, while dynamic components It can be used to detect movement or shaking of the subject during the measurement process, thereby eliminating measurement errors caused by shaking or prompting a remeasurement.

[0058] Posture Deviation Index The generation steps are as follows: first, use an IMU to obtain the gravity direction vectors of each key joint (cervical spine, thoracic spine, lumbar spine, knee joint, and ankle joint). With reference gravity vector ; Calculate the first The deviation angle of the point, then by weight. polymerization.

[0059] Specifically, based on Calculate the overall deviation index, where For the first Key point weight coefficients The higher the value, the more serious the deviation of the test subject's overall standing posture from the standard. Since it is a continuous quantity, decision-making requires discretization and hierarchical classification. If used directly... Threshold, for example If the measurement strategy is switched or a retest is prompted, it will lead to a sudden change in the boundary. It's normal. An error will be displayed. To achieve a smooth transition in membership functions, when hour, , , The system prioritizes matching strategies to avoid the risk of rigid threshold decisions.

[0060] In step S3 above, the specific process for calculating the fuzzy matching degree includes: Calculate the corresponding membership value by combining the membership functions with high matching degree, medium matching degree, and low matching degree; The membership function with high matching degree is expressed as: High-fit membership functions provide a reference standard for whether a body posture is close to a normal state.

[0061] The membership function of the matching degree is expressed as follows: The matching degree membership function provides a reference standard for whether the body posture deviates slightly from the normal state.

[0062] The membership function for low matching degree is expressed as: Low-match membership functions provide a reference standard for whether the body posture deviates significantly from the normal state.

[0063] Specifically, when making decisions, the deviation of the current body posture from the index will be considered. Substituting the values ​​into the three functions respectively to calculate the membership degree, we get... , , These correspond to three fuzzy sets, respectively. When the posture deviates from the index... After substituting the three membership functions, the current measurement matching level (high matching degree / medium matching degree / low matching degree) is determined according to the principle of maximum membership degree. Based on the determined matching level, the corresponding measurement or correction strategy is triggered (e.g., high matching degree corresponds to a normal posture; medium matching degree indicates the highest membership degree of the fuzzy combination, corresponding to a slightly deviated posture; low matching degree indicates the highest membership degree of the fuzzy combination, corresponding to a severely deviated posture). It should be noted that human posture deviation is not a binary state of "completely normal" or "completely abnormal," but a continuous and gradual process. Traditional "hard threshold" classification is prone to "critical value jitter," leading to frequent misjudgments. Therefore, a membership function is introduced to calculate the degree to which it "belongs" to one of the three fuzzy sets. The set corresponding to the highest membership degree is the category that is "closest" to the current state. The basis for “the output value of the membership function corresponds to the state” is that each function is mapped to a state during the design process. The interval division, membership function, and parameters are all fitted based on the distribution of real samples, rather than being set out of thin air (assuming that based on the body shape and actual results of 100 test subjects, the value is set to a reasonable range by measuring the membership function).

[0064] Set two reference thresholds and And satisfy ,set up , , Gaussian bandwidth factor ,midpoint .when , This indicates that the posture characteristics are close to normal.

[0065] when (i.e., 30°) is the degree of matching membership. =1, with smooth decay on both sides.

[0066] when , This indicates a significant deviation in posture characteristics, requiring readjustment of body posture.

[0067] After calculating the membership degrees of the three, the level with the highest membership degree is taken as the recognition result of the current body state, and the corresponding bone density detection strategy is triggered.

[0068] like If the value is at its maximum, switch to the high-matching bone density detection mode; like If the maximum value is reached, switch to the medium-matching bone mineral density detection mode; like If the maximum value is reached, switch to the low-matching bone density detection mode.

[0069] In step S3 above, the measurement mode is determined by combining the calculated membership function values ​​for high matching, medium matching, and low matching with a threshold. like Switch to standard mode for bone density testing, i.e., switch the bone density scanner to hip region measurement mode, hip and lumbar spine dual region measurement, and multidimensional data measurement. For example, scanning the hip is sufficient; if... If the value is the maximum, then switch to enhanced mode detection, that is, the bone density scanner switches to a dual-region measurement mode for the hip and lumbar spine; for example, scanning both the hip and lumbar spine regions. If the value is at its maximum, a refined detection mode is switched, meaning the bone density scanner switches to multi-dimensional data mode and records body posture data. Height data is obtained by acquiring spatial coordinate data from top to bottom via a laser ranging module, and calculations are performed based on this spatial coordinate data. Weight data is obtained through an integrated pressure sensor array.

[0070] Specifically, in cases of low matching (severe posture deviation), even simple dual-region scans of the hip and lumbar spine can be affected by posture errors. In such cases, refined detection can include multi-angle re-scanning, high-resolution small ROI scanning, and posture compensation algorithms. Multi-angle re-scanning involves repeatedly scanning the same region from different incident angles to build a correction model; high-resolution small ROI scanning reduces the scanning area and uses a high-resolution mode to minimize posture deviation interference; and the posture compensation algorithm combines IMU data to perform orientation correction and interpolation adjustments on BMD readings.

[0071] Based on literature reports, DEXA errors can exceed 3%–5% under extreme postures, while multi-angle or multi-mode scanning can reduce the error to within 1%. This application adopts this principle and incorporates upper-level compensation into the detection strategy to ensure measurement reliability.

[0072] Height is calculated by scanning the three-dimensional coordinates from the top of the head to the soles of the feet using laser ranging modules on both sides of the platform, while weight is acquired in real time by a hidden pressure sensor array and then filtered and averaged. The real-time measured height... ,weight Posture Deviation Index and its matching grade and corresponding bone mineral density value The data is integrated into a complete record and then uploaded to the subject's electronic record system via a local or cloud interface.

[0073] If the matching level is low, the generated parameter results are used to provide posture correction suggestions and retest reminders. If the matching level is medium, based on the numerical results, reference suggestions are provided for slight deviations and fine-tuning of posture. If the matching level is high, the guidance result corresponding to the parameter range provided is that the measurement data is normal and no additional intervention is required.

[0074] Example 3, based on Example 1, proposes a membership function triggering strategy principle based on actual application scenarios.

[0075] like Figure 2 As shown, the high-match membership function curve takes a value close to 1 when SI is small, indicating that the attitude basically meets the standard. As SI approaches T1, the function begins to decrease, and after T2, it rapidly approaches 0. The medium-match membership function curve... The distribution is centrally symmetrical, and it decreases rapidly at both ends of T1 and T2, indicating that the deviation of the attitude characteristics is small. The low-match membership function curve (with a value close to 0 when SI is small, and a rapid increase in function value as SI approaches T2, then tending towards 1, indicating a large deviation in posture characteristics). It should be noted that the three curves in the figure do not represent three different independent variables, but rather three curves obtained by mapping the same independent variable SI through three different membership functions. The independent variable is only the posture deviation index SI, but we need to evaluate the membership degree of SI in the three fuzzy sets of high-match, medium-match, and low-match. Plotting the three functions on the same coordinate system yields three curves. The final decision (e.g., choosing a measurement strategy) is not based on which curve exists, but on comparing the values ​​of the three curves at that SI, and determining the association based on the principle of maximum membership degree. For example, if the horizontal axis represents age and the vertical axis represents physical fitness, then a single "high-match" curve can express the relationship between age and physical fitness. However, to visually represent the range of "physical fitness" corresponding to different "ages," function curves with medium and low matching degrees are set. The three curves are interchangeable.

[0076] This embodiment uses a practical application scenario for scoliosis assessment as an example. Range 0–100, =25, =45. Substituting the current posture deviation index SI into the high, medium, and low matching degree membership functions, we obtain... , , The current measurement matching level (high matching, medium matching, or low matching) is determined based on the principle of maximum membership, which is used to control subsequent measurement procedures and correction strategies.

[0077] when hour, This indicates that the current measurement conditions are highly consistent with the system's built-in standard vertical geometry, the measurement conditions are determined to have a high degree of matching, and the preset measurement reliability requirements are met. The system can directly input and report the measurement results. place, At this point, the curve just crosses 0.5, indicating that it is at the critical position from high matching to the transition zone; when , Below 0.1, nearby, This indicates that the measurement conditions have deviated from the standard range, requiring further correction or alerts.

[0078] Preferred, in ,and If the system determines that the result meets the high matching threshold, it can skip the additional posture correction steps and directly input the measurement results; when When the score is below 0.9, the matching level can be switched to medium or low matching level, thereby triggering the corresponding strategy.

[0079] The peak value of 1.00 appeared at ,exist and At both ends, , and both ends and At the same time, a fuzzy interval is formed when the value is approximately 0.5, which meets the medium matching degree threshold. or hour, This indicates that the current match does not fall within the medium matching range, and the system will assign the matching level as high or low based on the maximum membership degree.

[0080] then, This indicates that the measurement conditions are close to the standard upright state, and the results can be directly entered; place, At this point, the curve enters an upward inflection point; when , = greater than 0.9, curve If the readings are near and tend to stabilize, it indicates that the measurement conditions deviate significantly from the standard, and the measurement results may not meet the reliability requirements. Compensation or retesting measures should be implemented.

[0081] Preferred, hour, The system will record low matching scores and perform one or more operations, such as prompting the operator or subject to adjust their posture, pausing the current scan and prompting a retest, enabling the laser path compensation algorithm, or triggering multi-angle retakes to reduce measurement errors caused by posture.

[0082] when ,correspond The system identifies this as a medium-match warning zone, issues a posture adjustment prompt, and suggests a short-term adjustment followed by a retest; if If the value is the maximum, the system enters standard mode and records it as a high match.

[0083] Of course, the present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the appended claims.

Claims

1. A data-driven bone densitometer measurement mode switching system, comprising a measurement platform, a posture monitoring module, a data processing module, and a control module, characterized in that: The measurement platform integrates a pressure sensor array for acquiring pressure distribution data and weight data of various parts of the body. Laser ranging modules are set on both sides of the measurement platform to acquire spatial coordinate data from top to bottom and calculate height based on the spatial coordinate data. The posture monitoring module includes an IMU sensor group for capturing spinal curvature data, limb bending angle data and body position offset data in real time. The data processing module receives data from the pressure sensor array, the laser ranging module, and the IMU sensor group, and connects to the control module; The inertial measurement unit data is used to generate the posture deviation index SI based on the spinal curvature data, limb bending angle data, and body position offset data obtained by the IMU sensor group. The data processing module calculates the fuzzy matching degree based on the body posture deviation index SI, and obtains the membership functions with high matching degree, medium matching degree, and low matching degree respectively. The control module determines the measurement mode based on the calculated membership functions with high matching degree, medium matching degree, and low matching degree, combined with a threshold. Based on the determination result, it switches the measurement mode of the bone density scanner. The measurement modes include hip region measurement, hip and lumbar spine dual region measurement, and multidimensional data measurement. The multidimensional data measurement includes multi-angle re-image, high-resolution small ROI scanning, and combined measurement with body posture compensation algorithm.

2. The data-driven bone densitometer measurement mode switching system according to claim 1, characterized in that, The membership function with high matching degree is: ； in, express The membership value of the body posture deviation index at a high degree of match. The postural deviation index is calculated from IMU data. This represents the lower threshold for determining the matching degree. This indicates the upper limit threshold for determining the matching degree. express The slope parameter of the function.

3. The data-driven bone densitometer measurement mode switching system according to claim 1, characterized in that, The membership function of the matching degree is: ； in, express The membership value of the body posture deviation index in the degree of matching. Represents the central value, i.e. This indicates the center point of a slight deviation. This indicates the width of the Gaussian function. This indicates a deviation from the body posture index. This represents the lower threshold for determining the matching degree. This indicates the upper limit threshold for determining the matching degree.

4. The data-driven bone densitometer measurement mode switching system according to claim 1, characterized in that, The membership function for low matching degree is: ； in, express The membership value of the body deviation index at low matching degree. This indicates a deviation from the body posture index. This represents the lower threshold for determining the matching degree. This indicates the upper limit threshold for determining the matching degree. express The slope parameter of the function.

5. A method of operating the data-driven bone densitometer measurement mode switching system according to any one of claims 1-4, characterized in that, Includes the following steps: S1: Data acquisition and standard attitude reference modeling. The three-axis angular velocity and acceleration of key parts are acquired through the IMU sensor group. The corresponding attitude data vector is established according to the system's built-in reference attitude model. S2: Calculate the posture deviation index based on the posture data vector. First, the degree of posture deviation of each part is calculated. Then, the deviations of all key parts are aggregated to form a posture deviation index. ; S3: Based on Fuzzy matching degree calculation is performed to obtain the membership function values ​​of high matching, medium matching and low matching respectively. Based on the calculated membership function values ​​of high matching, medium matching and low matching, the measurement mode is judged in combination with the threshold. The measurement mode of the bone density scanner is switched and controlled according to the judgment result. S4: Output three-dimensional measurement results, including height data, weight data, and bone density data.

6. The operating method according to claim 5, characterized in that, In step S1 above, the key areas include the cervical spine, thoracic spine, lumbar spine, knee joint, and ankle joint. The system has a built-in reference posture model. Based on the vectors of various parts of the human body in a standard posture, and combined with a reference posture model, posture data vectors are established. ;in, Represents the reference pose dataset. Indicates the first Reference acceleration vectors at key points Indicates the first Reference gravity direction vectors at key points This represents the actual pose dataset currently being collected. Indicates the first The current acceleration vector of each joint. Indicates the first The current gravity direction vector of each joint.

7. The operating method according to claim 6, characterized in that, In step S2 above, the formula for calculating the degree of posture deviation of each joint is: Posture Deviation Index The formula is ;in, Indicates the first The angle of deviation of the posture of each joint. This represents the dot product of two gravitational direction vectors. Representing vectors The length of the mold, The posture deviation index is used to quantify the overall degree of posture deviation. Indicates the first The weighting coefficient of each joint, This indicates the total number of key components involved in the calculation.

8. The operating method according to claim 5, characterized in that, In step S3 above, the specific process for calculating the fuzzy matching degree includes: Calculate the corresponding membership value by combining the membership functions with high matching degree, medium matching degree, and low matching degree; The membership function with high matching degree is expressed as: ； The membership function of the matching degree is expressed as follows: ； The membership function for low matching degree is expressed as: 。 9. The operating method according to claim 8, characterized in that, In step S3 above, the measurement mode is determined by combining the calculated membership functions with high, medium, and low matching degrees with a threshold. like Switch the bone density scanner to hip region measurement mode, hip and lumbar spine dual region measurement, and multidimensional data measurement; simply scan the hip. To reach the maximum value, the bone density scanner switched to a dual-region measurement mode for the hip and lumbar spine. At the maximum value, the bone density scanner switches to multidimensional data mode.

10. The operating method according to claim 9, characterized in that, In step S4 above, the height data is obtained by acquiring spatial coordinate data from top to bottom through a laser ranging module, and calculated based on the spatial coordinate data. The weight data is obtained by integrating a pressure sensor array.

Images