Method for calculating and correcting errors of lymphedema limb volume

Abstract

This invention relates to a method for calculating and correcting the volume of a limb with lymphedema. The method involves acquiring the outer contour data of the limb with lymphedema under a first and second standard measurement position, and dividing the limb into segments along its length with equal intervals to obtain two sets of corresponding volume segments. The volume of each corresponding segment is calculated, and the volume difference for the same segment number is determined. The volume segment whose absolute value of the volume difference is greater than a preset threshold is identified as the target volume segment. Based on the correspondence of the local volume distribution of the target volume segment in the two sets of volume segments, the volume difference is determined to be either a solid increment or a positional drift increment. Using the volume of each volume segment under the first standard measurement position as a benchmark, the solid increment is retained in the target volume segment, and the positional drift increment is corrected. The volumes of each volume segment after correction are summed to obtain the corrected volume of the limb with lymphedema.

Description

Technical Field

[0001] This invention relates to the field of medical measurement and data processing technology, specifically to a method for calculating and correcting the error in the volume of limbs affected by lymphedema. Background Technology

[0002] Current technologies, such as the IoT-based data management system and method for measuring the volume of lymphedema limbs (publication number CN115732068A), primarily rely on acquiring three-dimensional graphic volume data through infrared scanning equipment. After simple calibration, this data serves as the measurement result, and the overall volume change is compared and analyzed across different treatment cycles. However, this method still has significant shortcomings in terms of volume calculation accuracy control and error source differentiation. From the volume acquisition perspective, this document only establishes a volume calibration model by recording the number of scans and scanning speed, and makes overall corrections to the calculation results from 3DMAX software. Its calibration approach is essentially error compensation at the equipment level, without structural analysis of the morphological changes in the patient's limb under different measurement positions, nor does it consider the spatial distribution shift caused by differences in posture, limb placement angles, or local tissue pressure within the same measurement cycle. In reality, lymphedema limbs exhibit significant local volume distribution shifts under different positions. If only the overall volume value or the difference between simple adjacent cycles is compared, it is easy to misjudge the false volume fluctuations caused by positional changes as real changes in the condition, thus affecting the accuracy of subsequent treatment assessments.

[0003] This document only judges the increase or decrease in volume based on the overall volume data at the beginning and end of the measurement cycle during the volume change analysis, and uses the volume increase or decrease ratio as the statistical basis for treatment effectiveness. It lacks segmental modeling and spatial correspondence analysis along the length of the limb, and cannot identify the specific distribution characteristics of volume changes in different regions of the limb, nor can it distinguish between the actual aggravation of local edema and the cross-sectional misalignment caused by overall posture adjustment. Therefore, in actual clinical application, there may be a risk of misjudgment due to body position drift. Although the document proposes a formula for calculating the volume calibration value, the formula mainly averages the influence of the number of scans and the scan speed on the spectral signal. Its calibration logic is a global linear correction, and it does not introduce spatial structured processing mechanisms such as segmental comparison, judgment of local peak positions, or unified analysis of continuous abnormal areas. Therefore, it is difficult to ensure the spatial consistency and repeatability of volume data when facing complex morphological changes. The core innovation of this document lies in calculating the first effective value, the second effective value, and the comprehensive effective value of treatment based on historical data, and triggering early warnings through continuous anomalies. Its essence belongs to a data management and decision support system, rather than an algorithm optimization for the accuracy of volume measurement itself. Therefore, the reliability of its volume data as basic input data directly affects the subsequent effective value calculation results. Once the basic volume data has a systematic deviation due to differences in body position, all treatment effect evaluation indicators constructed based on this data may be subject to cascading errors. Summary of the Invention

[0004] The purpose of this invention is to provide a method for calculating and correcting the error of limbs with lymphedema, thereby addressing some of the drawbacks and shortcomings pointed out in the background art.

[0005] The present invention adopts the following technical solution to solve the above-mentioned technical problems: a method for calculating and correcting the volume of a lymphedema limb, comprising: collecting the outer contour data of the lymphedema limb to be measured under the first standard measurement position and the second standard measurement position respectively, and dividing it into segments along the length of the limb at the same segment interval to obtain two sets of corresponding volume segments.

[0006] Calculate the volume of each corresponding volume segment and determine the volume difference for the same segment number; determine the volume segment whose absolute value of the volume difference is greater than a preset threshold as the target volume segment; based on the correspondence of the local volume distribution of the target volume segment in the two sets of volume segments, determine whether the volume difference is the entity increment or the body position drift increment.

[0007] Using the volume of each volume segment under the first standard measurement position as a benchmark, the entity increment is retained in the target volume segment, and the positional drift increment is corrected by assignment; the volumes of each volume segment after assignment correction are accumulated to obtain the corrected volume of the lymphedema limb to be tested.

[0008] Further, determining whether the volume difference is an entity increment or a body position drift increment includes: in the two sets of volume segments, forming a first local determination interval and a second local determination interval with the target volume segment and adjacent segments respectively, determining the representative segment number corresponding to the largest segment volume in each local determination interval; when there are multiple largest segment volumes, taking the segment number with the smallest number as the representative segment number; when the two representative segment numbers are the same, determining that the volume difference is an entity increment; when the two representative segment numbers are different, determining that the volume difference is a body position drift increment.

[0009] Further, the positional drift increment is assigned and corrected, including: determining the offset direction of the positional drift increment based on the representative segment number of the first local determination interval and the representative segment number of the second local determination interval; allocating the positional drift increment to volume segments adjacent to the target volume segment according to the offset direction; when allocating to multiple adjacent volume segments, the allocation is based on the proportion of the absolute value of the volume difference between each adjacent volume segment to the sum of the absolute values ​​of the volume differences between the multiple adjacent volume segments.

[0010] Further, the absolute values ​​of the volume differences are sorted in ascending order; the first and last volume differences are removed after sorting; the remaining volume differences are averaged to obtain the average volume difference; the product of the average volume difference and a preset multiple is determined as the preset threshold.

[0011] Furthermore, the local determination interval is composed of the target volume segment and its two adjacent volume segments before and after it; when the target volume segment is located at the first or last segment of the volume segment sequence, the local determination interval is composed of the target volume segment and two adjacent volume segments on the same side.

[0012] Furthermore, when there are two or more consecutive target volume segments, the consecutive target volume segments are used as consecutive determination segments, and the adjacent volume segments on both sides of the consecutive determination segment together with the consecutive determination segment constitute an extended local determination interval; based on the representative segment number in the extended local determination interval, the volume difference corresponding to the consecutive target volume segments is uniformly determined as entity increment or body position drift increment.

[0013] Furthermore, the offset direction of the body position drift increment is determined by comparing the representative segment numbers of the first local determination interval and the second local determination interval; when the representative segment number of the second local determination interval is greater than the representative segment number of the first local determination interval, the offset direction is determined to be the direction of increasing segment number; when the representative segment number of the second local determination interval is less than the representative segment number of the first local determination interval, the offset direction is determined to be the direction of decreasing segment number.

[0014] Furthermore, when the target volume segment is located at the first or last segment, the body position drift increment is allocated to the volume segment adjacent to the target volume segment on the same side; when the target volume segment is located within a continuous target volume segment, the body position drift increment is allocated to the next volume segment located outside the continuous target volume segment along the offset direction.

[0015] Furthermore, when the body position drift increment is distributed to multiple adjacent volume segments, it is distributed sequentially according to the offset direction, and the distribution amount of each volume segment does not exceed the absolute value of the corresponding volume difference.

[0016] Furthermore, when there are multiple adjacent volume segments available for allocation in the offset direction, and the absolute value of the volume difference between the multiple adjacent volume segments is equal, the body position drift increment is preferentially allocated to the volume segment with the smallest distance from the target volume segment; when the multiple adjacent volume segments are at the same distance from the target volume segment, the body position drift increment is allocated to the volume segment in the direction of increasing segment number.

[0017] The beneficial effects of this invention are as follows: By collecting outer contour data under both a first and second standard measurement position, and constructing corresponding volume segments with the same segment spacing, the volume data under different positions have a strict spatial correspondence. Based on this, by calculating the volume difference between segments with the same sequence number and combining it with a statistical method based on sorting and removing extreme values ​​to determine a preset threshold, abnormal change areas are identified from the overall fluctuation level, effectively avoiding interference from individual extreme errors on the judgment results. Furthermore, by constructing local judgment intervals and extracting representative segment numbers, the consistency or offset of the peak position of the local volume distribution is used to distinguish between the entity increment and the positional drift increment, enabling accurate differentiation between real tissue volume changes and spatial misalignment caused by position, thereby improving the authenticity and reliability of the volume measurement results.

[0018] A correction mechanism based on a combination of offset direction determination, orderly allocation of adjacent segments, and proportional allocation is proposed to address the positional drift increment. Clear definitions are made regarding the allocation priorities for boundary segments, continuous target segments, and equal-difference cases, ensuring the correction process is deterministic, continuous, and repeatable. By retaining the entity increment based on the first standard measurement position and reasonably transferring the drift increment, the corrected volume, free from the influence of positional errors, is ultimately obtained. Attached Figure Description

[0019] Figure 1 This is a logic diagram for judging and correcting errors in the volume of limbs affected by lymphedema, as presented in this invention.

[0020] Figure 2 This is a comparison diagram of the segmental volume under two standard body positions in Embodiment 1 of the present invention.

[0021] Figure 3 This is a schematic diagram of the ΔV clipping mean and threshold screening in Embodiment 1 of the present invention.

[0022] Figure 4 This is a comparison and misalignment determination diagram of the segment numbers representing the local determination interval in Embodiment 1 of the present invention.

[0023] Figure 5 This is a schematic diagram of the sequential allocation of the drift increment of the continuous target segment and the upper limit truncation in Embodiment 2 of the present invention.

[0024] Figure 6 This is a schematic diagram showing the proportional allocation of multiple adjacencies of boundary targets and the unconstrained solution satisfying box constraints in Embodiment 2 of the present invention.

[0025] Figure 7 This is a schematic diagram illustrating the changes in indicators before and after the selection and correction of parallel rule candidates in Embodiment 2 of the present invention. Detailed Implementation

[0026] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0027] Combined with appendix Figure 1 The present invention discloses a method for calculating and correcting the volume of limbs with lymphedema. The first standard measurement position is a pre-set baseline position used as a reference for volume calculation and subsequent error correction. The second standard measurement position is another stable position set while maintaining consistent measurement conditions, used to identify volume drift errors caused by changes in position.

[0028] After setting the body position, the external contour data of the limb with lymphedema to be measured was collected under both the first and second standard measurement positions. The external contour data can be acquired using a 3D scanning device, structured light imaging equipment, laser contour scanning equipment, or a multi-camera visual reconstruction system to form 3D point cloud data or surface mesh data reflecting the true shape of the limb. During the acquisition process, the scanning path was kept to cover the entire length of the limb to ensure continuous and complete data and avoid missing areas. The acquired raw data underwent noise reduction and smoothing to improve the accuracy of subsequent volume calculations.

[0029] The limb length direction is defined from the proximal to the distal end of the limb, specifically determined based on anatomical landmarks to ensure consistency in the segmentation direction under both body positions. The segment spacing is a pre-set fixed value, which remains identical under both the first and second standard measurement positions, ensuring spatial consistency between the two sets of data. Along the length direction, starting from the same reference point, segments are sequentially taken according to the fixed segment spacing, dividing the entire limb contour data into multiple continuous volumetric segments. Each volumetric segment corresponds to an interval along the limb length direction. Within this interval, a closed cross-section is constructed based on the contour data, and independent segmental volumetric units are formed by combining the spatial positional relationships of adjacent cross-sections.

[0030] After obtaining the two sets of segment volumes, a one-to-one correspondence was established according to the segment number, and the difference in volume was calculated for the same segment number under the two standard measurement positions. The volume difference is the result of subtracting the segment volume under the first standard measurement position from the segment volume under the second standard measurement position.

[0031] To identify areas of abnormal variation, the absolute value of each volume difference is taken and compared with a preset threshold. The preset threshold is a discrimination standard pre-set based on statistical laws or empirical parameters, used to distinguish between normal fluctuations and abnormal deviations. When the absolute value of the volume difference of a certain segment is greater than the preset threshold, the volume change of that segment is considered to exceed the normal measurement error range, and that segment is determined as the target volume segment.

[0032] By comparing the relative size patterns and peak position relationships of the volumes of each segment within two sets of local intervals, the study analyzes whether the target volume segment maintains the same local volume distribution characteristics under different body positions. If the peak positions of the volume distribution within the two sets of local intervals remain consistent, it indicates that the target volume segment corresponds to the same actual anatomical location under different body positions, and its volume change originates from changes in the actual tissue volume. Therefore, the volume difference is determined to be the solid volume increment. If the peak positions of the volume distribution within the two sets of local intervals shift, it indicates that changes in body position cause a redistribution of volume between adjacent segments. The volume change of the target volume segment mainly originates from spatial positional drift rather than actual volume change. Therefore, the volume difference is determined to be the positional drift increment.

[0033] The volumes of each segment under the first standard measurement position are used as the baseline data for the overall calculation. The first standard measurement position is a pre-set reference position, and its volume data has high stability and repeatability. Therefore, the volume segment sequence formed under this position is used as the basis for the correction process. For the target volume segment determined to be an increase in volume, it is assumed that its volume difference originates from the actual volume change of lymphedema tissue. Therefore, the increase in volume is directly superimposed on the corresponding segment.

[0034] The offset direction is determined based on the difference in the representative segment number determined by the local judgment interval, and the body position drift increment is allocated to adjacent volume segments along the offset direction. During the allocation process, the principle of segment-by-segment correction is followed, prioritizing allocation to segments adjacent to the target volume segment and conforming to the offset direction, and controlling the allocation amount of each segment to not exceed the absolute value of its corresponding volume difference to prevent the generation of new abnormal differences.

[0035] After preserving the solid increments and correcting for positional drift increments for all target volume segments, a corrected sequence of volume segments is formed. Each segment volume in this sequence is based on the volume measured in the first standard position and updated in conjunction with solid increment corrections and positional drift adjustments. The corrected volumes of all volume segments are then summed sequentially according to their segment numbers to obtain the overall corrected volume of the limb with lymphedema.

[0036] After determining the target volume segment, a local volume distribution analysis is performed to determine whether the volume difference belongs to the entity increment or the positional drift increment. In the volume segment sequences corresponding to the first and second standard measurement positions, adjacent segments are selected to form local decision intervals, centered on the target volume segment with the same segment number. Under the first standard measurement position, the target volume segment and its adjacent segments form the first local decision interval; under the second standard measurement position, the target volume segment with the corresponding number and its adjacent segments form the second local decision interval.

[0037] The segment number corresponding to the largest segment volume is determined as the representative segment number of the local decision interval. The representative segment number reflects the peak position of the volume distribution within the local interval and is used to characterize the degree of volume concentration and spatial distribution characteristics within the interval. When there are two or more segment volumes with equal values ​​within a local decision interval, to ensure the uniqueness and stability of the decision result, the segment with the smallest segment number is taken as the representative segment number of the local decision interval.

[0038] When the two representative segment numbers are the same, it indicates that the peak position of the volume distribution within this local interval remains consistent under both standard measurement positions. The target volume segment corresponds to the same actual anatomical location under different positions, and the volume difference originates from changes in the actual tissue volume. Therefore, the volume difference is determined to be the solid volume increment. When the two representative segment numbers are different, it indicates that the peak position of the local volume distribution shifts under different positions. The volume change manifests as a spatial transfer between adjacent segments rather than a change in the actual tissue volume. Therefore, the volume difference is determined to be the positional drift increment.

[0039] The absolute values ​​of the volume differences obtained for each segment are taken to eliminate the influence of positive and negative directions on the statistical results, ensuring that each value reflects the magnitude of volume change. Subsequently, all the absolute values ​​of volume differences are sorted according to their numerical magnitude, arranged in ascending order, thus forming an ordered sequence of differences.

[0040] The smallest volume difference at the top and the largest volume difference at the bottom of the sorted results are removed. Removing these two values ​​effectively reduces the impact of individual measurement errors or localized fluctuations on the threshold calculation, improving the stability and representativeness of the statistical results. After removing the first and last differences, the remaining volume differences are calculated as an arithmetic mean. The sum of these remaining differences is then divided by their respective numbers to obtain the average volume difference. The average volume difference reflects the general fluctuation level of volume change under the current measurement conditions.

[0041] After obtaining the average volume difference, it is multiplied by a pre-set factor to calculate a preset threshold for anomaly screening. The factor is set according to the actual measurement accuracy requirements and clinical application scenarios to adjust the screening sensitivity. When the factor is larger, the preset threshold is increased accordingly, and the screening results are more conservative; when the factor is smaller, the preset threshold is decreased accordingly, and the screening results are more sensitive.

[0042] For a target volume segment located in the middle of the volume segment sequence, the local decision interval is formed by the target volume segment and its adjacent preceding and following volume segments in the segment sequence. Centered on the target volume segment, the preceding and following segments are included in the analysis range, thus forming a local decision interval containing three consecutive segments. By selecting adjacent segments on both sides of the target volume segment, the changing trend of the local volume distribution can be reflected within a smaller range, which is helpful in identifying the location of the volume distribution peak and whether it has shifted.

[0043] When the target volume segment is at the beginning of the volume segment sequence, since it has no preceding adjacent segment, to ensure that the local decision interval still has sufficient data support, the local decision interval consists of the target volume segment and the two consecutive volume segments following it. In this case, the local decision interval is composed of the target volume segment and its two adjacent segments on the same side, maintaining the number of segments to ensure consistency in the length of the decision interval. When the target volume segment is at the end of the volume segment sequence, since it has no following adjacent segment, the local decision interval consists of the target volume segment and the two consecutive volume segments preceding it, similarly forming a local decision interval containing three consecutive segments.

[0044] When two or more target volume segments are found to be adjacent to each other in the segment sequence after filtering by a preset threshold, each target volume segment is no longer judged independently. Instead, the multiple consecutive target volume segments are regarded as a single continuous judgment segment. A continuous judgment segment consists of multiple volume segments that are consecutive in segment number and all satisfy the condition that the absolute value of their volume difference is greater than a preset threshold. This continuous judgment segment reflects a relatively concentrated or continuous abnormal volume change within adjacent intervals.

[0045] An adjacent segment is selected before the starting segment of the continuous decision segment and after the ending segment of the continuous decision segment, so that the extended local decision interval spatially covers the continuous anomaly region and the transition regions on both sides, thus fully reflecting the overall shape of the local volume distribution. When the continuous decision segment is located at the beginning or end of the segment sequence, only the adjacent segment on one side where it exists is selected to form the extended local decision interval.

[0046] After constructing the extended local judgment intervals for both body positions, the volume values ​​of each volume segment within each extended local judgment interval are statistically analyzed. The segment with the largest volume within each interval is then identified, and its corresponding segment number is determined as the representative segment number for that extended local judgment interval. Subsequently, the representative segment numbers obtained for the two body positions are compared. When the two representative segment numbers are the same, it indicates that the peak position of the volume distribution remains consistent within the continuous abnormal region and its adjacent regions. The volume difference corresponding to the continuous target volume segments originates from changes in the actual tissue volume. Therefore, the volume difference of all target volume segments within this continuous judgment interval is uniformly determined as the entity increment. When the two representative segment numbers are different, it indicates that the peak value of the local volume distribution has shifted overall under different body positions. The continuous abnormal region is essentially a spatial drift phenomenon caused by changes in body position. Therefore, the volume difference of all target volume segments within this continuous judgment interval is uniformly determined as the body position drift increment.

[0047] The representative segment number determined by the first local judgment interval corresponding to the first standard measurement position and the representative segment number determined by the second local judgment interval corresponding to the second standard measurement position are compared. If the representative segment number of the second local judgment interval is greater than the representative segment number of the first local judgment interval, it indicates that the peak value of the local volume distribution under the second standard measurement position has shifted along the direction of increasing segment number. Therefore, the direction of the positional drift increment is determined to be the direction of increasing segment number. If the representative segment number of the second local judgment interval is less than the representative segment number of the first local judgment interval, it indicates that the peak value of the local volume distribution has shifted along the direction of decreasing segment number. Therefore, the direction of the positional drift increment is determined to be the direction of decreasing segment number.

[0048] After determining the offset direction, the positional drift increment is not retained in the original target volume segment. Instead, it is allocated to volume segments adjacent to the target volume segment and located along the offset direction. By reassigning the volume increment along the actual movement direction of the volume distribution, it better reflects the spatial correspondence of real anatomical structures in different positions. When the positional drift increment is large, or when multiple adjacent volume segments exist along the offset direction, the positional drift increment can be allocated to multiple adjacent volume segments. The sum of the absolute values ​​of the volume differences between adjacent volume segments along the offset direction is calculated. Then, the proportion of the absolute value of the volume difference between each adjacent volume segment in the sum is determined, and the positional drift increment is allocated to the corresponding segment according to this proportion.

[0049] After constructing the local decision intervals and determining the representative segment numbers for the first and second local decision intervals, the two representative segment numbers are directly compared. The first local decision interval corresponds to the local volume distribution under the first standard measurement position, and the second local decision interval corresponds to the local volume distribution under the second standard measurement position. The representative segment number reflects the spatial location of the volume distribution peak within each local interval; therefore, by comparing the magnitude of the two representative segment numbers, the direction of movement of the local volume distribution between different positions can be determined.

[0050] When the representative segment number of the second local judgment interval is greater than that of the representative segment number of the first local judgment interval, it indicates that under the second standard measurement position, the peak value of the volume distribution has shifted relative to the first standard measurement position along the direction of increasing segment number. Since the segment numbers are arranged sequentially according to the limb length direction, and the direction of increasing number corresponds to a fixed spatial direction along the limb length direction, the offset direction of the positional drift increment can be determined as the direction of increasing segment number. In this case, during the subsequent attribution correction process, the positional drift increment should be allocated along the direction of increasing segment number.

[0051] When the representative segment number of the second local judgment interval is less than that of the representative segment number of the first local judgment interval, it indicates that under the second standard measurement position, the peak value of the volume distribution has shifted relative to the first standard measurement position along the direction of decreasing segment number. Therefore, the offset direction of the positional drift increment can be determined as the direction of decreasing segment number. In subsequent attribution correction, the positional drift increment should be allocated along the direction of decreasing segment number.

[0052] After determining the offset direction of the body position drift increment, it is also necessary to consider the positional relationship of the target volume segment within the overall segment sequence to limit the specific allocation of the body position drift increment. When the target volume segment is located at the beginning of the volume segment sequence, since it does not have any adjacent segments in the decreasing order of the segment sequence, regardless of how the offset direction is determined, the body position drift increment can only be allocated to the adjacent segment on one side, that is, the volume segment adjacent to the target volume segment on the same side. This adjacent volume segment on the same side is the segment that is directly adjacent to the target volume segment in terms of segment number and is located in the increasing order of the segment number.

[0053] When the target volume segment is located at the end of the volume segment sequence, since there are no adjacent segments in the segment sequence with increasing numbers, the body position drift increment can only be allocated to the adjacent segment on one side, that is, the volume segment that is adjacent to the target volume segment on the same side and located in the decreasing number direction.

[0054] When the target volume segment is located inside a continuous target volume segment, that is, when the target volume segment belongs to a continuous decision segment composed of two or more continuous target volume segments, the positional drift increment is not allocated to other target volume segments inside the continuous decision segment when allocating the positional drift increment. Instead, the positional drift increment is allocated to the next volume segment located outside the continuous target volume segment along the determined offset direction.

[0055] After determining the offset direction of the body position drift increment, the first volume segment located in that offset direction and directly adjacent to the target volume segment is selected as the priority allocation target. A portion of the body position drift increment is allocated to this first volume segment. After the allocation of this segment is completed, if there are still unallocated body position drift increments, the next adjacent volume segment is selected along the offset direction for sequential allocation until all body position drift increments are allocated.

[0056] During the allocation process, to prevent the allocation result from exceeding the reasonable variation range of adjacent volume segments, an allocation upper limit is set for each volume segment. The allocation upper limit is the absolute value of the volume difference corresponding to that volume segment. That is, the positional drift increment borne by each volume segment must not exceed the volume difference range generated under the two standard measurement positions.

[0057] Under the condition that the absolute values ​​of the volume differences are equal, the spatial distances between each adjacent volume segment and the target volume segment are compared. The distance is represented by the absolute value of the difference between the segment numbers; the smaller the difference in segment numbers, the closer the interval between the two segments in the limb length direction. Priority is given to allocating the body position drift increment to the volume segment with the smallest distance from the target volume segment, so that the attribution of body position drift is concentrated in the nearest neighbor region as much as possible, thus better reflecting the actual characteristics of local volume distribution translation.

[0058] When two or more adjacent volume segments are equidistant from the target volume segment, and the absolute values ​​of their volume differences are also equal, a priority order is further determined according to the segment number to avoid uncertainty in the allocation result. In this case, the body position drift increment is preferentially allocated to volume segments in the direction of increasing segment number. By setting a unified priority rule for the sequence number direction, a unique allocation path can still be obtained under symmetry conditions, thereby ensuring the repeatability of the algorithm execution process and the stability of the results.

[0059] Example 1:

[0060] On the morning of a routine follow-up day, the subject, a patient with upper limb lymphedema, underwent two standardized measurements using a limb 3D scanning device equipped with a structured light depth camera. The first standard measurement position was defined as the limb hanging naturally with the elbow extended; the second standard measurement position was defined as the limb abducted and slightly pronated to expose the axillary contour. Due to the influence of gravity and the shoulder-elbow angle, soft tissues may experience slight displacement of the overall contour along its length. Therefore, the second position may show a misalignment of the peak position of the segmental volume compared to the first position, introducing positional drift error. (Appendix) Figure 2 The comparison curves of the volume of each segment under two standard body positions are shown, and the target segments that meet the threshold conditions are marked. The second body position shows a significant peak in segment 8 and the opposite difference with segment 7 appears at the same time, which reflects the visual feature of the peak being misaligned along the segment number.

[0061] External contour data of the same limb were collected under both the first and second standard measurement positions. External contour data was defined as the arc length parameter along the limb's centerline. After camera calibration and point cloud denoising, the equivalent radius of the cross-sectional profile point set is estimated on each cross-section. To obtain a smooth and differentiable radius function, cubic B-splines are used to process the discrete radius samples. Fitting, definition:

[0062]

[0063] in For cubic basis functions, The control coefficient is used. Measurement noise is defined as the standard deviation of the radius estimate. Example .

[0064] The limb is divided into segments along its length with equal intervals. The interval between segments is defined as the difference in arc length between the centerlines of adjacent sections. ,Pick The segments are numbered sequentially from proximal to distal as segment 1 to segment 12. A set of corresponding volume segments is obtained for each body position. The volume of the segment for the first body position is denoted as... The volume of the second body position segment is .

[0065] Calculate the volume of the truncated section and provide two reproducible computational paths. Path A uses a frustum approximation, letting the first... The equivalent radii at both ends of the truncated segment are respectively and The truncated volume is defined as:

[0066]

[0067] Path B uses numerical integration, and the cross-sectional area function is defined as follows: The truncated volume is defined as:

[0068]

[0069] And use the compound Simpson's rule or the trapezoidal method to analyze each segment. Integration. The data is given by path A. and During verification, path B is used to recalculate the key segments, and the relative difference between the two is usually less than 1% to 2% to prove that alternative implementations are possible.

[0070] Error propagation is given to quantify the effect of measurement noise on volume. For the first-order approximation of the frustum formula under small perturbations, the variance of the truncated volume can be written as:

[0071]

[0072] in:

[0073]

[0074] This allows us to obtain the uncertainty range for each segment volume, which can be used to explain the robustness of threshold selection and outlier identification. The code calculation gives the uncertainty range for segment 10. Order of magnitude of volume standard deviation under the given conditions Its magnitude is significantly lower than the threshold described later. This is used to support a stable filtering interpretation of outlier differences using thresholds.

[0075] Calculate the volume difference and select the target volume segment. The volume difference is defined as Δ. and calculate The segment volume and difference data are shown in the table below, in mL.

[0076] Segment number D 1 180 205 25 25 2 200 205 5 5 3 220 235 15 15 4 240 238 -2 2 5 260 255 -5 5 6 280 270 -10 10 7 300 280 -20 20 8 290 330 40 40 9 270 275 5 5 10 250 269 19 19 11 230 228 -2 2 12 210 205 -5 5

[0077] right Sort by ascending order to obtain the sequence After removing the minimum value of 2 and the maximum value of 40, the average of the remaining 10 values ​​is... Preset multiplier A value of 1.7 is chosen because it falls above normal fluctuations caused by noise, while preserving the peak misalignment cutoff caused by body positional drift. The threshold is defined as:

[0078]

[0079] Therefore, satisfied. The target volume segments are segment 1, segment 7, segment 8, and segment 10. (See attached text.) Figure 3 It shows histogram distribution and threshold line And use texture to identify the satisfaction The target cutoff segment is used to separate the abnormal cutoff segment related to peak misalignment from the normal fluctuations.

[0080] The entity increment or body position drift increment is determined based on the correspondence of local volume distribution. The first local determination interval is defined as the interval formed by the target segment and adjacent segments in the first body position, and the second local determination interval is defined as the interval formed by the corresponding segments in the second body position. The representative segment number is defined as the segment number with the largest volume in each local determination interval; if multiple largest volumes exist, the smallest number is taken. If the representative segment numbers of two body positions are the same, the difference in the target segment value is determined as the entity increment; otherwise, it is determined as the body position drift increment. (Appendix) Figure 4 By comparing the representative segment numbers of boundary targets, ordinary targets, and continuous target segments in two body positions using step lines, the criteria for determining whether the representative segments are consistent or misaligned can be visually demonstrated.

[0081] An example of determining a typical single target segment is given for target segment 10. Its local determination interval is segments 9 to 11. The first body position interval has a volume of 270, 250, 230, with the maximum value corresponding to segment 9. The second body position interval has a volume of 275, 269, 228, with the maximum value corresponding to segment 9. Since the two segments have the same sequence number, therefore... If it is determined to be an entity increment, it means that the actual volume at that location has increased rather than the peak has shifted.

[0082] Since the target is located in the first segment, the local determination interval is taken from two adjacent segments on the same side, with the interval being segment 1 to segment 3. The volume of the first body position is 180, 200, 220, with the maximum value corresponding to segment 3. The volume of the second body position is 205, 205, 235, with the maximum value corresponding to segment 3. Since the two represent segment numbers being the same, therefore... The determination that it is an entity increment indicates that the difference in the first segment mainly comes from the entity deformation caused by real swelling changes or real posture rather than overall drift.

[0083] Since segments 7 and 8 are continuous and both are target volume segments, they are defined as continuous decision segments. The extended local decision interval is defined as the continuous decision segment and its two adjacent segments, i.e., segments 6 to 9. The first body position interval has volumes of 280, 300, 290, and 270, with the maximum value corresponding to segment 7. The second body position interval has volumes of 270, 280, 330, and 275, with the maximum value corresponding to segment 8. Since the segment numbers are different, the continuous decision segment... and The result was uniformly determined to be an incremental shift in body position, which manifested as a misalignment of the peak volume along the segment number between the two body positions.

[0084] Intermediate results are generated based on the first standard measurement position, while retaining the solid increment. The reference volume is defined as follows: The entity increment list is segment 1 with an increment of 25 mL and segment 10 with an increment of 19 mL. The body position drift increment list is segment 7 with a drift of 20 mL and segment 8 with a drift of 40 mL, and records the extended local judgment interval representing the segment number changing from 7 to 8. For body position drift increments, this embodiment only outputs a list to be corrected and does not assign them at this point. Subsequent assignment corrections will be completed based on the offset direction and the ratio of the difference between adjacent segments. Figure 2 With appendix Figure 4 Together they are used to explain that drift is manifested as a shift in the peak position rather than an overall increase in the volume of the entity.

[0085] The efficacy is described using quantitative indicators to provide verifiable conclusions. Without prior assessment, the second position, relative to the first position, exhibits an isolated peak of 40 mL at segment 8, accompanied by a reverse difference of 20 mL at segment 7. This could easily be misinterpreted as rapid local deterioration. This phenomenon is observed in the appendix... Figure 2 It presents itself in the form of peak misalignment, and in the appendix Figure 3 The abnormal bars appear as exceeding the threshold. After assessment, the difference between truncated segments 7 and 8 is uniformly classified as positional drift increment, ensuring that the true increment is mainly concentrated at the two entity increment locations: truncated segments 1 and 10. Taking the target set as an example, the total entity increment is 44 mL, and the total drift increment is 60 mL, which is marked as pending correction. This achieves logical separation between the true swelling change and the peak misalignment caused by positional drift, providing consistent input for subsequent correction and correction volume accumulation.

[0086] Example 2:

[0087] The same subject underwent a second retest during the same follow-up procedure. The operator used a structured light depth camera to acquire data in two positions a second time to verify the repeatability of the segmented volume calculation. Because the second standard measurement position differs from the first standard measurement position in terms of limb angle, local soft tissues are prone to overall contour displacement along their length under gravity and traction, resulting in misalignment of the peak position of the local volume distribution. (Appendix) Figure 5The diagram illustrates the sequential allocation process of the drift increment of continuous target segments in the offset direction, and simultaneously displays the upper limit of each segment's acceptable amount and the cumulative allocation amount, to visually demonstrate the mechanism of upper limit truncation and the sequential allocation of the remaining amount.

[0088] The input data directly reuses the output list of body position drift increments. The drift increment is defined as the absolute value of the volume difference between target segments determined to be part of the body position drift. For example, segment 7 has a drift of 20 mL and segment 8 has a drift of 40 mL, with the segment number changing from 7 to 8 within the extended local determination interval. To perform the allocation, a supplementary allocation criterion for adjacent segments is added, defined as the absolute value of the volume difference between adjacent segments that can receive drift increments under the two body positions. Example: Take segment 9 =5mL and 10% of the truncated segment =19mL, used for multiple adjacency allocation and upper limit constraint.

[0089] Boundary targets are defined as target segments located at the beginning or end of the segment and identified as having drift increments. For example, the drift volume of target segment 1 in the beginning segment is 10 mL, and the adjacent segment on the same side is segment 2. =5mL and 3 segments and =15mL. The objective of the parallel rule is defined as the existence of multiple adjacent segments available for allocation along the offset direction, and these segments... In the case of equal values, for example, the drift volume of target segment 6 is 10 mL, and the drift direction is the direction of increasing segment number. Segment 7 and segment 8 are available as candidates for allocation, and are set as follows: = =20mL, used to trigger the parallel rule of distance priority and sequence number increasing direction priority.

[0090] The offset direction is determined by the difference in segment numbers. The offset direction is defined as the migration direction of the body position drift increment along the segment number axis, denoted as [the segment number is missing from the original text]. The second local decision interval represents the segment number as follows: ,but:

[0091] Δ

[0092] When Δ When the value is greater than 0, the offset direction is the direction of increasing segment number. When Δ When the value is less than 0, the offset direction is the direction of decreasing segment number. Taking continuous target segments 7 to 8 as an example, the local judgment interval is expanded. , Therefore Δ =1, the offset direction is the direction of increasing segment number, this calculation is used to satisfy the offset direction determination rule.

[0093] Further, an example of directional inference is given for the second objective. For example parallel rule objective segment 6, the segment number representing the first local decision interval is taken. The second local decision interval represents the segment number. , then Δ =1, and the offset direction is also the direction of increasing segment number. For boundary target segment 1, the inference of its offset direction does not change the boundary assignment constraints, even if Δ exists. The inference results still prioritize the allocation of adjacent segments on the same side of the first segment, and the allocation result of the multi-adjacent ratio of this boundary is determined by... Figure 6 It is presented visually.

[0094] The attribution correction is implemented using steps T1 to T6. Step T1 reads the drift increment list and forms a set of adjacent segments available for allocation for each drift target segment, with elements sorted by offset direction. Step T2: If the target segment is located at the beginning or end of a segment, only adjacent segments on the same side are retained according to boundary rules. Step T3: If the target segment is located within a continuous target segment, only the next segment located outside the continuous segment along the offset direction is retained as a set according to continuous segment rules. Step T4: Multiple segments within the set are allocated proportionally, with weights defined as follows:

[0095]

[0096] And define the unconstrained assign as:

[0097]

[0098] in The set of adjacent segments available for allocation. This is the drift increment. Step T5 introduces an upper limit constraint and distributes it sequentially according to the offset direction. The constraint is as follows: If a segment reaches its upper limit, it is truncated, and the remaining amount is distributed to the next segment along the offset direction. Step T6: If there are multiple candidates in the set and If they are equal, they will be assigned to the segment with the smallest distance from the target segment according to the tie rule. If the distances are also the same, they will be assigned to the segment in the direction of increasing segment number.

[0099] The assignable segment for a drifting target within a continuous target segment is defined as the next segment located outside the continuous segment along the offset direction, where the offset direction is the direction of increasing segment number. Therefore, the set... ={9}. Following steps T3 and T5, the total drift volume of 60 mL, consisting of 20 mL and 40 mL drift volumes, is allocated starting from segment 9 according to the rules. This satisfies the upper limit constraint. Therefore, segment 9 is allocated first. The remaining drift volume is 55 mL, and it is further distributed in the direction of drift to segment 10, with an upper limit of [missing value]. =19mL, therefore, the fraction was divided into 10 portions. =19mL, remaining drift volume is 36mL. Continue dispensing along the offset direction to segment 11, its upper limit is =2mL, dispensing =2mL, the remaining amount is 34mL, continue to the cutoff segment 12, the upper limit is =5mL, dispensing =5mL, the remaining amount is 29mL. At this point, there are no more segments available for allocation within the given data range. The remaining amount is recorded as the unassigned drift amount and is retained as the compensation amount for subsequent retests, without changing the allocation rule closed loop.

[0100] To demonstrate the proportional allocation of multiple adjacencies and the solution with constraints, an example drift of 10 mL is used for calculations on the boundary target segment 1. Under the boundary rule, the set is taken from adjacent segments on the same side. ,That =5mL, =15mL, therefore the weight is , The unconstrained allocation is , mL. Figure 6 The upper limit of the candidate cutoff segments is displayed simultaneously in a bar chart comparison format. With allocation And overlay the unconstrained quantities with a broken line. This proves that the unconstrained solution in this example satisfies the box constraint. The example also provides an optimization expression with an upper bound, letting... To assign the actual vector, define:

[0101]

[0102] Satisfy constraints:

[0103] ,

[0104] when =10mL and =5mL, When the volume is 15 mL, the unconstrained solution satisfies the box constraint. Therefore , No truncation or redistribution is required.

[0105] Assuming the offset direction is the direction of increasing segment number, the candidate segments to be assigned are segment 7 and segment 8. = =20mL equal, the distances from target segment 6 to segment 7 and segment 8 are 1 and 2 respectively, therefore, according to the minimum distance priority, the drift volume of 10mL is preferentially allocated to segment 7, resulting in , If we further construct a parallel case with the same distance, let the candidates for target segment 7 be segment 6 and segment 8, and both... = If the segments are equal and the distance is 1, the drift amount will be preferentially allocated to segment 8 according to the principle of increasing segment number. This rule is used to eliminate allocation ambiguity and maintain the determinism of allocation results under symmetric conditions.

[0106] After completing the attribution correction, update the correction volume. The correction volume is defined as the set of volumes of the second body position segment after removing the drift increment from the original target segment and attributing it to the receiving segment. The correction increment of the receiving segment set is denoted as... The corrected volume of the second body position segment is:

[0107]

[0108] Meanwhile, the drift amount corresponding to the original drift target segment is no longer included in the segment difference in this embodiment, thus making the peak position of the corrected difference closer to the local peak position of the first body position. The peak value of the difference of the target segment set before correction is =40mL, after correction, the drift peak is no longer retained at cutoff 8 and the peak misalignment is transferred to the adjacent cutoff along the offset direction, forming an interpretable attribution result.

[0109] To provide a quantitative comparison index, the total deviation is defined as the absolute value of the difference between the sum of the volumes of the two body positions in the segmented sections, and the peak value of the segmented difference is defined as... And define the repeatability index as the root mean square of the difference between the repeated tests:

[0110]

[0111] Figure 7 The comparison of indicators before and after correction is overlaid, providing calculation results for total deviations of 65mL to 96mL, peak values ​​of segmented differences of 40mL to 40mL, and RMSE values ​​of 16.88 to 19.47, which can be used as a basis for... Updated quantization output. After correction, because the drift peak remains within the original target cutoff and the added attribution increases the difference in the later segments, the peak value of the segmented difference remains unchanged, but the RMSE increases. This result is consistent with... Figure 7 The indicators are consistent.

[0112] By determining the offset direction and using the rule of the next segment outside the continuous segment, the peak misalignment caused by drift is shifted back to the adjacent segment and allocated in an interpretable manner. Through proportional allocation and a box-constrained projection mechanism, it is ensured that the allocation amount for each received segment does not exceed its... The upper limit can be redistributed sequentially along the offset direction after exceeding the limit. This is achieved through parallel rules. When the values ​​are equal, the nearest segment is prioritized, and when the distance is the same, the segment number is prioritized in the direction of increasing. This ensures that the allocation result is definite and reproducible, thereby achieving stable processing of boundary scenes and continuous segment scenes.

Claims

1. A method for calculating and correcting the error of limb volume in cases of lymphedema, characterized in that... include: Under the first and second standard measurement positions, the outer contour data of the limb to be measured for lymphedema were collected respectively, and the limb was divided into segments along the length direction with the same segment interval to obtain two sets of corresponding volume segments. Calculate the volume of each corresponding volume segment to determine the volume difference for the same segment number; The volume segment whose absolute value of the volume difference is greater than a preset threshold is determined as the target volume segment; Based on the correspondence of the local volume distribution of the target volume segment in the two sets of volume segments, the volume difference is determined to be either the entity increment or the body position drift increment. Using the volume of each volume segment under the first standard measurement position as a benchmark, the entity increment is retained in the target volume segment, and the positional drift increment is corrected for attribution. The volumes of each volume segment after attribution correction are summed to obtain the corrected volume of the limb with lymphedema to be tested.

2. The method for calculating and correcting the volume of lymphedema limbs according to claim 1, characterized in that... Determining whether the volume difference is a solid increment or a body position drift increment includes: in two sets of volume segments, forming a first local determination interval and a second local determination interval with the target volume segment and adjacent segments respectively, determining the representative segment number corresponding to the largest segment volume in each local determination interval; when there are multiple largest segment volumes, taking the segment number with the smallest number as the representative segment number; when the two representative segment numbers are the same, determining that the volume difference is a solid increment; when the two representative segment numbers are different, determining that the volume difference is a body position drift increment.

3. The method for calculating and correcting the volume of lymphedema limbs according to claim 2, characterized in that... The assignment correction for the body position drift increment includes: determining the offset direction of the body position drift increment based on the representative segment number of the first local determination interval and the representative segment number of the second local determination interval; allocating the body position drift increment to volume segments adjacent to the target volume segment according to the offset direction; when allocating to multiple adjacent volume segments, the allocation is based on the proportion of the absolute value of the volume difference between each adjacent volume segment to the sum of the absolute values ​​of the volume differences between the multiple adjacent volume segments.

4. The method for calculating and correcting the volume of lymphedema limbs according to claim 1, characterized in that... Sort the absolute values ​​of the volume differences in ascending order; Remove the volume difference between the first and last digits after sorting; The average volume difference is obtained by averaging the remaining volume differences; the product of the average volume difference and a preset multiple is determined as the preset threshold.

5. The method for calculating and correcting the volume of lymphedema limbs according to claim 2, characterized in that... The local determination interval is composed of the target volume segment and its two adjacent volume segments before and after it; when the target volume segment is located at the first or last segment of the volume segment sequence, the local determination interval is composed of the target volume segment and two adjacent volume segments on the same side.

6. The method for calculating and correcting the volume of lymphedema limbs according to claim 2, characterized in that... When there are two or more consecutive target volume segments, the consecutive target volume segments are used as consecutive decision segments, and the adjacent volume segments on both sides of the consecutive decision segment together with the consecutive decision segment constitute an extended local decision interval. Based on the representative segment number within the extended local determination interval, the volume difference corresponding to the continuous target volume segments is uniformly determined as either entity increment or body position drift increment.

7. The method for calculating and correcting the volume of lymphedema limbs according to claim 3, characterized in that... The offset direction of the body position drift increment is determined by comparing the representative segment numbers of the first local judgment interval and the second local judgment interval; when the representative segment number of the second local judgment interval is greater than the representative segment number of the first local judgment interval, the offset direction is determined to be the direction of increasing segment number; when the representative segment number of the second local judgment interval is less than the representative segment number of the first local judgment interval, the offset direction is determined to be the direction of decreasing segment number.

8. The method for calculating and correcting the volume of lymphedema limbs according to claim 3, characterized in that... When the target volume segment is located at the first or last segment, the body position drift increment is allocated to the volume segment adjacent to the target volume segment on the same side; when the target volume segment is located within a continuous target volume segment, the body position drift increment is allocated to the next volume segment located outside the continuous target volume segment along the offset direction.

9. The method for calculating and correcting the volume of lymphedema limbs according to claim 3, characterized in that... When the body position drift increment is distributed to multiple adjacent volume segments, it is distributed sequentially according to the offset direction, and the distribution amount of each volume segment does not exceed the absolute value of the corresponding volume difference.

10. The method for calculating and correcting the volume of lymphedema limbs according to claim 7, characterized in that... When there are multiple adjacent volume segments available for allocation in the offset direction, and the absolute value of the volume difference between the multiple adjacent volume segments is equal, the body position drift increment is preferentially allocated to the volume segment with the smallest distance from the target volume segment; when the multiple adjacent volume segments are at the same distance from the target volume segment, the body position drift increment is allocated to the volume segment in the direction of increasing segment number.

Images