CN117694864A - Lymphedema volume intelligent measurement method - Google Patents
Lymphedema volume intelligent measurement method Download PDFInfo
- Publication number
- CN117694864A CN117694864A CN202311726197.2A CN202311726197A CN117694864A CN 117694864 A CN117694864 A CN 117694864A CN 202311726197 A CN202311726197 A CN 202311726197A CN 117694864 A CN117694864 A CN 117694864A
- Authority
- CN
- China
- Prior art keywords
- data
- lymphedema
- volume
- model
- analysis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 206010025282 Lymphoedema Diseases 0.000 title claims abstract description 134
- 208000002502 lymphedema Diseases 0.000 title claims abstract description 134
- 238000000691 measurement method Methods 0.000 title claims abstract description 27
- 238000011871 bio-impedance analysis Methods 0.000 claims abstract description 83
- 238000003384 imaging method Methods 0.000 claims abstract description 51
- 238000005259 measurement Methods 0.000 claims abstract description 50
- 206010030113 Oedema Diseases 0.000 claims abstract description 38
- 238000009825 accumulation Methods 0.000 claims abstract description 22
- 238000004364 calculation method Methods 0.000 claims abstract description 19
- 238000005516 engineering process Methods 0.000 claims abstract description 18
- 238000012544 monitoring process Methods 0.000 claims abstract description 14
- 230000004660 morphological change Effects 0.000 claims abstract description 11
- 239000007788 liquid Substances 0.000 claims abstract description 9
- 210000004880 lymph fluid Anatomy 0.000 claims abstract description 7
- 238000013499 data model Methods 0.000 claims abstract description 6
- 230000007774 longterm Effects 0.000 claims abstract description 6
- 238000001727 in vivo Methods 0.000 claims abstract description 3
- 238000000034 method Methods 0.000 claims description 82
- 230000006870 function Effects 0.000 claims description 48
- 230000004927 fusion Effects 0.000 claims description 46
- 238000004458 analytical method Methods 0.000 claims description 42
- 238000010801 machine learning Methods 0.000 claims description 35
- 238000013527 convolutional neural network Methods 0.000 claims description 27
- 238000012545 processing Methods 0.000 claims description 27
- 230000008859 change Effects 0.000 claims description 21
- 238000004422 calculation algorithm Methods 0.000 claims description 19
- 239000012530 fluid Substances 0.000 claims description 19
- 238000007405 data analysis Methods 0.000 claims description 13
- 238000013079 data visualisation Methods 0.000 claims description 13
- 238000013135 deep learning Methods 0.000 claims description 12
- 230000002452 interceptive effect Effects 0.000 claims description 12
- 230000010354 integration Effects 0.000 claims description 11
- 238000013136 deep learning model Methods 0.000 claims description 9
- 238000003909 pattern recognition Methods 0.000 claims description 9
- 238000011161 development Methods 0.000 claims description 8
- 238000001566 impedance spectroscopy Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 8
- 238000011269 treatment regimen Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 7
- 238000012800 visualization Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 5
- 238000009499 grossing Methods 0.000 claims description 5
- 239000000203 mixture Substances 0.000 claims description 5
- 238000003062 neural network model Methods 0.000 claims description 5
- 238000012546 transfer Methods 0.000 claims description 5
- 230000003190 augmentative effect Effects 0.000 claims description 4
- 238000003745 diagnosis Methods 0.000 claims description 4
- 238000013178 mathematical model Methods 0.000 claims description 4
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000003993 interaction Effects 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 claims description 3
- 210000001519 tissue Anatomy 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 3
- 230000006399 behavior Effects 0.000 claims description 2
- 238000005452 bending Methods 0.000 claims description 2
- 230000001186 cumulative effect Effects 0.000 claims description 2
- 238000013480 data collection Methods 0.000 claims description 2
- 230000001419 dependent effect Effects 0.000 claims description 2
- 238000002847 impedance measurement Methods 0.000 claims description 2
- 238000001453 impedance spectrum Methods 0.000 claims description 2
- 238000007500 overflow downdraw method Methods 0.000 claims description 2
- 230000000737 periodic effect Effects 0.000 claims description 2
- 238000011002 quantification Methods 0.000 claims description 2
- 238000013139 quantization Methods 0.000 claims description 2
- 230000004044 response Effects 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 239000013598 vector Substances 0.000 claims description 2
- 238000011282 treatment Methods 0.000 description 13
- 239000000284 extract Substances 0.000 description 4
- 238000013459 approach Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 230000002497 edematous effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000001225 therapeutic effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 210000002751 lymph Anatomy 0.000 description 1
- 230000001926 lymphatic effect Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000002503 metabolic effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000003205 muscle Anatomy 0.000 description 1
- 238000000053 physical method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000011285 therapeutic regimen Methods 0.000 description 1
Landscapes
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to an intelligent lymphedema volume measurement method. Comprising the following steps: firstly, using a multi-mode sensor array, including a pressure sensor, a temperature sensor and a conductivity measurement sensor, monitoring the pressure, temperature and conductivity changes of limb skin in real time, and providing preliminary data for physical characteristics of lymphedema; the sensor is integrated in the wearable device: smart bracelets or custom pressure socks for long-term and continuous monitoring; then combining 3D imaging techniques: laser scanning or structured light scanning, creating an accurate three-dimensional model for the affected limb of the patient, providing detailed volume measurement and morphological change recording of the oedema region; measuring the bioelectrical resistance and capacitance of the limb through the electrode patch on the surface of the limb by using a noninvasive bioelectrical impedance analysis technology to reveal the distribution of in-vivo liquid and the accumulation of lymph fluid; and finally, fusing and analyzing through an integrated software platform to obtain a calculation data model of the lymphedema volume.
Description
Technical Field
The invention relates to an intelligent lymphedema volume measurement method.
Background
The defects and shortages of the prior art in the lymphedema volume measurement are quite remarkable, and the accuracy, convenience and comfort level of patients of the measurement are affected. First, conventional measurement methods, such as measuring the circumference of the limb with a ruler or tape, present significant accuracy and repeatability problems. These methods often rely on the skill and experience of the operators, and skill differences between operators may lead to inconsistent results. In addition, these physical measurements often fail to provide detailed information about the location and depth of lymph fluid accumulation, making it difficult to fully assess the condition and treatment outcome. Second, many conventional measurement methods do not allow for dynamic monitoring of lymphedema. This means that they cannot continuously record changes in limb volume, which is important for assessing the effect of the treatment and adjusting the treatment regimen in time. Furthermore, some measurement methods may cause discomfort to the patient, especially for inconveniently moving patients, where the method of measuring volume by tightly wrapping the limb may be quite difficult to operate. Furthermore, there is a lack of standardization between current measurement methods and techniques. The measurement results of different methods and techniques are difficult to compare with each other, which limits the general applicability of the data and the popularization of the study to some extent. For example, different medical institutions may employ different measurement protocols and equipment, resulting in inconsistent treatment outcome assessment. There are also technical and equipment limitations that are an important issue. Some more advanced measurement techniques, such as certain types of imaging techniques, may be difficult to use widely due to their high cost, lack of popularity of equipment, or complexity of operation. These advanced techniques, while providing more accurate and comprehensive measurements, have obstacles in popularity and daily use.
In summary, existing lymphedema volume measurement methods have limitations in several respects, including accuracy and repeatability of the measurement, comprehensiveness, ease of operation, patient comfort, standardization between methods, and popularity of techniques and equipment. To overcome these limitations, researchers and medical professionals are striving to develop more advanced, accurate, and user-friendly measurement methods, such as intelligent measurement systems that combine deep learning and machine learning techniques, to improve the diagnostic and therapeutic effects of lymphedema.
Disclosure of Invention
The invention aims to provide an intelligent lymphedema volume measurement method, so as to solve part of defects and shortcomings pointed out in the background art.
The invention solves the technical problems as follows: an intelligent lymphedema volume measurement method, comprising: firstly, using a multi-mode sensor array, including a pressure sensor, a temperature sensor and a conductivity measurement sensor, monitoring the pressure, temperature and conductivity changes of limb skin in real time, and providing preliminary data for physical characteristics of lymphedema; the sensor is integrated in the wearable device: smart bracelets or custom pressure socks for long-term and continuous monitoring; then combining 3D imaging techniques: laser scanning or structured light scanning, creating an accurate three-dimensional model for the affected limb of the patient, providing detailed volume measurement and morphological change recording of the oedema region; measuring the bioelectrical resistance and capacitance of the limb through the electrode patch on the surface of the limb by using a noninvasive bioelectrical impedance analysis technology to reveal the distribution of in-vivo liquid and the accumulation of lymph fluid; and finally, fusing and analyzing through an integrated software platform, and utilizing a data processing algorithm: and (3) machine learning and pattern recognition, and comprehensively analyzing the multi-source data to obtain a calculation data model of the lymphedema volume.
Further, the multi-modal sensor array comprises an integrated pressure sensor, a temperature sensor and a conductivity measurement sensor; the multi-mode sensor array uses microminiaturization and flexible design, so that the sensor array is suitable for being used for being clung to skin for a long time, and the pressure, temperature and conductivity changes of limbs are monitored and recorded;
by the formula for calculating the edema rate R (t):
wherein (P), (T) and (C) represent real-time readings of pressure, temperature and conductance, respectively, (a), (b) and (C) are coefficients derived based on experimental data for adjusting the extent to which pressure, temperature and conductance changes affect edema rate;
and calculate the total edema change (V) over a period of time total ) Is a function of the integral formula:
(V total ): in a time period (t 0 ) To (t) n ) Total edema volume change between;
is performed on the edema rate R (t) at time (t 0 ) To (t) n ) Integration of the period provides a cumulative change in the amount of edema over the period of time.
Further, the implementation of the 3D imaging technique includes: firstly, capturing a high-precision three-dimensional model of an affected limb by using a structured light scanning or laser scanning technology in combination with a machine learning algorithm; and secondly, applying a volume calculation formula:
V＝∫∫∫ D f(x,y,z)dxdydz
wherein,
(V): the total volume of lymphedema area measured;
(∫∫∫ D ): triple integration of region D (i.e., the affected limb region) is performed for volume calculation;
f (x, y, z): this is a function used to describe the local nature of each point in three-dimensional space;
(dxdydz): these are tiny volume elements, tiny variations along the (x), (y) and (z) directions in three-dimensional space; applying the surface area calculation formula:
A＝∫∫ S g(x,y,z)dS
wherein,
(A) The method comprises the following steps The total surface area of the lymphedema area being measured;
(∫∫ S ): double integration of the surface S (i.e. the surface of the affected area) is performed for calculating the surface area;
g (x, y, z): this is a function describing the local properties of each point on the surface in three dimensions, related to the curvature properties of the surface;
(dS): this is a tiny surface element, tiny areas on a three-dimensional surface;
next, the change in limb morphology is analyzed using differential geometry and topology principles, including calculating morphology change index formulas:
M＝∫∫ S h(K,H)dS
(M): a quantization index of morphological changes;
h (K, H): this is a function of the gaussian curvature K and the mean curvature H based on a curved surface, for describing the complexity or significance of the morphological changes;
(K) And (H): the Gaussian curvature and the average curvature of the curved surface are respectively described, and the bending degree of the curved surface at different points is described;
(dS): representing a tiny area on a curved surface;
finally volume and morphology changes of lymphedema were monitored and assessed by periodic 3D scan data analysis.
Further, the technical method for non-invasive bioelectrical impedance analysis comprises the steps of firstly using a bioelectrical impedance analysis instrument with high sensitivity, wherein the bioelectrical impedance analysis instrument adopts a multi-frequency measurement technology; then using the mathematical model formula:
Z(f)＝A(f)·e (jφ(f))
to analyze multifrequency bioelectrical impedance data, wherein (Z (f) is a complex impedance at frequency (f), A (f) is a frequency dependent magnitude function, phi (f) is a phase angle function, and further developing an adaptive algorithm:
to adjust the electricityThe measured parameter of the polar patch, wherein P (f, ΔZ) is the measured parameter adjustment value, (ΔZ) is the measured impedance variation, Z 0 (f) Is the initial impedance value, and (K) and B (f) are adjustment coefficients; for adjusting the response of the algorithm according to different measurement conditions;
finally, a calculation formula is applied:
to estimate the liquid distribution and accumulation,
wherein,
l (x, y, z): the level of fluid accumulation at three dimensional spatial points (x, y, z) is a quantification of fluid distribution and accumulation in the body;
for a certain frequency range (f 1 ) To (f) 2 ) Integrating the electrical impedance in the circuit to consider measurement information at different frequencies;
Z (f): as before, electrical impedance at a particular frequency;
Z max (f) The method comprises the following steps Is the maximum electrical impedance value at frequency (f) for normalizing the electrical impedance measurements.
Further, the software platform performs fusion and analysis methods, which first include processing data from a variety of data sources: a data processing platform comprising 3D scan data, bioelectrical impedance data, and patient medical history; the processing platform applies a deep learning network and a high-level analysis model, in particular a custom deep neural network model:
for predicting the volume of lymphedema from the integrated data (x),
F lymphedema (x) The method comprises the following steps Outputting a predictive function of the lymphedema volume as a predicted lymphedema volume;
(x) The method comprises the following steps Input data, herein referred to as a feature set that integrates the multi-source data;
summing the outputs of all neurons;
(w i ): weights of (i) th neurons in the neural network;
(sigma): an activation function for introducing nonlinearities that enable the neural network to learn more complex patterns;
(a i •x+b i ): a weighted input to the (i) th neuron, where (a) i ) Is a weight (b) i ) Is biased;
in addition, the method for fusing and analyzing the software platform further comprises a feature extraction function:
for converting the raw data into meaningful feature vectors,
E (x): extracting a function of the features from the raw data (x);
this is a logic function for converting the raw data into values between 0 and 1, making it suitable for input to the machine learning model;
in addition, the method for fusing and analyzing the software platform further comprises a mode identification and data visualization algorithm:
for identifying and visually presenting the pattern of lymphedema,
V pattern (x) The method comprises the following steps Visualization of the progression pattern of lymphedema;
is an integral expression, wherein ∈>Is a gaussian function for smoothing and weighting the features E (x); this expression helps identify patterns in the data and convert them into a form that is visualized.
Further, another method for implementing the deep neural network model includes: first, a data processing platform integrating a plurality of machine learning models is developed, wherein each model M k (x) Specialized processing of different features and patterns of data; the output of the integrated model is given by a weighted sum, namely:
wherein F is ensemble (x) The method comprises the following steps This is the final output of the integrated model, representing a predicted value of lymphedema volume;
summing the outputs of all (K) different machine learning models;
(α k ): a weight of the (k) th model, the weight being dynamically assigned according to the behavior of the model;
M k (T (x)): the output of the (k) th machine learning model, where T (x) is a feature transfer function applied to the input data (x); the raw data is then converted into a form more suitable for model learning using a feature transfer function T (x):
wherein T (x): the characteristic conversion function is applied to the original input data (x) and is used for improving the applicability of the data and the prediction performance of the model;
(x 2 +λ): this is a conversion formula in which (x 2 ) The square of the original data increases the nonlinearity of the data, and (lambda) is a smoothing parameter, avoiding the error of dividing by zero;
and then dynamically distributing weights according to the performance of each model in the training process, wherein the specific formula is as follows:
wherein (alpha) k ): weights of the (k) th model;
exp(-E k ): is an exponential function for the error rate (E) of the conversion model (k) k ) Is the weight;
the exponential transformation summation of all model errors is used for normalizing the weights;
the meta learner (L) is applied at the top level of the integrated model to integrate the outputs of the different models as follows:
wherein,is the output of the (k) th integration model, and (L) is a meta learner for integrating these outputs.
Further, another implementation method of the pattern recognition and data visualization algorithm adopts the following steps: first, a deep learning analysis is performed on a 3D scan image using a Convolutional Neural Network (CNN) to accurately identify and distinguish lymphedema areas from normal tissues; the CNN model is described by the following formula:
CNN lymphedema (I) The method comprises the following steps Is the output of the CNN model for analysis of lymphedema, representing the lymphedema region identified from the 3D image (I);
summing all (n) convolution layers;
(W i ): is the weight of the (i) th convolutional layer;
(*): convolution operation, which is used for extracting the characteristics in the image;
(sigma): is an activation function for introducing nonlinearity;
(I) The method comprises the following steps Is an input 3D image;
(b i ): is the bias term for the (i) th convolutional layer;
secondly, a self-encoder technique is adopted: AE (AE) features (x) =dec (Enc (x)) to extract key features in 3D images;
wherein AE features (x) The method comprises the following steps Is the output from the encoder for extracting lymphedema features;
enc (x): converting the input 3D image (x) from the encoding part of the encoder into a more compact feature;
(Dec): reconstructing data from the encoded features from a decoding portion of the encoder;
next, an interactive 3D data visualization tool is developed whose chart update rules are controlled by the following formula:
VizUpdate(x,t)＝∫ 0 t γ(t-τ)·CNN lymphedema (x(τ))dτ
(VizUpdate (x, t)): a visual state at a point in time (t);
integrating the data in the time window to consider the history information;
gamma (t-tau): is a time decay function for giving different weights to the data at different points in time;
CNN lymphedema (x (τ)): is a CNN model applied on a 3D image x (τ) of time (τ).
Further, the bioelectrical impedance analysis and 3D imaging another technical scheme comprises the following key steps: first, a multidimensional Bioelectrical Impedance Analysis (BIA) technique is implemented that utilizes a multi-frequency and multi-electrode configuration to accurately measure the bioelectrical resistance value, as follows:
wherein,
BIA multi (f, θ): this is the output of a multi-dimensional bioelectrical impedance analysis, which represents the integrated resistance measurements at different frequencies (f) and electrode configurations (θ);
(∑ i ): summing the resistance measurements of all electrode configurations;
γ i (f) The method comprises the following steps Is a weight function of the (i) th electrode configuration at frequency (f) for adjusting the influence of the resistance values at different frequencies;
R i (θ, f): resistance measurements at a particular electrode configuration (θ) and frequency (f);
secondly, fusing 3D imaging technology with BIA data to provide comprehensive lymphedema volume information, wherein the fusion formula is as follows:
Fusion 3D_BIA (V 3D ,BIA multi )＝δ·V 3D +(1-δ)·Conv(BIA multi )
wherein,
(Fusion 3D_BIA ): this is the output of the 3D imaging data and bioelectrical impedance data fused, providing comprehensive lymphedema volume information;
(V 3D ): is lymphedema volume data acquired from 3D imaging techniques;
Conv(BIA multi ): is to convert bioelectrical impedance analysis data into volume An estimated convolution function;
(delta): is a fusion coefficient for balancing the contributions of the 3D imaging data and the BIA data;
finally, applying the deep learning model
DL analysis (Fusion 3D_BIA )＝DeepNet(Fusion 3D_BIA )
The data analysis is performed and the data is analyzed,
wherein:
(DL analysis ): the data analysis output of the deep learning enhancement is used for extracting key characteristics and trends of lymphedema from the fusion data;
DeepNet(Fusion 3D_BIA ): is a deep neural network applied to process and analyze the fused data (Fusion 3D_BIA )。
Further, the bioelectrical impedance analysis and 3D imaging further comprises the following key steps: first, bioelectrical Impedance Spectroscopy (BIS) is performed, which provides detailed information about body composition and fluid distribution by measuring bioelectrical impedance at a series of different frequencies; the mathematical expression of BIS is:
wherein,
BIS (f): bioelectrical impedance spectroscopy at frequency (f);
summing all the frequency points (N) to obtain an overall bioelectrical impedance spectrum;
(A k ): an amplitude parameter of a (k) th frequency point, the magnitude of the electrical impedance at that frequency;
a Gaussian function describing the frequency (f k ) NearbyWherein (f) k ) Is the center frequency (sigma) k ) Is the standard deviation of the frequency point;
Then, a machine learning Model (ML) is used in combination with the 3D imaging data and the BIS data model ) And (3) carrying out data fusion and analysis, wherein the formula is as follows:
Fusion 3D_BIS ＝ML model (V 3D ,BIS(f))
(Fusion 3D_BIS ): outputting fused 3D imaging data and BIS data, namely comprehensively analyzing the volume of lymphedema;
(ML model ): is a machine learning model for data fusion responsible for combining 3D imaging data (V 3D ) And BIS data BIS (f) to provide a final analysis result;
(V 3D ): volumetric data of lymphedema region obtained from 3D imaging techniques;
BIS (f): is the result of bioelectrical impedance spectroscopy analysis, providing detailed electrical impedance information about the lymphedema region.
An intelligent lymphedema volume measurement method specifically comprises the following steps:
s1, data collection:
3D imaging: 3D imaging using structured light scanning or laser scanning techniques to capture detailed three-dimensional morphology of the affected limb; this provides the basis data for subsequent volume calculations;
bioelectrical impedance analysis (BIA/BIS): bioelectrical impedance analysis, including multi-frequency and multi-electrode configurations, is used to measure the resistance and capacitance of the limb to infer fluid distribution in the body, particularly lymph fluid accumulation;
s2, data processing and fusion:
3D volume calculation: processing data obtained from the 3D scan, estimating the volume of the affected region using a volume calculation algorithm;
BIA/BIS data analysis: processing BIA/BIS data to obtain bioelectrical impedance characteristics of the lymphedema region; in the case of BIS, data at a series of frequencies are analyzed to provide more detailed body composition information;
data fusion: combining the 3D imaging data and the BIA/BIS data, integrating the two types of data using machine learning models or other data fusion techniques to provide a more comprehensive lymphedema analysis;
s3, intelligent analysis and prediction:
deep learning/machine learning model application: applying a deep learning model, such as a Convolutional Neural Network (CNN) or a self-encoder, to analyze the fused data; the models can identify the mode of lymphedema, predict the development trend of lymphedema and assist doctors in diagnosis;
pattern recognition and trend analysis: identifying a specific mode and a development trend of lymphedema by using a machine learning model, and providing accurate medical advice and treatment strategies;
s4, visualization and interaction:
data visualization: presenting the measurement and analysis results in a graphical form, providing an intuitive view using interactive 3D models or Augmented Reality (AR) techniques, helping doctors and patients to better understand the status of lymphedema;
Interactive feedback: a platform is provided to enable a physician to interact with the data, such as adjusting view angles, analyzing the data at different points in time.
The lymphedema volume intelligent measurement method of the invention has a plurality of beneficial effects, and the effects are mainly focused on improving the accuracy of measurement, enhancing the depth and breadth of data analysis and improving the experience of patients. In particular, these benefits include:
1. measurement accuracy is improved: in combination with 3D imaging techniques and bioelectrical impedance analysis (BIA/BIS), the present invention is able to more accurately capture volume changes and fluid distribution of the affected limb, thereby providing more accurate data than conventional measurement methods.
2. Comprehensive data analysis: by using the deep learning and machine learning models, the invention can carry out deep analysis on the collected data, identify the development mode of lymphedema and predict the change trend of the illness state, thereby providing more comprehensive information for doctors to formulate a treatment scheme.
3. Dynamic monitoring capabilities: the invention makes it possible to continuously monitor lymphedema, which is important for assessing the therapeutic effect and adjusting the therapeutic regimen in time.
4. Noninvasive and patient comfort: the present invention provides a non-invasive, more patient friendly measurement than some conventional measurement methods that may cause discomfort.
5. Data visualization and interactivity: the invention displays the measurement and analysis results in a graphical form through an interactive 3D model or an augmented reality technology, which not only enables doctors to understand data more intuitively, but also enables patients to understand their illness states more easily.
6. Standardization and replicability: by using standardized measurement and analysis procedures, the present invention improves data consistency and replicability, which is of great importance for clinical research and cross-center comparison.
Drawings
FIG. 1 is a flow chart of an intelligent lymphedema volume measurement method according to the present invention.
Detailed Description
The following describes the embodiments of the present invention in detail with reference to the drawings.
In one method of intelligent measurement of lymphedema volume, a multimodal sensor array is first used, which includes pressure sensors, temperature sensors and conductivity measuring sensors to monitor in real time the pressure, temperature and conductance changes of the limb skin. The pressure sensor monitors the change of the local pressure of the limb and can indicate the size and degree of the edema area; the temperature sensor captures the temperature change of limb skin and reflects the difference of local circulation and metabolic state; conductivity measuring sensors measure changes in the electrical conductivity of the limb, thereby revealing fluid distribution and lymphatic fluid accumulation. The data collected by these sensors provides preliminary but critical information on the physical characteristics of lymphedema, which in turn is combined with the results of other measurement techniques such as 3D scanning or bioelectrical impedance analysis (BIA/BIS) to provide a more comprehensive assessment of lymphedema volume and properties, which is critical for accurate diagnosis and effective treatment of lymphedema.
The integration of a multimodal sensor into a wearable device, such as a smart wristband or custom pressure sock, such a configuration makes monitoring lymphedema more convenient and efficient. The intelligent bracelet or the pressure sock is embedded with a pressure sensor, a temperature sensor and a conductivity measurement sensor, and the sensors can collect pressure, temperature and conductivity data of limb skin in real time. Because of their portability and comfort, wearable devices can be worn on the body for long periods of time, providing long-term and continuous monitoring. Such long-term monitoring is critical to tracking the progression of lymphedema, assessing the efficacy of the treatment, and adjusting the treatment regimen in time. In addition, the smart wristband or custom pressure sock can transmit the collected data to a smart device or medical center via wireless technology, enabling a doctor to remotely monitor the condition of the patient.
In conjunction with 3D imaging techniques, such as laser scanning or structured light scanning, to create an accurate three-dimensional model of the affected limb of the patient, this approach can provide detailed volumetric measurements and morphology change recordings of the edematous region. Using these high precision 3D imaging techniques, the precise shape and size of the affected limb can be captured, allowing doctors and medical professionals to not only measure the volume increase caused by lymphedema, but also to observe morphological changes in oedema over time. This is important for assessing the severity of oedema, monitoring the progression of the condition, and judging the effectiveness of the treatment. Furthermore, by periodically performing 3D scans, data at a series of time points can be collected, further supporting long-term monitoring and analysis of lymphedema progression.
The bio-resistance and capacitance of the limb are measured using non-invasive Bioelectrical Impedance Analysis (BIA), which is achieved by electrode patches attached to the limb surface. Bioelectrical impedance analysis is an effective method for assessing fluid distribution in the body, and is particularly useful for identifying and quantifying the accumulation of lymph fluid. During this process, the electrode patch produces a tiny, human-safe current through the limb. This method can be used to estimate the location and extent of fluid accumulation due to the different impedance (i.e., resistance and capacitance) of different tissues (e.g., muscle, fat, and fluid) to the current. In particular in the case of lymphedema, fluid accumulation in the body can lead to changes in electrical impedance, which can be detected and quantified by the BIA technique.
Integrated software platforms are used to fuse and analyze multi-source data, which may come from a variety of measurement techniques such as 3D imaging, bioelectrical impedance analysis, and the like. On the platform, a data processing algorithm, particularly a machine learning and pattern recognition technology is applied to comprehensively analyze various collected data. These algorithms can extract useful information from different types of data, identify characteristic patterns of lymphedema, and perform deep analysis of the data. For example, a machine learning model may learn from volume data of 3D imaging, conductance data of bioelectrical impedance, find characteristic markers of lymphedema, and predict changes in volume thereof. Finally, the results of these comprehensive analyses are integrated into a computational data model that accurately calculates and predicts the volume of lymphedema, providing important information to the physician regarding the progress of the condition. The method not only improves the accuracy and efficiency of lymphedema measurement, but also provides powerful support for lymphedema management and treatment through intelligent data processing.
Example 1: a patient, a woman, suffers from lymphedema and needs to monitor the edema of the affected right arm on a regular basis. In order to effectively monitor and record the pressure, temperature and conductance changes of its limb, a wearable device integrated with a multi-modal sensor array is used by a stretch woman.
The wearable equipment for the ladies comprises a pressure sensor, a temperature sensor and a conductivity measurement sensor, is miniaturized in design and flexible, and is convenient to cling to the skin for a long time. These sensors monitor in real time the pressure, temperature and conductance changes of her right arm, which is critical for assessing the status of lymphedema.
At some point, the rate of change of pressure recorded by the pressure sensorAt 0.2 units/hr, temperatureSensor recorded rate of temperature change +.>The conductivity change recorded by the conductivity sensor is 0.1 units/hour +.>Is 0.3 units/hour.
The formula applies:
wherein (a), (b) and (c) are adjustment coefficients obtained based on experimental data. These coefficients are (1.5), (2.0) and (1.0), respectively. Substituting the data, the edema rate R (t) calculated was 1.5×0.2+2.0×0.1+1.0×0.3=0.7 units/hour.
The present embodiment calculates the time from 8 am (t 0 ) To 5 pm (t) n ) Total edema volume variation V total 。
Using the integral formula:
the edema rate R (t) was kept constant during this period, and the total edema change calculated in this example was 0.7x9 (because of 9 hours) equal to 6.3 units.
In this example, a woman can continuously monitor her lymphedema status using her wearable device. By collecting data on pressure, temperature and conductance in real time, the formulas for calculating the rate of edema and the total amount of edema are combined.
Example 2: a patient suffering from lymphedema, grand woman, needs regular monitoring of her right leg for oedema. For this purpose, the medical team captures a three-dimensional model of the grand woman's right leg using structured light scanning techniques, and applies a calculation formula of volume and surface area and analysis of morphological changes.
First, a three-dimensional model of the right leg of a grand woman is captured using a structured light scanning technique.
The scan of this example shows that the affected area (D) of grand women is generally cylindrical, about 30cm in height and about 10cm in diameter.
The volume calculation formula is applied:
V＝∫∫∫ D f(x,y,z)dxdydz
for a simplified cylindrical model, the volume V can be calculated approximately as a cylindrical volume. The cylinder volume formula is (pi r 2 h) Where r is the radius and h is high.
Substituting data, radius r=5 cm, height h=30 cm, and volume v≡pi×5 2 X 30 is approximately 2356 cubic centimeters.
The formula is used:
A＝∫∫ S g(x,y,z)dS
to calculate the surface area a of the edematous zone.
For a cylindrical model, the surface area a can be calculated approximately as the cylindrical side area plus the top-bottom area.
The area formula of the side of the cylinder is 2 pi rh, and the area of the top and the bottom is 2 pi r 2 。
Substituting the data, the surface area A is approximately equal to 2pi×5×30+2pi×5 2 942 square cm.
Using a morphological change index formula:
(M＝∫∫ S h(K,H)dS)
in a simplified case, H (K, H) may be approximated as a constant, e.g., 1, and then M may be approximated as a surface area a.
Therefore, the morphological change index m≡942.
The right leg of grand women was then periodically scanned 3D to monitor and evaluate the volume and morphology changes of oedema.
By comparing the continuous scan data, the medical team can observe the trend of edema over time.
Example 3: a patient, called a Qin woman, is receiving treatment for lymphedema. In order to monitor her condition changes, medical teams have employed high sensitivity bioelectrical impedance analysis instruments with multi-frequency measurement capabilities for assessing fluid distribution and accumulation in the affected limbs of Qin women. The medical team uses bioelectrical impedance analysis instruments with multi-frequency measurement function to detect the affected limbs of Qin women.
The instrument measures the electrical impedance of the limbs of the Qin lady through the electrode patches and records data at different frequencies.
Applying a mathematical model:
Z(f)＝A(f)·e (jφ(f))
to analyze the measurement data.
In this embodiment, at a specific frequency, the measured amplitude a (f) is 50Ω, the phase angle Φ (f) is 0.5 radian, and the complex impedance Z (f) can be calculated.
The formula is used:
to adjust the measurement parameters of the electrode patch. In the subsequent measurement, the impedance change amount (ΔZ) was (5Ω), and the initial impedance value Z 0 (f) If the adjustment coefficients (K) and B (f) are set to 1 and 0.5, respectively, for 45Ω, then P (f, ΔZ) can be calculated.
Applying a formula;
to estimate the liquid distribution and accumulation, the present embodiment is performed over a frequency range (f 1 ) To (f) 2 ) Maximum electrical impedance value (Z) max (f) 60 Ω), the liquid accumulation level L (x, y, z) in the specific area can be estimated by integration.
In this example, by using multi-frequency bioelectrical impedance analysis techniques, the medical team can effectively assess the fluid distribution and accumulation of the affected limb of the Qin woman. By applying mathematical models and adaptive algorithms, they can accurately parse the data and adjust the measurement parameters, thereby obtaining more accurate results. Finally, by estimating the level of fluid accumulation through a computational formula, the medical team can better monitor the change of the disease of Qin women, providing her with a more personalized treatment regimen.
Example 4: patients with lymphedema are receiving treatment by a patient called a king woman. The medical team utilizes a comprehensive data processing platform that integrates the king women's 3D scan data, bioelectrical impedance data, and her medical history. The 3D scan data and bioelectrical impedance data of the king women, as well as her historical medical records, were collected.
Using deep neural network model F lymphedema (x) And (5) predicting.
In the model of this example, there are 3 neurons (n=3), weight w 1 ＝0.5,w 2 ＝1.0,w 3 Bias b=1.5 1 ＝0.1,b 2 ＝0.2,b 3 =0.3, the activation function (σ) is a Sigmoid function.
The integrated data (x) is set to (2.0) in this embodiment.
Substituting the formula to obtain: f (F) lymphedema (2.0)＝0.5·σ(0.5·2.0+0.1)+1.0·σ(1.0·2.0+0.2)+1.5·σ(1.5·2.0+0.3)。
Application of
The original data x in this embodiment is 3.0.
Substituting the formula to obtain
Pattern recognition and data visualization:
using V pattern (x) And (5) performing visualization.
The present embodiment (x) is in the range from (-2) to (2).
Substituting the formula to obtainIt is necessary to calculate by a numerical integration method.
In this example, by integrating the application of the data processing platform and the deep learning network, the medical team can effectively extract useful information from the multi-source data of the wang woman and predict her lymphedema volume. The application of the feature extraction function and the pattern recognition algorithm enables teams to present data in an intuitive manner to better understand the lymphedema development trend of king women.
Example 5: a patient is undergoing treatment for lymphedema. To more accurately monitor and predict his condition changes, medical teams developed a data processing platform that integrated multiple machine learning models.
Medical team integrates multiple machine learning models M k (x) Each model specifically handles a different feature and mode of data.
The platform of this embodiment integrates 3 models (k=3).
Using the formulaTo predict lymphedema volume.
The outputs of the three models of this embodiment are (2.0,1.5,3.0) (i.e., M 1 ＝2.0,M 2 ＝1.5,M 3 ＝3.0)。
Application ofWhere (λ) is a smoothing parameter, in this embodiment λ=1.
Feature conversion is performed on one original data point x=4, and calculation is performed
Then use the formula
To calculate the weight of each model. Error rate E of the three models of the present embodiment 1 ＝0.1,E 2 ＝0.2,E 3 =0.3, the weight of each model was calculated.
The element learner (L) is applied at the top level of the integrated model.
The present embodiment (L) can integrate the outputs of three modelsTo produce the final prediction F final (x)。/>
In this example, by integrating multiple machine learning models, a medical team can synthesize the features of the different models to more accurately predict the lymphedema volume of a patient. The feature transfer function and the strategy of dynamically assigning weights ensure efficient fusion of different data, while the application of the meta learner further improves the accuracy of the predictions.
Example 6: a patient with lymphedema, mr. Prune, receives regular 3D scans to monitor his oedema. To analyze these 3D scan data more accurately, medical teams use Convolutional Neural Networks (CNNs) and self-encoder techniques, as well as interactive 3D data visualization tools.
The medical team uses CNN to analyze mr. Prune's 3D scan images.
The CNN model of this embodiment includes 3 convolutional layers (n=3), the weights of each layer (W i ) And bias (b) i ) Has been determined by training. And inputting the 3D image (I) and extracting the characteristics of the edema region through CNN processing.
Then use the self-encoder AE features (x) =dec (Enc (x)) extracts key features.
The 3D image (x) is input from the encoder to obtain a compressed representation of the features, which is then used to reconstruct the data by the decoding process.
Next: medical teams developed an interactive 3D data visualization tool.
The chart update rules of the tool are represented by the formula:and (5) controlling. The time decay function gamma (t-tau) of this embodiment has been optimized based on historical data to balance recent and past observations.
In this example, by using CNN and self-encoder techniques, the medical team can accurately extract features of the edema region from mr. Prune's 3D scan images. These features are used to update interactive 3D data visualization tools, providing a dynamic view of lymphedema. This approach not only enhances the ability of the medical team to analyze and understand lymphedema, but also provides an intuitive, user-friendly way to demonstrate the progression of edema.
Example 7: a patient, a woman, is receiving treatment for lymphedema. To accurately monitor and evaluate her level of edema, medical teams use multidimensional Bioelectrical Impedance Analysis (BIA) techniques in combination with 3D imaging and analyze the fused data through a deep learning model.
S1, implementation of multidimensional Bioelectrical Impedance Analysis (BIA):
the medical team uses the BIA technique of the multi-frequency and multi-electrode configuration to measure the resistance value of the female.
The present embodiment is implemented at two different frequencies (f 1 ,f 2 ) And under the two electrode configurations, the measured resistance values are respectively:
R 1 (θ,f 1 )＝30Ω,R 2 (θ,f 2 )＝40Ω
application formula BIA multi (f,θ)＝∑ i γ i (f)·R i (θ, f) performing calculation.
S2.3D imaging and BIA data fusion:
3D imaging data of the female (V) 3D ) Fusion with BIA data.
Lymphedema volume V obtained by D imaging in this example 3 3D ＝2000cm 3 。
Fusion formula Fusion is applied 3D_BIA (V 3D ,BIA multi )＝δ·V 3D +(1-δ)·Conv(BIA multi ) Wherein δ=0.5.
S3, deep learning model application:
using deep learning model DL analysis (Fusion 3D_BIA )＝DeepNet(Fusion 3D_BIA ) Performing data analysis。
The deep neural network of this embodiment has been trained and is ready to process the fused data.
In this example, by combining multidimensional BIA technology with 3D imaging data, the medical team is able to more fully understand lymphedema status of the female. BIA technology provides detailed information about the liquid distribution, while 3D imaging gives accurate volume measurements. By fusing these two data, a medical team may obtain a more comprehensive edema analysis. The application of the deep learning model further enhances the accuracy of the data analysis, enabling the team to more effectively monitor and evaluate the trend of edema.
Example 8: a patient, a flood female, is receiving treatment for lymphedema. To accurately monitor and evaluate her edema status, medical teams employ Bioelectrical Impedance Spectroscopy (BIS) in combination with 3D imaging and analyze the fused data through a machine learning model.
S1, implementation of Bioelectrical Impedance Spectroscopy (BIS):
the medical team uses BIS technology to measure bioelectrical impedance of flood women at different frequencies.
Suppose that at three frequency points (f 1 ,f 2 ,f 3 ) The electrical impedance magnitudes under the test are respectively A 1 ＝50Ω,A 2 ＝60Ω,A 3 =55Ω, center frequency f 1 ＝100Hz,f 2 ＝200Hz,f 3 =300 Hz, standard deviation σ 1 ＝10Hz,σ 2 ＝20Hz,σ 3 ＝15Hz。
And calculating by applying a BIS formula.
S2.3D imaging data was fused with BIS data:
3D imaging data (V) of flood ladies 3D ) Fused with BIS data.
Let 3D imaging give a volume of lymphedema of V 3D ＝2500cm 3 。
Fusion using machine learning model 3D_BIS ＝ML model (V 3D BIS (f)) for data fusion and analysis.
In this example, by combining BIS technology with 3D imaging data, the medical team is able to more fully understand the lymphedema status of the flood woman. BIS technology provides detailed information about the liquid distribution, while 3D imaging gives accurate volume measurements. Through fusion analysis of machine learning models, a medical team may obtain a more comprehensive edema analysis. The application of the deep learning model further enhances the accuracy of the data analysis, enabling the team to more effectively monitor and evaluate the trend of edema.
The implementation steps of the lymphedema volume intelligent measurement method firstly comprise:
firstly, 3D imaging, namely, acquiring detailed three-dimensional forms of affected limbs by utilizing a structured light scanning or laser scanning technology, and providing necessary basic data for subsequent volume measurement; next is bioelectrical impedance analysis (BIA/BIS), a process involving measurement of multiple frequencies and multiple electrode configurations for accurately determining the resistance and capacitance of the limb, to infer the distribution of fluid in the body, particularly the extent of lymph accumulation.
In the lymphedema volume intelligent measurement method, the data processing and fusion stage comprises the steps of processing data obtained from 3D scanning, and estimating the volume of an affected area by using a volume calculation algorithm; secondly, processing bioelectrical impedance analysis (BIA/BIS) data to obtain bioelectrical impedance characteristics of the lymphedema region, particularly when BIS techniques are used, by analysing the data at different frequencies to provide more detailed body composition information; finally, the 3D imaging data and the BIA/BIS data are combined using machine learning models or other data fusion techniques to integrate the two data types, thereby providing a more comprehensive lymphedema analysis.
In the lymphedema volume intelligent measurement method, the intelligent analysis and prediction stage covers the application of deep learning and machine learning models, such as Convolutional Neural Networks (CNNs) or self-encoders, for detailed analysis of the fused data. The advanced models can effectively identify specific modes of lymphedema, predict the development trend of the disease, provide support for doctors and help them to make accurate diagnosis. Furthermore, by using machine learning models for pattern recognition and trend analysis, medical professionals can obtain accurate medical advice and treatment strategies regarding the development of lymphedema, thereby providing patients with more effective personalized treatment regimens.
In the lymphedema volume intelligent measurement method, the visualization and interaction phases mainly involve converting the measurement and analysis results into a graphical form for more visual presentation. This includes using interactive 3D models or Augmented Reality (AR) techniques to provide a clear, intuitive view, thereby helping doctors and patients to better understand the current state of lymphedema. In addition, the method provides an interactive platform that allows doctors to interact with the data, such as adjusting view angles or analyzing the data at different points in time, to gain more insight and better formulate treatment strategies.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made without departing from the spirit and scope of the invention, which is defined in the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (10)
1. An intelligent lymphedema volume measurement method is characterized by comprising the following steps of: comprising the following steps: firstly, using a multi-mode sensor array, including a pressure sensor, a temperature sensor and a conductivity measurement sensor, monitoring the pressure, temperature and conductivity changes of limb skin in real time, and providing preliminary data for physical characteristics of lymphedema; the sensor is integrated in the wearable device: smart bracelets or custom pressure socks for long-term and continuous monitoring; then combining 3D imaging techniques: laser scanning or structured light scanning, creating an accurate three-dimensional model for the affected limb of the patient, providing detailed volume measurement and morphological change recording of the oedema region; measuring the bioelectrical resistance and capacitance of the limb through the electrode patch on the surface of the limb by using a noninvasive bioelectrical impedance analysis technology to reveal the distribution of in-vivo liquid and the accumulation of lymph fluid; and finally, fusing and analyzing through an integrated software platform, and utilizing a data processing algorithm: and (3) machine learning and pattern recognition, and comprehensively analyzing the multi-source data to obtain a calculation data model of the lymphedema volume.
2. The method for intelligently measuring lymphedema volume according to claim 1, wherein the multi-modal sensor array includes integrated pressure sensors, temperature sensors and conductivity measurement sensors; the multi-mode sensor array uses microminiaturization and flexible design, so that the sensor array is suitable for being used for being clung to skin for a long time, and the pressure, temperature and conductivity changes of limbs are monitored and recorded;
by the formula for calculating the edema rate R (t):
wherein (P), (T) and (C) represent real-time readings of pressure, temperature and conductance, respectively, (a), (b) and (C) are coefficients derived based on experimental data for adjusting the extent to which pressure, temperature and conductance changes affect edema rate;
and calculate the total edema change (V) over a period of time total ) Is a function of the integral formula:
(V total ): in a time period (t 0 ) To (t) n ) Total edema volume change between;
is performed on the edema rate R (t) at time (t 0 ) To (t) n ) Integration of the period provides a cumulative change in the amount of edema over the period of time.
3. The method for intelligently measuring the volume of lymphedema according to claim 1, wherein the implementation of the 3D imaging technique includes: firstly, capturing a high-precision three-dimensional model of an affected limb by using a structured light scanning or laser scanning technology in combination with a machine learning algorithm; and secondly, applying a volume calculation formula:
V＝∫∫∫ D f(x,y,z)dxdydz
Wherein,
(V): the total volume of lymphedema area measured;
(∫∫∫ D ): triple integration of region D (i.e., the affected limb region) is performed for volume calculation;
f (x, y, z): this is a function used to describe the local nature of each point in three-dimensional space;
(dx dydz): these are tiny volume elements, tiny variations along the (x), (y) and (z) directions in three-dimensional space;
applying the surface area calculation formula:
A＝∫∫ S g(x,y,z)dS
wherein,
(A) The method comprises the following steps The total surface area of the lymphedema area being measured;
(∫∫ S ): double integration of the surface S (i.e. the surface of the affected area) is performed for calculating the surface area;
g (x, y, z): this is a function describing the local properties of each point on the surface in three dimensions, related to the curvature properties of the surface;
(dS): this is a tiny surface element, tiny areas on a three-dimensional surface;
next, the change in limb morphology is analyzed using differential geometry and topology principles, including calculating morphology change index formulas:
M＝∫∫ S h(K,H)dS
(M): a quantization index of morphological changes;
h (K, H): this is a function of the gaussian curvature K and the mean curvature H based on a curved surface, for describing the complexity or significance of the morphological changes;
(K) And (H): the Gaussian curvature and the average curvature of the curved surface are respectively described, and the bending degree of the curved surface at different points is described;
(dS): representing a tiny area on a curved surface;
finally volume and morphology changes of lymphedema were monitored and assessed by periodic 3D scan data analysis.
4. The intelligent lymphedema volume measuring method according to claim 1, wherein the non-invasive bioelectrical impedance analysis technical method comprises the steps of firstly using a bioelectrical impedance analysis instrument with high sensitivity, wherein the bioelectrical impedance analysis instrument adopts a multi-frequency measurement technology; then using the mathematical model formula:
Z(f)＝A(f)·e (jφ(f))
to analyze multifrequency bioelectrical impedance data, wherein (Z (f) is a complex impedance at frequency (f), A (f) is a frequency dependent magnitude function, phi (f) is a phase angle function, and further developing an adaptive algorithm:
to adjust the measured parameters of the electrode patch, wherein P (f, ΔZ) is the measured parameter adjustment value, (ΔZ) is the measured impedance variation, Z 0 (f) Is the initial impedance value, and (K) and B (f) are adjustment coefficients; for adjusting the response of the algorithm according to different measurement conditions;
finally, a calculation formula is applied:
to estimate the liquid distribution and accumulation,
wherein,
l (x, y, z): the level of fluid accumulation at three dimensional spatial points (x, y, z) is a quantification of fluid distribution and accumulation in the body;
For a certain frequency range (f 1 ) To (f) 2 ) Integrating the electrical impedance in the circuit to consider measurement information at different frequencies;
z (f): as before, electrical impedance at a particular frequency;
Z max (f) The method comprises the following steps Is the maximum electrical impedance value at frequency (f) for normalizing the electrical impedance measurements.
5. The method for intelligently measuring the volume of lymphedema according to claim 1, wherein the software platform performs fusion and analysis methods, including processing data from a plurality of data sources: a data processing platform comprising 3D scan data, bioelectrical impedance data, and patient medical history; the processing platform applies a deep learning network and a high-level analysis model, in particular a custom deep neural network model:
for predicting the volume of lymphedema from the integrated data (x),
F lymphedema (x) The method comprises the following steps Outputting a predictive function of the lymphedema volume as a predicted lymphedema volume;
(x) The method comprises the following steps Input data, herein referred to as a feature set that integrates the multi-source data;
summing the outputs of all neurons;
(w i ): weights of (i) th neurons in the neural network;
(sigma): an activation function for introducing nonlinearities that enable the neural network to learn more complex patterns;
(a i ·x+b i ): (i) th Weighted input of neurons, where (a i ) Is a weight (b) i ) Is biased;
in addition, the method for fusing and analyzing the software platform further comprises a feature extraction function:
for converting the raw data into meaningful feature vectors,
e (x): extracting a function of the features from the raw data (x);
this is a logic function for converting the raw data into values between 0 and 1, making it suitable for input to the machine learning model;
in addition, the method for fusing and analyzing the software platform further comprises a mode identification and data visualization algorithm:
for identifying and visually presenting the pattern of lymphedema,
V pattern (x) The method comprises the following steps Visualization of the progression pattern of lymphedema;
is an integral expression, wherein ∈>Is a gaussian function for smoothing and weighting the features E (x); this expression helps identify patterns in the data and convert them into a form that is visualized.
6. According to the weightsThe intelligent lymphedema volume measurement method according to claim 5, wherein the further method implemented by the deep neural network model comprises: first, a data processing platform integrating a plurality of machine learning models is developed, wherein each model M k (x) Specialized processing of different features and patterns of data; the output of the integrated model is given by a weighted sum, namely:
Wherein F is ensemble (x) The method comprises the following steps This is the final output of the integrated model, representing a predicted value of lymphedema volume;
summing the outputs of all (K) different machine learning models;
(α k ): a weight of the (k) th model, the weight being dynamically assigned according to the behavior of the model;
M k (T (x)): the output of the (k) th machine learning model, where T (x) is a feature transfer function applied to the input data (x);
the raw data is then converted into a form more suitable for model learning using a feature transfer function T (x):
wherein T (x): the characteristic conversion function is applied to the original input data (x) and is used for improving the applicability of the data and the prediction performance of the model;
(x 2 +λ): this is a conversion formula in which (x 2 ) The square of the original data increases the nonlinearity of the data, and (lambda) is a smoothing parameter, avoiding the error of dividing by zero;
and then dynamically distributing weights according to the performance of each model in the training process, wherein the specific formula is as follows:
wherein (alpha) k ): weights of the (k) th model;
exp(-E k ): is an exponential function for the error rate (E) of the conversion model (k) k ) Is the weight;
the exponential transformation summation of all model errors is used for normalizing the weights;
the meta learner (L) is applied at the top level of the integrated model to integrate the outputs of the different models as follows:
Wherein,is the output of the (k) th integration model, and (L) is a meta learner for integrating these outputs.
7. The method for intelligently measuring the volume of lymphedema according to claim 5, wherein the mode identification and data visualization algorithm further comprises the steps of: first, a deep learning analysis is performed on a 3D scan image using a Convolutional Neural Network (CNN) to accurately identify and distinguish lymphedema areas from normal tissues; the CNN model is described by the following formula:
CNN lymphedema (I) The method comprises the following steps Is used for analyzing showerThe output of the CNN model of baredemas, representing the lymphedema region identified from the 3D image (I);
summing all (n) convolution layers;
(W i ): is the weight of the (i) th convolutional layer;
(*): convolution operation, which is used for extracting the characteristics in the image;
(sigma): is an activation function for introducing nonlinearity;
(I) The method comprises the following steps Is an input 3D image;
(b i ): is the bias term for the (i) th convolutional layer;
secondly, a self-encoder technique is adopted: AE (AE) features (x) =dec (Enc (x)) to extract key features in 3D images;
wherein AE features (x) The method comprises the following steps Is the output from the encoder for extracting lymphedema features;
enc (x): converting the input 3D image (x) from the encoding part of the encoder into a more compact feature;
(Dec): reconstructing data from the encoded features from a decoding portion of the encoder;
next, an interactive 3D data visualization tool is developed whose chart update rules are controlled by the following formula:
(VizUpdate (x, t)): a visual state at a point in time (t);
integrating the data in the time window to consider the history information;
gamma (t-tau): is a time decay function for giving different weights to the data at different points in time;
CNN lymphedema (x (τ)): is a CNN model applied on a 3D image x (τ) of time (τ).
8. The method for intelligently measuring the volume of lymphedema according to claim 3 or 4, wherein the bioelectrical impedance analysis and 3D imaging are implemented according to another technical scheme, and the method comprises the following key steps: first, a multidimensional Bioelectrical Impedance Analysis (BIA) technique is implemented that utilizes a multi-frequency and multi-electrode configuration to accurately measure the bioelectrical resistance value, as follows:
wherein,
BIA multi (f, θ): this is the output of a multi-dimensional bioelectrical impedance analysis, which represents the integrated resistance measurements at different frequencies (f) and electrode configurations (θ);
(∑ i ): summing the resistance measurements of all electrode configurations;
γ i (f) The method comprises the following steps Is a weight function of the (i) th electrode configuration at frequency (f) for adjusting the influence of the resistance values at different frequencies;
R i (θ, f): resistance measurements at a particular electrode configuration (θ) and frequency (f);
secondly, fusing 3D imaging technology with BIA data to provide comprehensive lymphedema volume information, wherein the fusion formula is as follows:
Fusion 3D_BIA (V 3D ,BIA multi )＝δ·V 3D +(1-δ)·Conv(BIA multi )
wherein,
(Fusion 3D_BIA ): this is the output of the 3D imaging data and bioelectrical impedance data fused, providing comprehensive lymphedema volume information;
(V 3D ): is lymphedema volume data acquired from 3D imaging techniques;
Conv(BIA multi ): is to make bioelectrical impedanceAnalyzing the data to convert to a convolution function of the volume estimate;
(delta): is a fusion coefficient for balancing the contributions of the 3D imaging data and the BIA data;
finally, applying the deep learning model
DL analysis (Fusion 3D_BIA )＝DeepNet(Fusion 3D_BIA )
The data analysis is performed and the data is analyzed,
wherein:
(DL analysis ): the data analysis output of the deep learning enhancement is used for extracting key characteristics and trends of lymphedema from the fusion data;
DeepNet(Fusion 3D_BIA ): is a deep neural network applied to process and analyze the fused data (Fusion 3D_BIA )。
9. The method for intelligently measuring the volume of lymphedema according to claim 3 or 4, wherein the bioelectrical impedance analysis and 3D imaging further comprises the following key steps: first, bioelectrical Impedance Spectroscopy (BIS) is performed, which provides detailed information about body composition and fluid distribution by measuring bioelectrical impedance at a series of different frequencies; the mathematical expression of BIS is:
Wherein,
BIS (f): bioelectrical impedance spectroscopy at frequency (f);
summing all the frequency points (N) to obtain an overall bioelectrical impedance spectrum;
(A k ): an amplitude parameter of a (k) th frequency point, the magnitude of the electrical impedance at that frequency;
a Gaussian function describing the frequency (f k ) A nearby electrical impedance change, where (f k ) Is the center frequency (sigma) k ) Is the standard deviation of the frequency point;
then, a machine learning Model (ML) is used in combination with the 3D imaging data and the BIS data model ) And (3) carrying out data fusion and analysis, wherein the formula is as follows:
Fusion 3D_BIS ＝ML model (V 3D ,BIS(f))
(Fusion 3D_BIS ): outputting fused 3D imaging data and BIS data, namely comprehensively analyzing the volume of lymphedema;
(ML model ): is a machine learning model for data fusion responsible for combining 3D imaging data (V 3D ) And BIS data BIS (f) to provide a final analysis result;
(V 3D ): volumetric data of lymphedema region obtained from 3D imaging techniques;
BIS (f): is the result of bioelectrical impedance spectroscopy analysis, providing detailed electrical impedance information about the lymphedema region.
10. An intelligent lymphedema volume measuring method according to claims 8-9, comprising the specific steps of:
s1, data collection:
3D imaging: 3D imaging using structured light scanning or laser scanning techniques to capture detailed three-dimensional morphology of the affected limb; this provides the basis data for subsequent volume calculations;
Bioelectrical impedance analysis (BIA/BIS): bioelectrical impedance analysis, including multi-frequency and multi-electrode configurations, is used to measure the resistance and capacitance of the limb to infer fluid distribution in the body, particularly lymph fluid accumulation;
s2, data processing and fusion:
3D volume calculation: processing data obtained from the 3D scan, estimating the volume of the affected region using a volume calculation algorithm;
BIA/BIS data analysis: processing BIA/BIS data to obtain bioelectrical impedance characteristics of the lymphedema region; in the case of BIS, data at a series of frequencies are analyzed to provide more detailed body composition information;
data fusion: combining the 3D imaging data and the BIA/BIS data, integrating the two types of data using machine learning models or other data fusion techniques to provide a more comprehensive lymphedema analysis;
s3, intelligent analysis and prediction:
deep learning/machine learning model application: applying a deep learning model, such as a Convolutional Neural Network (CNN) or a self-encoder, to analyze the fused data; the models can identify the mode of lymphedema, predict the development trend of lymphedema and assist doctors in diagnosis;
pattern recognition and trend analysis: identifying a specific mode and a development trend of lymphedema by using a machine learning model, and providing accurate medical advice and treatment strategies;
S4, visualization and interaction:
data visualization: presenting the measurement and analysis results in a graphical form, providing an intuitive view using interactive 3D models or Augmented Reality (AR) techniques, helping doctors and patients to better understand the status of lymphedema;
interactive feedback: a platform is provided to enable a physician to interact with the data, such as adjusting view angles, analyzing the data at different points in time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311726197.2A CN117694864A (en) | 2023-12-14 | 2023-12-14 | Lymphedema volume intelligent measurement method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311726197.2A CN117694864A (en) | 2023-12-14 | 2023-12-14 | Lymphedema volume intelligent measurement method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117694864A true CN117694864A (en) | 2024-03-15 |
Family
ID=90147482
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311726197.2A Pending CN117694864A (en) | 2023-12-14 | 2023-12-14 | Lymphedema volume intelligent measurement method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117694864A (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150302594A1 (en) * | 2013-07-12 | 2015-10-22 | Richard H. Moore | System and Method For Object Detection Using Structured Light |
| US20160331314A1 (en) * | 2015-05-15 | 2016-11-17 | The Florida International University Board Of Trustees | Materials and methods for detecting and monitoring edema and related conditions |
| CN106175776A (en) * | 2016-07-13 | 2016-12-07 | 南京医科大学第附属医院 | A kind of lymphedema limbs volume measurement device |
| CN107122593A (en) * | 2017-04-06 | 2017-09-01 | 复旦大学 | Upper limb lymphedema monitoring system based on depth scan and information analysis |
| CN111163693A (en) * | 2017-06-28 | 2020-05-15 | 瑞桑尼有限公司 | Customization of health and disease diagnostics |
| CN111202523A (en) * | 2020-03-03 | 2020-05-29 | 浙江省肿瘤医院 | Visual measuring equipment for upper limb lymphedema after tumor operation |
| US20220125666A1 (en) * | 2020-10-23 | 2022-04-28 | Board Of Supervisors Of Louisiana State University And Agricultural And Mechanical College | Devices and methods for lymphedema treatment |
| CN116897023A (en) * | 2021-01-22 | 2023-10-17 | 西拉格国际有限公司 | Predicting tissue irregularities based on biomarker monitoring |
| WO2023212763A1 (en) * | 2022-05-03 | 2023-11-09 | Falko Skrabal | Diagnostic system |
| CN117043874A (en) * | 2021-01-22 | 2023-11-10 | 西拉格国际有限公司 | Predictive system adjustment based on biomarker trends |
-
2023
- 2023-12-14 CN CN202311726197.2A patent/CN117694864A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150302594A1 (en) * | 2013-07-12 | 2015-10-22 | Richard H. Moore | System and Method For Object Detection Using Structured Light |
| US20160331314A1 (en) * | 2015-05-15 | 2016-11-17 | The Florida International University Board Of Trustees | Materials and methods for detecting and monitoring edema and related conditions |
| CN106175776A (en) * | 2016-07-13 | 2016-12-07 | 南京医科大学第附属医院 | A kind of lymphedema limbs volume measurement device |
| CN107122593A (en) * | 2017-04-06 | 2017-09-01 | 复旦大学 | Upper limb lymphedema monitoring system based on depth scan and information analysis |
| CN111163693A (en) * | 2017-06-28 | 2020-05-15 | 瑞桑尼有限公司 | Customization of health and disease diagnostics |
| CN111202523A (en) * | 2020-03-03 | 2020-05-29 | 浙江省肿瘤医院 | Visual measuring equipment for upper limb lymphedema after tumor operation |
| US20220125666A1 (en) * | 2020-10-23 | 2022-04-28 | Board Of Supervisors Of Louisiana State University And Agricultural And Mechanical College | Devices and methods for lymphedema treatment |
| CN116897023A (en) * | 2021-01-22 | 2023-10-17 | 西拉格国际有限公司 | Predicting tissue irregularities based on biomarker monitoring |
| CN117043874A (en) * | 2021-01-22 | 2023-11-10 | 西拉格国际有限公司 | Predictive system adjustment based on biomarker trends |
| WO2023212763A1 (en) * | 2022-05-03 | 2023-11-09 | Falko Skrabal | Diagnostic system |
Non-Patent Citations (1)
| Title |
|---|
| 俞海平 等；: "肢体淋巴水肿磁共振淋巴造影技术方法的实验研究", 《生物磁学》, vol. 05, no. 04, 31 December 2005 (2005-12-31), pages 8 - 11 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6280395B1 (en) | System and method for determining muscle dysfunction | |
| Xue et al. | Full-field wrist pulse signal acquisition and analysis by 3D Digital Image Correlation | |
| CA3069891A1 (en) | Discovering novel features to use in machine learning techniques, such as machine learning techniques for diagnosing medical conditions | |
| KR20200006447A (en) | Method and program for stroke volume acquisition using arterial pressure waveform | |
| CN112806977B (en) | Physiological parameter measuring method based on multi-scale fusion network | |
| KR101576526B1 (en) | Apparatus and Method for Supporting Rehabilitation Exercise | |
| Hajian et al. | Deep multi-scale fusion of convolutional neural networks for EMG-based movement estimation | |
| Chen et al. | Machine learning method for continuous noninvasive blood pressure detection based on random forest | |
| CN111588353A (en) | Body temperature measuring method | |
| Shi et al. | Feasibility of controlling prosthetic hand using sonomyography signal in real time: Preliminary study | |
| CN110752032A (en) | Traditional Chinese medicine diagnosis method based on convolutional neural network and laser vibration measurement | |
| CN110575178B (en) | Diagnosis and monitoring integrated medical system for judging motion state and judging method thereof | |
| CN109545374A (en) | A kind of disease based on body gait feature big data automatically tracks, treats and supplying system | |
| Ostadabbas et al. | A knowledge-based modeling for plantar pressure image reconstruction | |
| Tong et al. | Sensitivity analysis of wearable textiles for ECG sensing | |
| CN105796091B (en) | A kind of intelligent terminal for removing electrocardiosignal vehicle movement noise | |
| Pombo et al. | Big data reduction using RBFNN: A predictive model for ECG waveform for eHealth platform integration | |
| Johns et al. | Force modelling of upper limb biomechanics using ensemble fast orthogonal search on high-density electromyography | |
| CN116269738B (en) | Intelligent control method, device, equipment and storage medium of radio frequency therapeutic apparatus | |
| CN117694864A (en) | Lymphedema volume intelligent measurement method | |
| WO2023119103A1 (en) | Method for analyzing breast signs on basis of orderliness, multiple modalities and symmetry breaking | |
| Eun et al. | Development of intelligent healthcare system based on ambulatory blood pressure measuring device | |
| Martinez et al. | Strategic attention learning for modality translation | |
| CN111387939B (en) | Method, device, computer equipment and storage medium for measuring heart displacement | |
| CN113520413B (en) | Lower limb multi-joint angle estimation method based on surface electromyogram signals |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |