WO2023119103A1 - 基于有序性、多模态及对称性破缺对乳房体征分析的方法 - Google Patents

基于有序性、多模态及对称性破缺对乳房体征分析的方法 Download PDF

Info

Publication number
WO2023119103A1
WO2023119103A1 PCT/IB2022/062441 IB2022062441W WO2023119103A1 WO 2023119103 A1 WO2023119103 A1 WO 2023119103A1 IB 2022062441 W IB2022062441 W IB 2022062441W WO 2023119103 A1 WO2023119103 A1 WO 2023119103A1
Authority
WO
WIPO (PCT)
Prior art keywords
sign
abnormal
signs
data
specific
Prior art date
Application number
PCT/IB2022/062441
Other languages
English (en)
French (fr)
Inventor
陈耀邦
康宏
Original Assignee
香港生物节律研究院有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 香港生物节律研究院有限公司 filed Critical 香港生物节律研究院有限公司
Publication of WO2023119103A1 publication Critical patent/WO2023119103A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/43—Detecting, measuring or recording for evaluating the reproductive systems
    • A61B5/4306—Detecting, measuring or recording for evaluating the reproductive systems for evaluating the female reproductive systems, e.g. gynaecological evaluations
    • A61B5/4312—Breast evaluation or disorder diagnosis
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/01—Measuring temperature of body parts ; Diagnostic temperature sensing, e.g. for malignant or inflamed tissue
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Definitions

  • the invention relates to a method for analyzing mammary gland signs, in particular to a method for analyzing mammary gland signs based on orderliness, multi-modality, symmetry breaking and the like.
  • CBE Clinical breast examination
  • BUS Ultrasonography
  • Breast line magnetic resonance imaging (breast MRI) MR imaging technology has the characteristics of excellent soft tissue resolution and no radiation, and has unique advantages for breast line inspection, especially with the development of special breast line coils and fast imaging sequences 4.
  • Digital breast tomography digital breast tom synthesis, DBT
  • DBT digital breast tom synthesis
  • the new tomographic imaging technology developed on the basis of photography principles combined with digital image processing technology can quickly collect milk lines from a series of different angles, obtain small-dose projection data at different projection angles, and retrospectively reconstruct the plane with the detector. X-ray density image of any depth slice parallel to the mammary line.
  • the present invention provides a method for analyzing breast signs based on orderliness, multi-modality and symmetry breaking, by combining various mathematical models to analyze and process the measured sign data , in order to be able to analyze and obtain the rule of mammary gland signs changing over time from a large amount of sign data with measurement noise.
  • the present invention provides the following technical solutions:
  • a method for analyzing breast signs based on order, multi-modality, and symmetry breaking including setting a variety of sign data sensors on the breast parts of the evaluation object, and continuously collecting various signs of the breast parts within a specified time interval through the sign data sensors.
  • Preliminary analysis is performed on the collected sign data and a corresponding weight value is assigned to each sign value of each sign data;
  • the specific steps for finding and evaluating whether there are abnormal confirmation features of specified signs in the corresponding abnormal rhythm model in multiple specific sign change equations or specific sign change curves of the subject's mammary gland are:
  • abnormal confirmation feature corresponding to this sign If it is confirmed that the abnormal confirmation feature corresponding to this sign is found, then within the specified time window corresponding to the abnormal confirmation feature, look for the specific sign change equation or specific sign change curve corresponding to another sign. Abnormal confirmatory features for specified signs in the abnormal rhythm model; repeat until complete finding abnormal confirmatory features for all signs.
  • the characteristic sign change equation of the evaluated object's mammary gland is fitted with the standard rhythm model to obtain the corresponding parameter values of the equation; an evaluation model is established based on the change of mammary gland signs under abnormal conditions, and the above parameter values are input into the evaluation model for evaluation. Calculate and output the result.
  • the sign measuring devices are symmetrically placed on the breast surfaces on both sides of the human body; each sign measuring device is equipped with at least two similar sign sensors, and each sign sensor independently collects its sign data, and establishes a specific sign change corresponding to it Equations and/or curves for specific signs.
  • the specific sign change equation or the specific sign change curve obtained from the characteristic data collected by the same sign sensor at the same specified time interval in different rhythm cycles is compared, and the above comparison value is input into the evaluation model as an input value for calculation.
  • the specific sign change equations or specific sign change curves corresponding to different sign sensors on the same side of the breast are compared to obtain a corresponding comparison value at a certain moment or an integral comparison value of a specified time interval, and input the above comparison value as an input value to evaluate the model.
  • the specific sign change equations or specific sign change curves corresponding to the sign sensors corresponding to the breast positions on both sides are compared to obtain a corresponding comparison value at a certain moment or an integral comparison value of a specified time interval, and the above comparison value is used as an input Values are input to the evaluation model for computation.
  • an intelligent orderly multi-modal dynamic mapping operation is performed on the specific sign change equation or characteristic sign change curve, so that the specific sign change equation or characteristic The sign change curve is mapped to the abnormal rhythm model.
  • an intelligent orderly multimodal dynamic mapping operation is performed on the specific sign change equations or characteristic sign change curves, so that multiple specific sign change equations or characteristic The sign change curves are mapped to each other.
  • the method of assigning the weight value is: comparing the currently collected sign data with the historically collected sign data, and calculating the change trend and rate of change of the current sign data relative to the historical sign data; A higher weight value assigns a lower weight value to the sign data with a decreasing change trend; a higher weight value is assigned to the sign data with a smaller change rate, and a lower weight value is assigned to the sign data with a larger change rate.
  • the method of finding the abnormal confirmation features and/or abnormal exclusion features of the specified signs in the abnormal rhythm model in the specific sign change equation or the specific sign change curve is as follows:
  • the analysis results of different signs are analyzed by orderly multi-sign integration to distinguish whether the related signs belong to the dependent variable or the independent variable, and calculate the time-ordered distance between the related signs for the dependent variable.
  • the intelligent and orderly multi-modal dynamic mapping operation is to map each sign data collected by the evaluation object with the corresponding abnormality confirmation features or abnormality exclusion features in the standard rhythm model and abnormal rhythm model, so that the sign data
  • the rate and magnitude of change are adapted to the corresponding abnormality confirmation feature or abnormality elimination feature.
  • the intelligent orderly multimodal dynamic mapping operation is to perform a mapping operation on the specific sign change equation and/or the specific sign change curve to be compared, so that the change rate of the specific sign change equation and/or the specific sign change curve correspond to the magnitude.
  • the senor installed on the breast collects various physical sign data, including but not limited to the temperature and humidity of the breast. Because breast cancer cells will cause changes in corresponding signs, these subtle changes are easily covered by noise generated by various human activities or different states, and in the past, only a single sign data, such as temperature, is used for analysis, which is difficult to carry out Accurate analysis.
  • the present invention collects a variety of physical sign data and establishes a variety of analysis modes correspondingly, so that the analysis of different modes can be more accurate and effective.
  • the orderliness mentioned in the present invention means that the data of various signs of the human body, especially the mammary gland, will show a certain periodicity over time, and the order in which some signs appear will also have a sequential relationship.
  • Sexual sign data includes both normal and abnormal conditions. Through orderly analysis, it is possible to find out whether there are signs in an abnormal state.
  • Multi-modality refers to the collection of different human body sign data through different sensors, and the corresponding establishment of multiple analysis modes. Since each mode has its unique order, and when an abnormal state occurs, different modes will produce corresponding abnormal features, and the order in which some modes appear will also have a sequential relationship. The abnormal feature analysis of each mode can effectively improve the accuracy of abnormal judgment.
  • the analysis method for multimodality is as follows:
  • the abnormal confirmation feature of the specified sign in the abnormal rhythm model corresponding to the sign is found in the corresponding specific sign change equation or specific sign change curve. If it is confirmed that the abnormal confirmation feature corresponding to this sign is found, it means that there is a possibility of abnormality. But judging only by one mode, the accuracy rate is relatively low.
  • an abnormal confirmation feature corresponding to a specified sign in the abnormal rhythm model If the abnormal confirmation feature of the other sign cannot be found, it means that the probability of abnormality is reduced. On the contrary, if the corresponding abnormal confirmation feature is also found in another sign, the probability of abnormality increases.
  • the abnormal confirmation features of all signs are found. That is to say, in different modalities, that is, in different sign data, the more corresponding abnormal confirmation features are always found, the higher the probability of evaluating the presence of abnormal breasts; otherwise, the lower the probability. Through multi-modal analysis, the accuracy of judgment can be greatly improved.
  • the current multimodality can include the following human body signs: temperature and humidity, and the above-mentioned modalities should also include the sign data collected by similar sensors in different positions in the same cycle or in different cycles, and different types of sensors in the same position. Sign data collected by sensors in the same period or in different periods.
  • the symmetry generally refers to that when the human body is in a normal state, that is, in a non-abnormal state, the physical sign data of the human body are orderly, that is, in the corresponding cycle, these characteristic data are symmetrical.
  • the sign data collected by a pair of mammary glands in the same period are also symmetrical.
  • the sign data collected from the same side of the breast in the same time period such as from 01:00 to 04:00 in the morning, but in different time intervals, such as three months to one year, also have symmetry.
  • the above symmetrical sign data will be asymmetrical, resulting in a broken symmetry.
  • Including mammary glands in different rhythm cycles will be produced, and by analyzing the symmetry breaking, it is also possible to judge whether there is an abnormal basis.
  • the corresponding sign sensor can be set in the bust so that the user can wear it at all times and collect data in real time.
  • the collected data can be sent to the server through wireless connection or other means for processing and analysis.
  • factors such as wearing comfort, it is impossible for the sensor to be close to the breast all the time. For example, the user's movement or posture change may cause the sensor to detach from the breast surface, resulting in errors or even errors in the collected data.
  • a preliminary analysis is performed on the collected sign data and a corresponding weight value is assigned to each sign value.
  • the weight value of the collected sign data can be assigned according to the above rules, and the degree of influence of the sign data on the subsequent data processing can be judged or determined by the weight value.
  • the data with a high weight value is relatively reliable or accurate data
  • those with low weight values are data with errors or errors.
  • the assignment rules of the weight value are as follows: compare the currently collected sign data with the previously collected sign data, and calculate the change trend and rate of change of the current sign data relative to the previous sign data; On the contrary, a lower weight value is assigned to the sign data with a decreasing change trend; a higher weight value is assigned to the sign data with a smaller change rate, and a lower weight value is assigned to the sign data with a larger change rate.
  • the various signs of the human body are orderly in the corresponding rhythm cycle, that is, by collecting the signs in a certain rhythm cycle of the human body under normal circumstances, and establishing a standard rhythm model of the sign, then real-time sampling and The sign data to be analyzed with weighted values are fitted with the standard rhythm model, , among
  • Fourier transform, Gaussian mixture model, wavelet transform, cosine fitting, sine fitting, etc. can be used, but not limited to, is the time series, is the sign data series, To fit the relevant parameters, is the error value.
  • a specific sign change equation and/or a specific sign change curve of the subject's mammary gland is obtained.
  • the specific sign change equation or specific characteristic change curve is used for subsequent analysis to find whether there are abnormal features in the above equation or curve.
  • the signs of abnormal human tissues will also show order in the corresponding rhythm cycle, so an abnormal rhythm model can also be established for breast signs under abnormal conditions. And usually there will be some abnormal features in the abnormal rhythm model, and there will also be features that are used to rule out abnormalities. Comparing the specific sign change equation or the specific characteristic change curve with the abnormal rhythm model and looking for abnormal confirmation features in the specific sign change equation or the specific characteristic change curve. Also by looking for abnormal features in the specific sign change equation or specific feature change curve.
  • the inventor In view of the huge amount of various sign data obtained through continuous sampling, a large number of data calculations are required in the above-mentioned analysis and judgment process.
  • the inventor also designed an evaluation model through a computer, and the evaluation model uses an artificial neural network for calculation.
  • the corresponding parameter values, abnormality confirmation features, and abnormality exclusion features of the equation are obtained after the fitting operation of the characteristic sign change equation of the breast of the evaluation object and the standard rhythm model.
  • the sign measurement devices are symmetrically placed on the breast surfaces on both sides of the human body; each sign measurement device is equipped with at least two similar sign sensors, and each sign sensor collects its sign data independently, and establishes a specific sign change equation corresponding to it and/or Or a specific sign change curve.
  • the description of symmetry breaking includes a variety of situations, such as the comparison of the data of the same sign obtained from different positions of the same breast, the comparison of the data of the same sign of different breasts (as shown in Figure 5), and the comparison of the data of the same sign Comparison of the characteristic data of the sensor in the same specified time interval in different rhythm cycles to determine whether there is a symmetry break.
  • the present invention also provides corresponding comparison and search methods, which can be implemented in the following ways, but not limited to:
  • the data in the data window and the jth position in the sign data matrix correspond to the sign data of length m , calculate the similarity value, and move the data window backward from the jth position and calculate to obtain the distance similarity matrix , where d i , j is the calculated similarity value between T i, m and T j, m , m, the smaller the value, the more similar, where j is a positive integer, 1 ⁇ j ⁇ n-m+1;
  • the data window moves from the first sign data of the sign data matrix to the n-m+1th, take the minimum value of each distance matrix, and obtain the minimum distance matrix ,in is the above distance similarity matrix, min( )for The minimum value in .
  • the repeated features in the sign data matrix are obtained by the minimum distance matrix; the repeated features in the sign data matrix are obtained by the minimum distance matrix;
  • FIG. 1 shows the specific practice of finding repeated features in data according to the method described above.
  • the above repeated features are compared and/or confirmed with the abnormality confirmation features and/or the abnormality exclusion features as described in S2-5.
  • This algorithm judges and finds whether the data group contains abnormal rhythm models from the data to be mined in an intelligent and effective way and makes a screening Mining, the algorithm allows the difference in speed and amplitude in the mapping comparison process but the same mode, such as , using algorithms to intelligently mine the approximate modes from the data in a dynamic mapping manner to mine relevant data blocks.
  • the method of mining relevant data blocks can be calculated and obtained by the following methods, but is not limited to the following methods. The following specific examples are used to describe the above.
  • the digging path starts at , the end point of the mining path is ;
  • Figure 2 illustrates the path planning mechanism.
  • mapping operation is applied to multimodality, the following steps should also be performed:
  • the start time of finding the minimum distance path in the matrix W corresponding to the specific sign change equation and/or the specific sign change curve is , based on orderly and multimodal considerations, another modality
  • the modality factors at the following times period occurs, and is a valid modal, the other modal
  • the comparison method can be based on steps a, b, c, and d, as shown in Figure 4, where the modality, that is, the number of signs , , .

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Veterinary Medicine (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Engineering & Computer Science (AREA)
  • Public Health (AREA)
  • Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Gynecology & Obstetrics (AREA)
  • Reproductive Health (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

一种基于有序性、多模态及对称性破缺对乳房体征分析的方法，通过设置在乳房的传感器收集多种体征数据，包括但不限于乳房的温度、湿度等。由于乳腺癌细胞会引起相应的变化，这些细微变化很容易被各种人体活动或不同的状态所产生的噪声所掩盖，并且以往只会使用单一的体征数据进行分析，如温度，很难进行精确分析。本方法通过收集多种体征数据，相应的建立多种分析模态，对不同的模态进行分析能够更加准确及有效。

Description

基于有序性、多模态及对称性破缺对乳房体征分析的方法
本发明涉及一种对乳腺的体征进行分析的方法，尤其是基于有序性、多模态以及对称性破缺等对乳腺的体征进行分析的方法。
乳腺疾病，尤其是乳腺癌，其发病率位居女性恶性肿瘤的首位。然而早期乳线癌的治愈率可以高达95%。通过早期的乳线癌筛查可以及早发现、诊断并治疗，以降低乳线癌的死亡率。目前常用筛查方法包括1.临床乳线检查（clinical breast examination，CBE）CBE简便、易行，可重复性强，但敏感度低，受主观因素影响较大；2.乳线超声检查（breast ultrasonography，BUS）BUS能较好地显示乳线肿块的特征，可鉴别在X线片上看不到，但可触及的肿物，也可用于不适用mammogram ，MAM检查的女性（如年轻女性和孕妇等），同时也适用于致密型乳线。3.乳线磁共振成像（breast MRI）MR成像技术具有极好的软组织分辨力和无辐射等特点，对乳线检查具有独到的优势，特别是随着专用乳线线圈及快速成像序列的开发应用，使乳线MR影像质量及诊断水平有了很大提高；4.数字乳线断层合成技术(digital breast tom synthesis，DBT)是一项基于平板探测器技术的高级应用，是在传统体层摄影原理的基础上结合数字影像处理技术开发的新型体层成像技术，通过一系列不同角度对乳线进行快速采集，获取不同投影角度下的小剂量投影数据，可回顾性重建出与探测器平面平行的乳线任意深度层面X线密度影像。
但是以上这些筛查方法的最大缺陷就是女性必须要去医院进行检查，而且检测成本较高。其中除了基因检测之外，其他成像检查技术还存在对致密性乳腺漏检的风险。近年，在对乳线癌的进一步研究中发现，由于乳线癌细胞的存在，从而引起乳腺细胞体征的节律发生异常。通过比较癌组织中节律变化模式与健康组织的差异，可以实现对乳线细胞是否存在癌变的风险进行监测。但是如何能够连续精确监测乳线细胞的体征变化存在技术壁垒，其中包括人体体征会存在变化周期、不同生理状态导致体征的变化、体征测量设备与人体之间的适配等等因素。
针对上述现有技术的不足，本发明提供了一种基于有序性、多模态及对称性破缺对乳房体征分析的方法，通过结合多种数学模型对测量获得的体征数据进行分析及处理，以便能够从大量带有测量噪声的体征数据中分析获得乳腺的体征随时间变化的规律。
为实现上述目的，本发明提供以下技术方案：
基于有序性、多模态及对称性破缺对乳房体征分析的方法，包括在评估对象的乳房部位设置多种的体征数据传感器，通过体征数据传感器连续采集指定时间区间内乳房部位的多种体征数据；
对所采集的体征数据分别进行初步分析并对每种体征数据的每个体征数值赋予相应的权重值；
针对每种体征数据分别执行以下分析步骤：
S1-1.依据正常情况下人体乳腺所对应体征建立标准节律模型，将赋予权重值的体征数据与该标准节律模型所对应的体征变化方程进行拟合运算，获得评估对象乳腺的特定体征变化方程和/或特定体征变化曲线；
S1-2.依据异常条件下的乳腺体征建立异常节律模型，寻找评估对象乳腺的特定体征变化方程或特定体征变化曲线中是否存在异常节律模型中的指定体征的异常确认特征和/或是否存在异常节律模型中的指定体征的异常排除特征。
优选地，寻找评估对象乳腺的多个特定体征变化方程或特定体征变化曲线中是否存在相应异常节律模型中的指定体征的异常确认特征的具体步骤为：
选取一个体征，并在其所对应的特定体征变化方程或特定体征变化曲线中寻找与该体征相对应异常节律模型中的指定体征的异常确认特征；
如确认找到此体征对应的异常确认特征，则在该异常确认特征相对应的指定时间窗口内寻找另一体征所对应的特定体征变化方程或特定体征变化曲线中是否存在与该另一体征相对应异常节律模型中的指定体征的异常确认特征；重复直至完成寻找所有体征的异常确认特征。
优选地，所述评估对象乳腺的特征体征变化方程与标准节律模型进行拟合运算后获得方程的对应参数值；依据异常条件下的乳腺体征变化建立评估模型，将上述参数值输入至评估模型进行运算并输出结果。
优选地，体征测量装置分别对称放置于人体两侧乳房表面；每个体征测量装置至少设有两个同类体征传感器，每个体征传感器都分别独立采集其体征数据，并建立与其对应的特定体征变化方程和/或特定体征变化曲线。
优选地，将同一体征传感器在不同节律周期中的同一指定时间区间所收集的特征数据获得的特定体征变化方程或特定体征变化曲线进行比较，将上述对比值作为输入值输入至评估模型进行运算。
优选地，将同一侧乳房不同体征传感器所对应的特定体征变化方程或特定体征变化曲线进行比较，获得某时刻相应的对比值或某指定时间区间的积分对比值，将上述对比值作为输入值输入至评估模型进行运算。
优选地，将两侧乳房位置对应的体征传感器所对应的特定体征变化方程或特定体征变化曲线进行比较，获得某时刻相应的对比值或某指定时间区间的积分对比值，将上述对比值作为输入值输入至评估模型进行运算。
优选地，在特定体征变化方程或特征体征变化曲线与异常节律模型比对前，对特定体征变化方程或特征体征变化曲线进行智能有序性多模态动态映射运算，使特定体征变化方程或特征体征变化曲线映射到异常节律模型中。
优选地，在多个特定体征变化方程或特征体征变化曲线比对前，对特定体征变化方程或特征体征变化曲线进行智能有序性多模态动态映射运算，使多个特定体征变化方程或特征体征变化曲线相互映射。
优选地，权重值的赋值方法为：将当前采集的体征数据与历史采集的体征数据进行比较，计算当前体征数据相对于历史体征数据的变化趋势及变化速率；对于变化趋势增大的体征数据赋予较高的权重值，对变化趋势减少的体征数据赋予较低的权重值；对变化速率较小的体征数据赋予较高的权重值，对变化速率较大的体征数据赋予较低的权重值。
优选地，通过在特定体征变化方程或特定体征变化曲线中寻找异常节律模型中的指定体征的异常确认特征和/异常排除特征的方法如下：
采集正常情况下乳腺的相关体征数据以及异常条件下的乳腺体征数据分别建立标准节律模型及异常节律模型的数据库；将评估对象所采集的每一个体征的有序性数据建立相应的数据矩阵，并以参数化的方法建立长度不同的数据窗口；将数据窗口与相对应标准节律模型及异常节律模型的数据库进行比对并计算相似值，找出异常确认特征和/或异常排除特征；
将不同体征的分析结果进行有序性多体征整合分析，分辨出相关体征属于因变量或自变量，对因变量计算出相关体征之间的时间有序性间距。
优选地，所述智能有序性多模态动态映射运算为将评估对象采集的各个体征数据与标准节律模型以及异常节律模型中对应的异常确认特征或异常排除特征进行映射运算，使体征数据的变化速率及幅度与对应的异常确认特征或异常排除特征相适应。
优选地，所述智能有序性多模态动态映射运算为将待比较的特定体征变化方程和/或特定体征变化曲线进行映射运算，使特定体征变化方程和/或特定体征变化曲线的变化速率及幅度相适应。
本发明的有益效果：设置在乳房的传感器收集多种的体征数据，包括但不限定于乳房的温度、湿度等。由于乳腺癌细胞会引起相应体征的变化，这些细微变化很容易被各种人体活动或不同的状态所产生的噪声所掩盖，而且以往只会使用单一的体征数据进行分析，如温度，很难进行精确分析。本发明通过收集多种体征数据，相应的建立多种分析的模态，对不同的模态进行分析能够更加准确及有效。
 图1
为特定体征变化曲线中分析找出重复异常特征的展示图；
图2
为智能有序性多模态动态映射运算方法数据块挖掘的展示图；
图3
为智能有序性多模态动态映射运算方法转化矩阵 B 及路径规划的展示图；
图4
为多模态的分析方法的展示图；
图5
为不同乳房同类体征数据的对比的展示图。
实施例
本发明中所提及的有序性，是指对人体，尤其是乳腺的各种体征数据随着时间会呈现一定的周期性，而且部分体征出现的次序也会有先后的关系，这种周期性的体征数据包括正常状态下的人体体征，也包括异常状态下的人体体征。通过有序性的分析，可以寻找出是否存在异常状态下的体征。
多模态是指通过不同的传感器对不同人体体征数据进行收集，并相应的建立多种分析的模态。由于每种模态都有其独特的有序性，而且当出现异常状态下，不同的模态都会产生对应的异常特征，而且部分模态出现的次序也会有先后的关系，通过联合对多个模态的异常特征分析，能够有效的提高异常判断的准确性。
对于多模态的分析方法如下：
首先通过不同的传感器对不同人体体征数据进行收集，并相应的建立多种分析的模态，也就是对不同体征数据分别建立特定体征变化方程或体征变化曲线，
然后选定一个体征，通过在其所对应的特定体征变化方程或特定体征变化曲线中寻找与该体征相对应异常节律模型中的指定体征的异常确认特征。如确认找到此体征对应的异常确认特征，就意味着存在异常的可能。但是只凭一种模态进行判断，准确率相对较低。
接着，需要对另一个体征进行分析判断，在上一体征的异常确认特征相对应的指定时间窗口内寻找该另一体征的特定体征变化方程或特定体征变化曲线中是否存在与该另一体征相对应异常节律模型中的指定体征的异常确认特征。如不能找到该另一体征的异常确认特征，也就意味着异常的概率降低。相反，如在另一体征也找到对应的异常确认特征，异常的概率提高。如此类推，直至完成寻找所有体征的异常确认特征。也就是说在不同模态中，即不同的体征数据总寻找到的对应异常确认特征越多，则评估乳腺存在异常的概率越高；反则，概率越低。通过多模态的分析，可以大大提高判断的准确率。
针对乳腺而言，目前的多模态可以包括以下人体体征：温度和湿度，而且上述模态还应该包括不同位置的同类传感器在同一周期或不同周期内所分别收集的体征数据，同一位置不同种类传感器在同一周期或不同周期内所分别收集的体征数据等。
所述对称性，一般指人体在正常状态下，即非异常状态下，人体的体征数据是有序的，也就是在对应的周期内，这些特征数据是对称的。对于乳腺而言，在同一周期的一对乳腺收集的体征数据也具有对称性。此外， 同一时间段， 例如凌晨01:00至04:00期间， 但不同的时间区间， 例如三个月至一年的同一侧乳房收集的体征数据也具有对称性。但是在异常状态下，如一侧乳腺出现异常，那么以上的对称的体征数据就会不对称，产生对称性破缺。包括在不同节律周期内的乳腺都会产生，通过分析对称性破缺，也能够判断是否出现异常的依据。
为能够方便及实时对乳腺部位的体征数据进行连续采集，可将相应的体征传感器设置在胸围内，以便于使用者能够时刻佩戴并实时采集数据。采集的数据可通过无线连接等方式发送至服务器等进行处理分析。考虑到佩戴舒适等方面的因数，传感器不可能时刻都能够紧贴乳房，如使用者的运动或者姿势的变换都可能导致传感器脱离乳房表面，从而导致所采集的数据存在误差甚至错误。为此，对所采集的体征数据进行初步分析并对每个体征数值赋予相应的权重值。
考虑到正常人体的体征数据在一个节律周期内变化的幅度有限，而且某一较短的时间区间内变化速率相对较小。由此可以通过上述规律对所采集的体征数据的权重值赋值，通过权重值的高低判断或决定该体征数据对后续数据处理的影响程度，一般来说权重值高的为相对可靠或精确的数据，相反权重值低的为存在误差或错误的数据。权重值的赋值规则如下：将当前采集的体征数据与之前采集的体征数据进行比较，计算当前体征数据相对于之前体征数据的变化趋势及变化速率；对于变化趋势增大的体征数据赋予较高的权重值，相反对变化趋势减少的体征数据赋予较低的权重值；对变化速率较小的体征数据赋予较高的权重值，相反对变化速率较大的体征数据赋予较低的权重值。
由于人体的各种体征在相应的节律周期内是具有有序性的，即可以通过采集正常情况下人体的某个节律周期内的体征，并建立该体征的标准节律模型，然后将实时采样并赋予权重值的待分析的体征数据与该标准节律模型进行拟合，
Figure pctxmlib-appb-M000001
， 当中
Figure pctxmlib-appb-M000002
为拟合函数， 可以但不限于使用傅立叶变换、高斯混合模型、小波变换、余弦拟合、正弦拟合等，
Figure pctxmlib-appb-M000003
为时间数列，
Figure pctxmlib-appb-M000004
为体征数据数列，
Figure pctxmlib-appb-M000005
为拟合相关参数，
Figure pctxmlib-appb-M000006
为误差值。通过节律模型拟合，从而获得评估对象乳腺的特定体征变化方程和/或特定体征变化曲线。该特定体征变化方程或特定特征变化曲线用于后续分析，寻找上述方程或曲线中是否存在异常的特征。
同样，对于产生异常的人体组织的体征也会在相应的节律周期内表现出有序性，为此也可以针对异常条件下的乳腺体征建立异常节律模型。而且通常在异常的节律模型中会存在某些异常的特征，同样也会存在用于排除异常的特征。将特定体征变化方程或特定特征变化曲线与异常节律模型进行比较并寻找特定体征变化方程或特定特征变化曲线中是否存在异常确认特征。同样通过在特定体征变化方程或特定特征变化曲线中寻找异常排除特征。
鉴于，通过连续采样获得的各种体征数据的十分庞大，因此在上述分析判断过程中需要进行大量的数据运算。为提高运算速度及运算精度，发明人还通过计算机设计了评估模型，该评估模型采用人工神经网络进行运算。将评估对象乳腺的特征体征变化方程与标准节律模型进行拟合运算后获得方程的对应参数值、异常确认特征、异常排除特征等作为评估模型的输入，经评估模型运算输出运算结果。
体征测量装置分别对称放置于人体两侧乳房表面；每个体征测量装置至少设有两个同类体征传感器，每个体征传感器都分别独立采集其体征数据，并建立与其对应的特定体征变化方程和/或特定体征变化曲线。如前所述的对称性破缺说明中包括了多种情况，如同一乳房不同位置所获得的同类体征数据的对比，以及不同乳房同类体征数据的对比（如图5所示），以及同一体征传感器在不同节律周期中同一指定时间区间内的特征数据的对比，以判断是否存在对称性破缺。对比方式为
Figure pctxmlib-appb-M000007
，当中 C 为对比值，
Figure pctxmlib-appb-M000008
 为第
Figure pctxmlib-appb-M000009
 个因素的权重，
Figure pctxmlib-appb-M000010
，s 为对比计算指定开始时间，v 为对比计算指定结束时间，
Figure pctxmlib-appb-M000011
为对比函数，λ 为体征数据数列， 可以为两侧乳房或同一侧乳房指定时间的数据。由此，需要将同一侧乳房不同体征传感器所对应的特定体征变化方程或特定体征变化曲线进行比较，获得某时刻相应的对比值或某指定时间区间的积分对比值，将上述对比值作为输入值输入至评估模型进行运算。同样，也需要将两侧乳房位置对应的体征传感器所对应的特定体征变化方程或特定体征变化曲线进行比较，获得某时刻相应的对比值或某指定时间区间的积分对比值，将上述对比值作为输入值输入至评估模型进行运算。
另外，还应该针对同一乳房在不同节律周期内的同类体征数据是否存在对称性破缺，即将同一体征传感器在不同节律周期中的同一指定时间区间所收集的特征数据获得的特定体征变化方程或特定体征变化曲线进行比较，将上述对比值作为输入值输入至评估模型进行运算。
为能够快速准确的寻找到特定体征变化方程或特定体征变化曲线是否存在异常确认特征、异常排除特征，本发明还提供了相应的对比及寻找方法，可以但不限于用以下的方式实施：
S2-1.将赋予权重值的体征数据按照时序T建立体征数据矩阵，其中
Figure pctxmlib-appb-M000012
 ；tn为体征数据的采样时刻，n为矩阵的数据个数，正整数；
S2-2.从体征数据矩阵的第i个位置开始依顺序取出m个体征数据构成长度为m的数据窗口，
Figure pctxmlib-appb-M000013
，其中i，m为正整数，且1 ≤ i ≤ n-m+1，m ≤ n；
S2-3.将数据窗口的数据与体征数据矩阵中第j个位置开始对应m长度的体征数据
Figure pctxmlib-appb-M000014
，计算相似值，且将数据窗口从第j个位置向后移动并计算，获得距离相似度矩阵
Figure pctxmlib-appb-M000015
，其中di ， j为Ti，m与Tj，m，m的相似度计算值，数值越小，越为相似，其中j为正整数，1 ≤ j ≤ n-m+1；
S2-4.数据窗口从体征数据矩阵首个体征数据开始移动至第n-m+1个，取出每个距离矩阵的最小值，并获得最小距离矩阵
Figure pctxmlib-appb-M000016
，其中
Figure pctxmlib-appb-M000017
为上述距离相似度矩阵，min(
Figure pctxmlib-appb-M000018
)为
Figure pctxmlib-appb-M000019
中的最小值。通过最小距离矩阵得出体征数据矩阵中的重复特征；通过最小距离矩阵得出体征数据矩阵中的重复特征；
S2-5.将以上重复特征与异常确认特征和/或异常排除特征进行比较和/或确认。
S2-6.将不同体征的分析结果进行有序性多体征整合分析，分辨出相关体征属于因变量或自变量，对因变量计算出相关体征之间的时间有序性间距θ，即上述提及的指定时间窗口。
下面以具体的实施例对上述对比及寻找方法作出详细的解析：
以下为上述系统性独立因素智能模态数据分析的实施例，如S2-1及S2-2的描述，下表以n=12，m=4，i=2为例，数据窗口为T2，4的情况；
Figure pctxmlib-appb-I000001
如S2-3所描述，j=1位置且m=4长度的体征数据T1，4与数据窗口T2，4比对计算得出距离相似度矩阵
Figure pctxmlib-appb-M000020
，如下表；
Figure pctxmlib-appb-I000002
如S2-4所描述，最小距离矩阵为P=[0.5， 0.5，……， 0.4]，如下表:
Figure pctxmlib-appb-I000003
其中，图1为根据以上所描述的方法，在数据内找出重复特征的具体实践情况。
如S2-5所述将以上重复特征与异常确认特征和/或异常排除特征进行比较和/或确认。
其后如S2-6所述，基于不同系统性独立因素智能模态数据分析的结果，进行有序性多因素智能整合分析，分办出因素属于因变量或自变量，对于因变量计算出相关因素之间的时间有序性间距θ。
考虑到，不同侧乳房在的同类体征数据的变化速率可能会存在不同，不同使用者的同类体征数据变化速率肯定会存在不同，因此就会导致将标准节律模型、异常节律模型、某个使用者的特定体征变化方程或特定特征变化曲线进行相互比对时存在问题。因此在比对前，需要对特定体征变化方程或特定特征变化曲线进行智能有序性多模态动态映射运算，该算法目的在于对所收集的大量人体体征数据的大数据中进行模态近似比对，当中涉及标准节律模型和异常节律模型与待进行模态对比发掘的数据，此算法以智能有效的方式从待进行模态挖掘数据中判断并找是否含有异常节律模型的数据组并作出筛选挖掘，该算法容许映射比对过程中存在速度及波幅上的不同但模态相同，如 ，以算法方式智能地将近似模态以动态映射方式从数据上将相关数据块挖掘，相关数据块挖掘的方法可以用以下方法计算获取，但不限于以下方法，下面以具体的实施例对上述对智能有序性多模态动态映射运算方法作出详细的解析，可以但不限于用以下的方式实施：
S3-1.多模态所涉及的体征数据数进行权重值赋值运算后，建立特定体征变化方程和/或特定体征变化曲线，将待进行智能有序性多模态动态映射运算的方程/模型/矩阵分别转化为矩阵，构成被映射矩阵
Figure pctxmlib-appb-M000021
，其中被映射矩阵长度为q；相关体征所对应其中一个模态f的映射矩阵为
Figure pctxmlib-appb-M000022
，其中映射矩阵的长度为p；其中p，q为正整数；
S3-2.构建p×q的矩阵B，其中矩阵B中的矩阵元素
Figure pctxmlib-appb-M000023
为被映射矩阵的中的元素
Figure pctxmlib-appb-M000024
与映射矩阵的元素
Figure pctxmlib-appb-M000025
之间进行对比计算获得的对比值；其中k，l为正整数
S3-3.按照以下规则挖掘上述矩阵B中的最小距离路径矩阵
Figure pctxmlib-appb-M000026
，其中最小距离路径矩阵G中的第u个元素为矩阵B中的第x， y个元素，即
Figure pctxmlib-appb-M000027
 并且 max(p，q)≤v<p+q-1：其中u，v，x，y为正整数，max(p，q)为p，q中的最大值
a．挖掘路径的起点为
Figure pctxmlib-appb-M000028
，挖掘路径的终点为
Figure pctxmlib-appb-M000029
；
b.如目前路径点为
Figure pctxmlib-appb-M000030
，则下一路径点为
Figure pctxmlib-appb-M000031
，且相邻的两点必需满足以下条件 0≤α-α'≤1 及 0≤β-β'≤1；其中，α，β，α'，β'为正整数
c.路径的规划计算除满足b所述的要求外，也要满足选择最高近似值/最低距离值的条件。其中图2阐述了路径规划的机制。
d.将最小距离路径中的所对应的矩阵元素形成被映射矩阵及映射矩阵之间的映射关系表，重复进行步骤b及c，累积的距离成本计算方程为
Figure pctxmlib-appb-M000032
，依此计算获得该路径与模态f的对比近似度，如 。
如将上述映射运算应用于多模态时，还应执行以下步骤：
e.模态
Figure pctxmlib-appb-M000033
 在特定体征变化方程和/或特定体征变化曲线对应的矩阵W中找到最小距离路径的开始时间为
Figure pctxmlib-appb-M000034
，基于有序性及多模态考虑，另一模态
Figure pctxmlib-appb-M000035
 模态的因素于以下时间
Figure pctxmlib-appb-M000036
 时段发生，且为有效模态，该另一模态
Figure pctxmlib-appb-M000037
 的对比方法可依据a，b，c，d步骤，如图4，其中模态，即体征数量
Figure pctxmlib-appb-M000038
，
Figure pctxmlib-appb-M000039
，
Figure pctxmlib-appb-M000040
。
f.重复步骤a，b，c，d，e，直至所有因素模态计算完毕，相关计算值会依次填入并成为特征数矩阵。

Claims (13)

  1. 基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：包括在评估对象的乳房部位设置多种的体征数据传感器，通过体征数据传感器连续采集指定时间区间内乳房部位的多种体征数据；
    对所采集的体征数据分别进行初步分析并对每种体征数据的每个体征数值赋予相应的权重值；
    针对每种体征数据分别执行以下分析步骤：
    S1-1.依据正常情况下人体乳腺所对应体征建立标准节律模型，将赋予权重值的体征数据与该标准节律模型所对应的体征变化方程进行拟合运算，获得评估对象乳腺的特定体征变化方程和/或特定体征变化曲线；
    S1-2.依据异常条件下的乳腺体征建立异常节律模型，寻找评估对象乳腺的特定体征变化方程或特定体征变化曲线中是否存在异常节律模型中的指定体征的异常确认特征和/或是否存在异常节律模型中的指定体征的异常排除特征。
  2. 根据权利要求1所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：
    寻找评估对象乳腺的多个特定体征变化方程或特定体征变化曲线中是否存在相应异常节律模型中的指定体征的异常确认特征的具体步骤为：
    选取一个体征，并在其所对应的特定体征变化方程或特定体征变化曲线中寻找与该体征相对应异常节律模型中的指定体征的异常确认特征；
    如确认找到此体征对应的异常确认特征，则在该异常确认特征相对应的指定时间窗口内寻找另一体征所对应的特定体征变化方程或特定体征变化曲线中是否存在与该另一体征相对应异常节律模型中的指定体征的异常确认特征；重复直至完成寻找所有体征的异常确认特征。
  3. 根据权利要求1所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：所述评估对象乳腺的特征体征变化方程与标准节律模型进行拟合运算后获得方程的对应参数值；依据异常条件下的乳腺体征变化建立评估模型，将上述参数值输入至评估模型进行运算并输出结果。
  4. 根据权利要求1所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：体征测量装置分别对称放置于人体两侧乳房表面；每个体征测量装置至少设有两个同类体征传感器，每个体征传感器都分别独立采集其体征数据，并建立与其对应的特定体征变化方程和/或特定体征变化曲线。
  5. 根据权利要求4所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：将同一体征传感器在不同节律周期中的同一指定时间区间所收集的特征数据获得的特定体征变化方程或特定体征变化曲线进行比较，将上述对比值作为输入值输入至评估模型进行运算。
  6. 根据权利要求4所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：将同一侧乳房不同体征传感器所对应的特定体征变化方程或特定体征变化曲线进行比较，获得某时刻相应的对比值或某指定时间区间的积分对比值，将上述对比值作为输入值输入至评估模型进行运算。
  7. 根据权利要求4所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：将两侧乳房位置对应的体征传感器所对应的特定体征变化方程或特定体征变化曲线进行比较，获得某时刻相应的对比值或某指定时间区间的积分对比值，将上述对比值作为输入值输入至评估模型进行运算。
  8. 根据权利要求1所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：在特定体征变化方程或特征体征变化曲线与异常节律模型比对前，对特定体征变化方程或特征体征变化曲线进行智能有序性多模态动态映射运算，使特定体征变化方程或特征体征变化曲线映射到异常节律模型中。
  9. 根据权利要求5或6或7所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：在多个特定体征变化方程或特征体征变化曲线比对前，对特定体征变化方程或特征体征变化曲线进行智能有序性多模态动态映射运算，使多个特定体征变化方程或特征体征变化曲线相互映射。
  10. 根据权利要求1所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：权重值的赋值方法为：将当前采集的体征数据与历史采集的体征数据进行比较，计算当前体征数据相对于历史体征数据的变化趋势及变化速率；对于变化趋势增大的体征数据赋予较高的权重值，对变化趋势减少的体征数据赋予较低的权重值；对变化速率较小的体征数据赋予较高的权重值，对变化速率较大的体征数据赋予较低的权重值。
  11. 根据权利要求1或2所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：通过在特定体征变化方程或特定体征变化曲线中寻找异常节律模型中的指定体征的异常确认特征和/异常排除特征的方法如下：
    采集正常情况下乳腺的相关体征数据以及异常条件下的乳腺体征数据分别建立标准节律模型及异常节律模型的数据库；将评估对象所采集的每一个体征的有序性数据建立相应的数据矩阵，并以参数化的方法建立长度不同的数据窗口；将数据窗口与相对应标准节律模型及异常节律模型的数据库进行比对并计算相似值，找出异常确认特征和/或异常排除特征；
    将不同体征的分析结果进行有序性多体征整合分析，分辨出相关体征属于因变量或自变量，对因变量计算出相关体征之间的时间有序性间距。
  12. 根据权利要求8所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：所述智能有序性多模态动态映射运算为将评估对象采集的各个体征数据与标准节律模型以及异常节律模型中对应的异常确认特征或异常排除特征进行映射运算，使体征数据的变化速率及幅度与对应的异常确认特征或异常排除特征相适应。
  13. 根据权利要求9所述的基于有序性、多模态及对称性破缺对乳房体征分析的方法，其特征在于：所述智能有序性多模态动态映射运算为将待比较的特定体征变化方程和/或特定体征变化曲线进行映射运算，使特定体征变化方程和/或特定体征变化曲线的变化速率及幅度相适应。
PCT/IB2022/062441 2021-12-20 2022-12-19 基于有序性、多模态及对称性破缺对乳房体征分析的方法 WO2023119103A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111562362.6 2021-12-20
CN202111562362.6A CN114468996B (zh) 2021-12-20 2021-12-20 基于有序性、多模态及对称性破缺对乳房体征分析的方法

Publications (1)

Publication Number Publication Date
WO2023119103A1 true WO2023119103A1 (zh) 2023-06-29

Family

ID=81494448

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/IB2022/062441 WO2023119103A1 (zh) 2021-12-20 2022-12-19 基于有序性、多模态及对称性破缺对乳房体征分析的方法

Country Status (2)

Country Link
CN (1) CN114468996B (zh)
WO (1) WO2023119103A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117857261A (zh) * 2024-01-18 2024-04-09 安徽即刻智能科技有限公司 一种基于边缘计算的工业智能网关

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114468996B (zh) * 2021-12-20 2024-01-23 香港生物节律研究院有限公司 基于有序性、多模态及对称性破缺对乳房体征分析的方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100312136A1 (en) * 2007-06-22 2010-12-09 Advanced Functional Systems, Llc Systems and methods for thermographic-guided medical treatment
CN108720814A (zh) * 2018-08-13 2018-11-02 脱浩东 一种乳腺健康远程监测系统及方法
CN108777161A (zh) * 2018-06-19 2018-11-09 云士康（深圳）智能科技有限公司 利用瞬时分布式图像监测乳房数据信息的方法
CN109008986A (zh) * 2018-08-21 2018-12-18 深圳市浓华生物电子科技有限公司 一种胸部检测方法、电子设备和健康监测仪
CN110432866A (zh) * 2019-08-01 2019-11-12 福州智睿伊科技有限公司 一种基于温度监测乳腺健康的系统和方法
CN114468996A (zh) * 2021-12-20 2022-05-13 香港生物节律研究院有限公司 基于有序性、多模态及对称性破缺对乳房体征分析的方法

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6179786B1 (en) * 1998-10-02 2001-01-30 Profemme Ltd. System for thermometry-based breast cancer risk-assessment
JP3781686B2 (ja) * 2001-02-23 2006-05-31 松下電器産業株式会社 遺伝子診断装置及び遺伝子診断方法
US20130123885A1 (en) * 2010-04-16 2013-05-16 New York University Apparatus, systems, computer-accessible medium and methods for facilitating radio frequency hyperthermia and thermal contrast in a magnetic resonance imaging system
CN105686810A (zh) * 2015-07-31 2016-06-22 上海温尔信息科技有限公司 一种温度测量方法及装置
CN105740614A (zh) * 2016-01-28 2016-07-06 上海温尔信息科技有限公司 基于肿瘤细胞异常的温湿度场模型的比对评估方法及装置
CN105760701A (zh) * 2016-03-22 2016-07-13 上海温尔信息科技有限公司 一种肿瘤细胞温度异常代谢热的捕获方法及装置
US20190290903A1 (en) * 2018-03-23 2019-09-26 The Board Of Regents Of The University Of Oklahoma System and Method of Electric-Induced Acoustic Tomography for Electrotherapy Monitoring
CN108720812A (zh) * 2018-08-13 2018-11-02 脱浩东 一种利用乳腺外部温度数据的乳腺健康评估装置

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100312136A1 (en) * 2007-06-22 2010-12-09 Advanced Functional Systems, Llc Systems and methods for thermographic-guided medical treatment
CN108777161A (zh) * 2018-06-19 2018-11-09 云士康（深圳）智能科技有限公司 利用瞬时分布式图像监测乳房数据信息的方法
CN108720814A (zh) * 2018-08-13 2018-11-02 脱浩东 一种乳腺健康远程监测系统及方法
CN109008986A (zh) * 2018-08-21 2018-12-18 深圳市浓华生物电子科技有限公司 一种胸部检测方法、电子设备和健康监测仪
CN110432866A (zh) * 2019-08-01 2019-11-12 福州智睿伊科技有限公司 一种基于温度监测乳腺健康的系统和方法
CN114468996A (zh) * 2021-12-20 2022-05-13 香港生物节律研究院有限公司 基于有序性、多模态及对称性破缺对乳房体征分析的方法

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117857261A (zh) * 2024-01-18 2024-04-09 安徽即刻智能科技有限公司 一种基于边缘计算的工业智能网关

Also Published As

Publication number Publication date
CN114468996B (zh) 2024-01-23
CN114468996A (zh) 2022-05-13

Similar Documents

Publication Publication Date Title
US7734077B2 (en) Method of assessing localized shape and temperature of the human body
US10650512B2 (en) Systems and methods for topographical characterization of medical image data
Etehad Tavakol et al. Analysis of breast thermography using fractal dimension to establish possible difference between malignant and benign patterns
JP2018081714A (ja) 自動特徴解析、比較、及び異常検出
US20120330447A1 (en) Surface data acquisition, storage, and assessment system
CN106600585A (zh) 基于灰度共生矩阵的皮肤状态定量评价方法
US20080154154A1 (en) Internet-based system and a method for automated analysis of tactile imaging data and detection of lesions
CN108151670B (zh) 一种后背三维对称度评价装置及方法
JP2022507955A (ja) 画像を使用して乳癌リスクを評価するためのシステムと方法
WO2022011616A1 (zh) 一种通过影像组学特征判断癌症治疗反应的方法及系统
WO2023119103A1 (zh) 基于有序性、多模态及对称性破缺对乳房体征分析的方法
CN105678734B (zh) 一种图像匹配系统的异源测试图像标定方法
de Sá Gomes et al. Scan time, reliability and accuracy of craniofacial measurements using a 3D light scanner
Kim et al. Detection and severity assessment of peripheral occlusive artery disease via deep learning analysis of arterial pulse waveforms: Proof-of-concept and potential challenges
CN105686810A (zh) 一种温度测量方法及装置
CN111588353A (zh) 一种体温测量方法
KR101886000B1 (ko) 뇌 자기공명 영상에 기반한 뇌 구조 비정상도 지수 및 이를 이용한 퇴행성 뇌질환 진단 방법
Li et al. A unified approach to nonparametric comparison of receiver operating characteristic curves for longitudinal and clustered data
Dubovskaya et al. Neural Network Modeling in the Construction of Multiple Nonlinear Regression of RTM-Diagnostic Data
Štrobl et al. Methods for automatic estimation of the number of clusters for K-means algorithm used on eeg signal: feasibility study
Inoue et al. Modeling disease progression with longitudinal markers
Gutenko et al. AnaFe: Visual Analytics of Image-derived Temporal Features—Focusing on the Spleen
CN109192301A (zh) 通过互联网智能医疗健康设备进行患者诊断工作方法
She et al. Lesion classification using 3D skin surface tilt orientation
Connelly Bland-Altman plots.

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22910321

Country of ref document: EP

Kind code of ref document: A1