Medical Signal Research And Application Based On Empirical Mode Decomposition

There are interactions and mutual adaptations between various physiological activities of the human body,which leads to nonlinear and non-stational characteristics of medical signals corresponding to physiological activities.Traditional signal analysis methods such as Fourier transform have not been able to effectively meet the needs of medical signal processing.With the rapid development of nonlinear-and-non-stational signal processing methods in recent years,empirical mode decomposition is one of the best tools that can accurately and effectively analyze that such signals.Empirical modal decomposition algorithm has been proven to have great potential and application prospects in the field of medical signal processing.However,the most of the current research are still limited to the use of empirical modal decomposition in analyzing and processing the medical signals,in which the potential factors of the signals would be easily neglected leading to the wrong results.In some cases,the analysis of the medical signals is based on the selecting and judgement of the components by human observation.Therefore,in order to further promote and deepen the application of the empirical mode decomposition algorithm in the field of medical signal processing,this paper carries out the following different research according to three different specific application scenarios:1.In order to distinguish the difference of heart rate variability in different physiological health conditions,this paper presents a novel method to analyzing heart rate variability time series based on complementary ensemble empirical mode decomposition algorithm and modified permutation entropy algorithm.This method has overcome the shortcomings of analyzing with single time-scale of heart rate variability in the previous permutation entropy method.Finally,the experimental results based on the MIT-BIH arrhythmia database show that the proposed method has statistically significant performance in distinguishing the patients with obvious abnormalities of heart rate.2.Photoplethysmography(PPG)is a non-invasive monitoring technique that has been shown to have lower implementation costs in monitoring cardiopulmonary activity than electrocardiographic techniques.Therefore,in the fifth chapter,we combine the great property of the complementary ensemble empirical mode decomposition algorithm,independent component analysis and non-negative matrix factorization to propose a novel method that can be used to separate the surrogate cardiorespiratory signals from PPG signals which are used to estimate heart rate and respiratory rate,respectively.The experimental results show that the proposed method has more stable and accurate estimation results than the previous method based on empirical mode decomposition.3.Diabetes has become the one of the majors threatens to human health.Therefore,in order to continuously monitor patients’ blood glucose value for effectively preventing and treating the diabetes,we propose a joint empirical decomposition and singular spectrum analysis based pre-processing method for wearable non-invasive blood glucose estimation.The non-invasive blood glucose estimation method can effectively suppress the influence of the motion artifact received by the wearable device during the signal acquisition,and guarantee the standard for the accuracy of the estimation result.