Photoplethysmographic Dynamic Errors of Fractional-Order Chaos Synchronization and Color Relational Analysis in Computer-assisted Diagnosis of Peripheral Arterial Disease

  • 李 健明

Student thesis: Doctoral Thesis

Abstract

Lowe-limb peripheral arterial disease (PAD) is a prevalent and insidious disease caused by atherosclerosis potentially leading to total occlusion of artery and gangrene of leg The most common risk factors for this limb-threatening disease are type 2 diabetes mellitus ageing and cigarette smoking but the pathogenesis of atherosclerosis may include hypertension dyslipidemia and homocysteinemia Underdiagnosis and under-treatment of PAD are partly because of its stealthy progress and lack of screening device for early diagnosis and active monitoring Typical clinical manifestations of this vascular disorder are intermittent claudication and leg pain with unknown degree of arterial obstruction In tertiary hospitals PAD can be accurately diagnosed with invasive angiography and ankle-brachial pressure index (ABI) currently deemed as a standard for evaluating severity The severity is generally divided into normal (ABI > 0 9) low-grade (ABI = 0 9-0 5) and high-grade (ABI > 0 9) based on the ABI Unfortunately it is not sensitive enough to be used for prognostic evaluation of patients with hardened arteries commonly observed in diabetic and uremic Photoplethysmography (PPG) is an optic technique based on the association of blood volume change with light-tissue interaction In current clinical setting measurement of peripheral oxygenation and cardiac rhythm highly rely on this noninvasive technique Previous studies have reported that the differences of pulse transit time (PTT) of photoplethysmograms between the right and left feet are significantly greater in PAD patients than those in healthy persons However a PPG-based assistant device for the diagnosis of PAD is still lacking The objective of this thesis is to develop a computer-assisted tool as an assistant tool for PAD diagnosis The prototype was designed by combining dynamic error system of fractional-order chaos synchronization and pattern recognition techniques to test its potential in clinical applications In the first experiment photoplethysmographic signals were captured from 25 recruited subjects including 11 healthy adults and 14 diabetic patients with three degrees of disease severity Taking R wave of electrocardiogram as the reference parameters △PTTp △PTTf and △RT are determined from big toes of two feet The results show that the dynamic errors of the fractional-order chaos synchronization and color relational analysis are able to effectively classify 3 different degrees of PAD severity In the second experiment 15 healthy persons and 17 diabetic patients were recruited for PPG measurement The classification of PAD severity is also effectively conducted using support vector machine to design the classifier with wolf pack search algorithm adopted for obtaining optimal model parameters Compared to the artificial neural network and other machine learning techniques the proposed method is demonstrated to be efficient and effective to estimate the PAD severity Photoplethysmography combined with proper classification techniques consisting of color relational analysis and support vector machine is potential to design a portable computer-assisted tool for real-time PAD diagnosis Its advantages include only small sample data are required for model training; patterns are expandable; signal capture and model training are efficient; parameter assignment is not necessary; and trapped in local minima can be avoided Furthermore electrocardiogram may be neglected for a PPG to be analyzed with dynamic error system of fractional-order chaos synchronization In conclusion PPG analyzed with color relation and modeled with support vector machine is potential in designing an assistant tool to screen and monitor the PAD in communities and hospitals
Date of Award2014 Jan 27
Original languageEnglish
SupervisorTainsong Chen (Supervisor)

Cite this

Photoplethysmographic Dynamic Errors of Fractional-Order Chaos Synchronization and Color Relational Analysis in Computer-assisted Diagnosis of Peripheral Arterial Disease
健明, 李. (Author). 2014 Jan 27

Student thesis: Doctoral Thesis