Automated myocardial infarction identification based on interbeat variability analysis of the photoplethysmographic data. (March 2020)

• Record Type: Journal Article
• Authors: Chakraborty, Abhishek; Sadhukhan, Deboleena; Pal, Saurabh; Mitra, Madhuchhanda
• Is Part Of: Biomedical signal processing and control. Volume 57(2020)
• Publication Date: 2020-03
• Subjects: Automated diagnosis -- Myocardial infarction (MI) -- Photoplethysmography (PPG) -- PPG derivatives -- Feature extraction -- Variability analysis -- Expert health systems; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2019.101747
• Languages: English
• ISSNs: 1746-8094

Abstract

Abstract: Background and objective: Myocardial infarction (MI) remains a major cause of mortality around the world for decades. Timely detection followed by instant medical intervention is strongly recommended to minimize MI related death threats. Generally, electrocardiogram (ECG) based automated techniques are preferred to ensure early diagnosis of MI. Recently, Photoplethysmogram (PPG) signal is evolving as a promising diagnostic tool for several cardiac monitoring applications because of its low cost, reliable and easy acquisition technology. However, the use of PPG for MI detection has not been much explored till date. Hence, in the present work, a method for MI detection is proposed based on the use of the PPG signal only. Method: The pathophysiological alteration due to MI induces a beat-to-beat variation in the PPG beat morphology. Different time-plane parameters from the PPG signal and its derivatives are extracted to represent the individual beat morphologies and their variations are then analyzed as features over the cardiac cycles. By using different feature optimization techniques, finally five features are selected that presents discriminating variability to identify MI. Results: The proposed method is evaluated with PPG records collected from 52 hospitalized MI patients and 52 normal subjects. The optimized five feature set, representing the inter beat morphological variations exhibits significant performance along with SVM (linear) classification technique with an average sensitivity, positive predictivity and detection accuracy of 92.70 %, 98.10 %, and 95.40 %, respectively. Conclusions: Compared to other related works, use of PPG signal for MI detection is studied in this research for the first time. The promising result obtained establishes the utility of PPG signal for MI detection with a potential of implementation in the personal healthcare systems.