Automatic identification of asthma from ECG derived respiration using complete ensemble empirical mode decomposition with adaptive noise and principal component analysis. (August 2022)

• Record Type: Journal Article
• Authors: Sarkar, Surita; Bhattacherjee, Saptak; Bhattacharyya, Parthasarathi; Mitra, Madhuchhanda; Pal, Saurabh
• Is Part Of: Biomedical signal processing and control. Volume 77(2022)
• Publication Date: 2022-08
• Subjects: Asthma -- Complete ensemble empirical mode decomposition with adaptive noise -- ECG derived respiration -- Principal component analysis -- Supervised classifier; 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.2022.103716
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: Good temporal and spectral reconstruction of respiration is derived with CEEMDAN-PCA. Morphological variation of normal and asthmatic EDR leads to extraction of distinguishable statistical features. Proposed method enhances patient comfort and cost effectiveness by reducing sensor attachments with patient body. Abstract: In spite of the increasing prevalence of asthma worldwide, it often remains underdiagnosed and untreated, causing permanent damage to lungs. In this scenario, early diagnosis and continuous monitoring is extremely necessary to control this by maintaining proper pulmonary function. Despite their wide applications in diagnosing asthma, conventional methods often suffer from bulky setup, high cost, huge dependency on patient effort, and intrusion in natural breathing. This leads to a greater demand to develop non-invasive, low-cost and reliable detection method of asthma without causing any patient discomfort. Therefore, in this study an unobtrusive, cost-effective method for automatic detection of asthma using single-lead electrocardiogram (ECG) has been presented. The proposed method utilizes complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) technique with principal component analysis (PCA) for extracting ECG derived respiration (EDR). The derived respiration signal demonstrated morphological variations caused due to pathophysiological changes in lungs during asthma. Features extracted from EDR were fed to supervised classifiers for classifying asthma from normal subjects. Classification performance assessed on a total of 94 subjects (both normal and asthma subjects) demonstrated that a maximum of 98.94% accuracy with 97.87% sensitivity and 100% specificity could be achieved by k-nearest neighbor (kNN) classifier. The results suggest that the proposed method can be used as a viable and promising alternative for identification of asthma with good accuracies using single-lead ECG only.