Novel and robust auxiliary indicators to ankle-brachial index using multi-site pulse arrival time and detrended fluctuation analysis for peripheral arterial disease assessment. (August 2022)

• Record Type: Journal Article
• Authors: Geng, Guoqiang; Zheng, Yali; Hong, Jingyuan; Liu, Qing; Poon, Carmen C.Y.
• Is Part Of: Biomedical signal processing and control. Volume 77(2022)
• Publication Date: 2022-08
• Subjects: Peripheral arterial disease -- Pulse arrival time -- Detrended fluctuation analysis -- Ankle-brachial index; 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.2022.103865
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: The novel indicator α1 ratio from multi-site PAT by DFA is robust to posture change. The sensitivity in detecting PAD was improved from 78.6% to 92.9% by combing ABI with α1 ratio. Multi-site PAT provides superior indicators for PAD detection than R-R interval. Abstract: The ankle-brachial index (ABI) test is a common tool used for screening peripheral arterial disease (PAD), but its accuracy has been doubted especially in patients with incompressible arteries attributed to calcification. Advanced wearable sensing technologies have potential to offer novel solutions for PAD assessment. This study investigated several novel indicators using the Detrended Fluctuation Analysis (DFA) on beat-to-beat RR interval (RRI) and pulse arrival time (PAT) from the four limbs of 7 young healthy and 7 elderly PAD subjects in supine and sitting postures. The auxiliary value of these indicators in addition to ABI for PAD assessment was also studied using a binary logistic regression model. The results showed that the short-scale DFA coefficients (α1 ) of RRI, α1 of PATs of fingers and α1 _ratio of PAT (the ratio of α1 of PAT between the ipsilateral finger and toe) in supine posture, were significantly different between the PAD and healthy groups (1.26 ± 0.24 vs 0.76 ± 0.21, 1.03 ± 0.30 vs 0.68 ± 0.19, and 0.84 ± 0.20 vs 1.42 ± 0.47, respectively, p < 0.05), and the α1 of RRI and α1 _ratio of PAT were not significantly influenced by postural change. Moreover, using ABI together with the supine α1 _ratio of PAT was more sensitive in detecting PAD (sensitivity = 92.9%) than ABI alone (78.6%) with the same specificity of 100%. The short-scale DFA coefficients of multi-site PAT in supine posture, which can be measured by wearable devices, is an auxiliary to ABI for detecting PAD.