Statistical Feature-based Cerebral Infarction Detection from Photoplethysmogram Signals

• Published in: International Conference on Computing, Communication, and Networking Technologies (Online) pp. 1 - 5
• Main Authors: Chowdhury, Aditta, Das, Diba, Chowdhury, Monika, Hasan, Kamrul, Chowdhury, Mehdi Hasan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 24.06.2024
• Subjects: Accuracy; Cardiovascular disease (CVD); Cerebral infarction; Computed tomography; Feature extraction; Machine learning; Machine learning algorithms; Magnetic resonance imaging; Nearest neighbor methods; Optical imaging; Photoplethysmogram (PPG); Point of care; Real-time systems; Reliability
• ISSN: 2473-7674
• DOI: 10.1109/ICCCNT61001.2024.10726251

Cerebral infarction is one of the leading reasons of death in the world which is also known as ischemic stroke. An early treatment can reduce its effect to a great extent. However, the popular ways of detecting this disease are magnetic resonance imaging (MRI) and computed tomography (CT) scan which are costly and time-consuming processes. Photoplethysmogram (PPG) is an optical signal that detects the changes in blood volume. Cerebral infarction causes disruption in normal blood flow, hence there is a potential to use PPG signal to detect this condition. So far only a few studies have used PPG signal to detect Cerebral infarction from PPG signal. Also, the use of statistical features for detecting the disease is yet to be explored. In this study, an equal amount of PPG segments from 30 healthy subjects and 20 cerebral infarction patients are chosen. Different feature importance tests are done to find out the best features among 9 statistical features. After that, several machine learning algorithms are applied to determine the most effective model. Among the various models tested, KNN performed the best providing 98.33% accuracy. This lightweight approach, marked by its high accuracy, may pave the way for a reliable point-of-care solution offering a potential breakthrough in enhancing patient care and treatment strategies within the realm of strokes.