|M.Sc Student||Gliner Vadim|
|Subject||Mobile Application Embedded with Novel Real Time R|
Peak Detector for Patients with Atrial
|Department||Department of Biomedical Engineering||Supervisor||Professor Yael Yaniv|
Cardiac diseases are one of the leading causes of morbidity and mortality in the western world, where atrial fibrillation (AF) is the most common sustained arrhythmia. Because AF can lead to stroke and ,very rarely, ventricular fibrillation, early detection of these episodes has an enormous clinical impact. Up to date there are no real-time devices that can precisely detect the R peaks in the ECG signal before, during and after AF events. The PhysioNet/CinC 2017 Challenge aimed to trigger a design of an algorithm that accurately classifies short single ECG lead record to 4 categories: normal rhythm, atrial fibrillation, noisy segment or other arrhythmias. The algorithm was optimized on randomly selected records out of the challenge learning set (8528 records after reassuring it includes 60.43% of normal records, 0.54% of noisy records, 9.04% of AF records and 30% of other rhythm disturbance) and tested on hidden test database.
Main objectives of this research is to design an algorithm that accurately detects the R peaks from ECG strips during AF event in the present of noise, movement and other associated arrhythmias, and to develop an artificial intelligence algorithm that accurately classifies short single ECG lead record to 4 categories: normal rhythm, atrial fibrillation, noisy segment or other arrhythmias.
One of the methods that was used in this research is a novel algorithm development which subtracts motion artifacts, electrical drift and breathing oscillations by low pass filtering. It also fixes the signal polarity, filters environmental noise, and deal with electrical spikes and premature beats by heuristic filter. The algorithm was tested on MITDB Physionet database. The R peak detector described above was used to accurately detect the R peaks. Based on the R peak annotation, the T, P, Q and S peaks were detected and ECG beat morphology was extracted. Machine learning techniques that include combination of 62 features were used for classification into 4 groups. Quadratic SVM classifier was used to classify the short ECG record to one of the four categories mentioned above. For records, which were classified as "normal" additional neural network classifier was applied.
The results of this research are that on average, our algorithm can precisely detect the R peaks with 0.26% of negative and positive false detection for a sensitivity of 99.69% and positive prediction of 99.74 % . The algorithm performs similarly on AF and non-AF patient data. Our arrhythmia classification algorithm reached results of total score (F1) of 0.8 (ranked 24 out a total of 90 open-source software entries), whereas normal rhythm score (F1n) was 0.9, AF rhythm score (F1a) of 0.81, and other rhythm score (F1o) of 0.69.