2017/10/10 by Jonathan Rubin, Rubin, Jonathan, Saman Parvaneh +8
Medicine · Neuroscience · #Atrial Fibrillation Management and Outcomes #Computer Vision and Pattern Recognition (cs.CV) #ECG Monitoring and Analysis #EEG and Brain-Computer Interfaces #FOS: Computer and information sciences #FOS: Electrical engineering #Machine Learning (stat.ML) #Signal Processing (eess.SP) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1710.05817
openalex publication_date 2017/10/10 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
The development of new technology such as wearables that record high-quality\nsingle channel ECG, provides an opportunity for ECG screening in a larger\npopulation, especially for atrial fibrillation screening. The main goal of this\nstudy is to develop an automatic classification algorithm for normal sinus\nrhythm (NSR), atrial fibrillation (AF), other rhythms (O), and noise from a\nsingle channel short ECG segment (9-60 seconds). For this purpose, signal\nquality index (SQI) along with dense convolutional neural networks was used.\nTwo convolutional neural network (CNN) models (main model that accepts 15\nseconds ECG and secondary model that processes 9 seconds shorter ECG) were\ntrained using the training data set. If the recording is determined to be of\nlow quality by SQI, it is immediately classified as noisy. Otherwise, it is\ntransformed to a time-frequency representation and classified with the CNN as\nNSR, AF, O, or noise. At the final step, a feature-based post-processing\nalgorithm classifies the rhythm as either NSR or O in case the CNN model's\ndiscrimination between the two is indeterminate. The best result achieved at\nthe official phase of the PhysioNet/CinC challenge on the blind test set was\n0.80 (F1 for NSR, AF, and O were 0.90, 0.80, and 0.70, respectively).\n