liyuejul / Heartbeat-Classification

Classify ECG signals into predefined categories based on heartbeat abnormality by transforming time series to images

Heartbeat Classification

Classify ECG heartbeat signals into predefined categories based on heartbeat abnormality by transforming time series to images

Datasets:

• MIT-BIH Dataset (https://www.physionet.org/content/mitdb/1.0.0/): contains 5 categories: N, S, V, F, Q (see https://arxiv.org/pdf/1805.00794.pdf for details), 87k train records, 21k test records
• PTB Dataset (https://www.physionet.org/content/ptbdb/1.0.0/): contains 2 categories: Normal and Abnormal, 14k records

ECG heartbeat signals are time series data (187 signals per heartbeat in the given data)

Approaches:

• Gramian Angular Field (GAF)
• Recurrence Plot (RP)
• Markov Transition Field (MTF)
• Intermediate fusion