ECG-based Cardiovascular Disease Classification using Modified Inception

Cardiovascular diseases are appearing more and more frequently. There is a growing concern about the severity of these diseases. Medical practitioners normally utilize Electrocardiogram (ECG) to detect onset of the disease and give initial diagnosis. Unfortunately, it is costly and laborious for the medical practitioners to analyze a large amount of ECG data. Current methods still lack performance across different recording sessions. On the other hand, manual evaluation ECG signals is a big challenge for inexperienced cardiologists. Recently, application of Deep Learning in ECG-based cardiac disease classification has been achieved. Its performance now is reaching that of the cardiologists. In this paper, a new method to classify ECG-based CVD diseases is proposed . The original ECG signals are transformed into 2D domain and classified by a modified Inception model. The experiments show that the proposed method attains satisfactory performance surpassing recent works.