Detecting the Types of Arrhythmia Based on the Extraction of Features from the Fractional Fourier Transform of Segmented ECG Signals

Cardiovascular diseases are highly dangerous and are spreading widely. The early diagnosis of these diseases by one of the simplest heart tests can contribute to timely measures. Thus, researchers detect and classify abnormal heart rhythms using electrocardiogram signals. There are different tools to classify heart rhythms. The current study used the Fractional Fourier transform as a new tool to classifiy abnormal heart rhythms. The purpose of this study was to answer the question whether using fractional Fourier transform improves accuracy and reduces errors. At first, the data were obtained from the MIT-BIH Arrhythmia database. After applying the filters and scanning each heartbeat, the fractional Fourier transform coefficients were extracted. After the superior features of the database were extracted, they were classified using the SVM and KNN classifiers. The data were first classified into two classes and then into five classes. First, the optimal alpha was found in each category and then the effect of having references was studied. The results showed that using the SVM classifier with the alpha of 0.7 and 20 superior features gave the best result with the accuracy of 98.97% ± 0.35.

Detecting the Types of Arrhythmia Based on the Extraction of Features from the Fractional Fourier Transform of Segmented ECG Signals Keywords:

Fractional Fourier Transform , SVM , KNN , superior features , electrocardiogram , classification ,