Automated ECG Arrhythmia Classification Using Feature Images with Common Matrix Approach-Based Classifier.

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Cardiac arrhythmias pose a significant diagnostic challenge.
• Accurate classification of arrhythmias from electrocardiogram (ECG) recordings is crucial for patient management.
• Existing methods may require extensive data or exhibit limitations in accuracy.

Purpose of the Study:

• To develop an effective and accurate method for classifying cardiac arrhythmias using a limited number of ECG recordings.
• To propose a novel technique combining Fourier Decomposition Method (FDM) and Common Matrix Approach (CMA).

Main Methods:

• ECG recordings were processed using FDM to generate time-frequency (T-F) representations.
• Data matrices were constructed by concatenating ECG signals, Fourier transforms, and T-F representations.
• Two-dimensional Principal Component Analysis (2DPCA) was applied to create feature images for classification.
• A CMA-based classifier model was utilized with the generated feature images.

Main Results:

• The proposed method achieved a mean overall accuracy of 99.81% on the MIT-BIH database for inter-patient classification.
• Performance exceeded 99% on five metrics for V- and S-class arrhythmia recognition.
• Mean overall accuracies of 99.76% (raw) and 99.45% (de-noised) were obtained for the Chapman database.
• An accuracy of 98.71% was achieved for classifying five arrhythmia types from the PTB-XL database.

Conclusions:

• The proposed FDM-CMA hybrid method demonstrates high efficacy and accuracy in classifying cardiac arrhythmias.
• The technique effectively utilizes limited ECG data, outperforming many recent approaches.
• This method offers a robust solution for automated arrhythmia detection and classification.