A new approach for heart disease detection using Motif transform-based CWT's time-frequency images with DenseNet deep...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Artificial Intelligence in Healthcare

Background:

• Electrocardiogram (ECG) signals are crucial for diagnosing cardiac conditions like congestive heart failure (CHF) and cardiac arrhythmias (ARR).
• Accurate analysis of ECG signals is vital for timely disease diagnosis and management.
• Advanced computer-aided systems are needed to enhance the analysis of complex ECG data.

Purpose of the Study:

• To develop and evaluate a novel ECG data pattern recognition algorithm for improved cardiac condition diagnosis.
• To address limitations of existing signal preprocessing techniques in ECG analysis.
• To leverage deep learning for accurate classification of ECG patterns.

Main Methods:

• Proposed a novel ECG signal preprocessing model using Continuous Wavelet Transform (CWT).
• Introduced Motif Transformation (MT) to overcome CWT limitations like boundary effects and improve time-frequency localization.
• Classified ECG signals using DenseNET deep transfer learning on scalogram images generated by MT and CWT.

Main Results:

• The integrated MT + CWT + DenseNET approach demonstrated significant effectiveness in ECG data pattern recognition.
• The algorithm achieved a high success rate of 99.31% in classifying cardiac conditions from ECG signals.
• The study highlights the potential of combining signal transformation techniques with deep learning for ECG analysis.

Conclusions:

• The proposed MT + CWT + DenseNET algorithm offers a highly accurate and effective method for ECG signal analysis.
• This approach shows promise for facilitating the early and accurate diagnosis of various cardiac conditions.
• The study validates the synergy between advanced signal processing and deep learning in cardiovascular diagnostics.