Morphological processing of physiological signals for feature extraction

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Physiological signals are often complex, nonlinear, and non-stationary, posing challenges for traditional analysis.
• Accurate feature extraction is crucial for diagnosing various medical conditions from biomedical data.

Purpose of the Study:

• To propose and validate a novel method for extracting features from physiological signals.
• To enhance the characterization of complex biomedical signals using intrinsic mode decomposition and morphological signal processing.

Main Methods:

• Decomposition of physiological signals into intrinsic mode functions (IMF) using intrinsic mode decomposition (IMD).
• Application of morphological signal processing (MSP) to each IMF to extract features like pattern spectrum entropy.
• Extraction of additional features including energy and sample entropy from individual IMFs and their cumulative sums.

Main Results:

• The proposed method successfully extracts shape-size complexity features (pattern spectrum entropy) from decomposed signal components.
• Features derived from IMFs and their cumulative sums effectively characterize the physiological signals.
• The method's efficacy is demonstrated using heart sound signals from patients with different cardiac conditions.

Conclusions:

• The combination of IMD and MSP offers a robust approach for feature extraction from complex physiological signals.
• This novel method provides valuable insights into signal characteristics, aiding in the diagnosis of cardiac conditions.
• The technique holds potential for broader applications in biomedical signal analysis.