Analysis of Systolic Murmurs in Heart Sounds Using Multiresolution Complex Gabor Dictionary

The automatic diagnosis of heart problems requires a deep understanding of murmurs that indicate various cardiac pathologies. We propose a novel analysis method focusing on systolic murmurs in heart sounds, using a multiresolution complex Gabor dictionary. The dictionary consists of localized complex-valued Gabor functions with various time-frequency scales or resolutions, enabling detailed signal examination. The complex orthogonal matching pursuit (COMP) algorithm is used to represent a systolic murmur segment as a complex vector in terms of the functions in the dictionary. We then calculate the energy of the real and imaginary components of the vector for each resolution to create an energy distribution for all scales. Analyzing these distributions allows for effectively identifying the patterns of various systolic murmurs, which improves diagnostic accuracy. The experimental analysis of a dataset composed of systolic heart murmurs with four patterns proves the effectiveness of the proposed analysis dictionary and methodology.