Cardiovascular Sound Classification via Sub-segment Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting coronary artery stenosis using auscultatory sounds are inefficient due to noise interference and variability among individuals, making it difficult to accurately diagnose coronary artery disease (CAD) with conventional techniques, especially when using digital stethoscopes which provide poor sound quality and limited data.
Innovation Solution
A method that classifies cardiovascular sounds by dividing diastolic and systolic segments into sub-segments with good signal-to-noise ratios, discarding noisy segments, and using multiple signal parameters to account for individual variations, allowing for robust classification with minimal recorded data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional auscultatory techniques are used to detect coronary artery stenosis, then diagnostic capability is provided, but measurement precision deteriorates due to noise interference and individual variability
Solution Approach 1:
The patent divides the cardiovascular sound signal into multiple segments and further subdivides each segment into sub-segments. This segmentation allows the system to identify and discard noisy sub-segments while retaining clean ones for analysis, thereby improving measurement precision by reducing the impact of noise interference and individual variability in the overall signal.
2Ease of operation
If digital stethoscopes are used to acquire auscultatory sounds, then ease of operation is improved, but sound quality deteriorates due to additional noise and limited data
Solution Approach 1:
The patent extracts and discards noisy sub-segments from the recorded cardiovascular sound, separating the useful signal components from the harmful noise. By taking out only the clean sub-segments for analysis, the system maintains ease of operation with digital stethoscopes while improving sound quality and reliability of the diagnostic data.
3Measurement precision
If long recording duration is used to improve classification accuracy, then measurement precision is improved, but loss of time increases
Solution Approach 1:
By segmenting the recording into multiple sub-segments and selecting only the clean ones for analysis, the patent achieves high classification accuracy without requiring long recording durations. This segmentation approach allows the system to reach reliable diagnostic conclusions from shorter recordings by concentrating analysis on high-quality signal portions.
4Measurement precision
If conventional analysis techniques are used, then device complexity is reduced, but measurement precision deteriorates due to inability to account for individual variations
Solution Approach 1:
The patent extracts multiple signal parameters from different sub-segments and uses these varying parameters to account for individual variations in cardiovascular sounds. By changing and analyzing multiple parameters across segmented sub-segments, the system achieves high measurement precision while managing device complexity through systematic parameter extraction and comparison.
Data Source
Figure 1~2
Figure 3
Figure 4a~4b
AI summary
The present invention relates to a method for classifying a cardiovascular sound recorded from a living subject. The method comprises the steps of: identifying diastolic and/or systolic segments of said cardiovascular sound; dividing at least one of said identified diastolic and/or systolic segments into a number of sub-segments comprising at least a first sub-segment and at least a second sub-segment; extracting from said first sub-segment at least a first signal parameter characterizing a first property of said cardiovascular sound, extracting from said second sub-segment at least a second signal parameter characterizing a second property of said cardiovascular sound; classifying said cardiovascular sound using said at least first signal parameter and said at least second signal parameter in a multivariate classification method. Furthermore, the invention relates to a system, stethoscope and server for classifying a cardiovascular sound recorded from a living subject, where the above-described method has been implemented.