PCG Signal Quality Classification via mRMR Feature Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Affordable smartphones often lack microphone quality for consistent medical-grade phonocardiogram (PCG) recordings, and medically untrained personnel may not adhere to proper protocols, leading to background noise and poor signal quality, which existing methods struggle to address effectively across different devices and environments.
Innovation Solution
A system and method that preprocesses PCG signals using a minimum redundancy maximum relevance (mRMR) algorithm to extract relevant features, rank them, and classify signal quality as diagnosable or non-diagnosable using a random forest classifier, ensuring only high-quality signals are analyzed for cardiac health monitoring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If smartphone microphones are used for PCG recording, then cost is reduced and accessibility is improved, but signal quality deteriorates due to insufficient microphone quality
Solution Approach 1:
A classification system acts as an intermediary between the smartphone microphone and the cardiac analysis process. The system pre-processes and evaluates the raw audio signal to identify diagnostically useful segments, separating the low-quality recording device from the analysis requirements through an automated quality assessment layer.
Solution Approach 2:
The system changes the parameter of signal evaluation by using multiple quality metrics (signal-to-noise ratio, presence of heart sounds, absence of artifacts) rather than relying on the inherent quality of the microphone. This transforms the approach from device-dependent quality to algorithm-dependent quality assessment.
2Measurement precision
If manual quality assessment is used, then signal quality control is improved, but time consumption increases and productivity decreases
Solution Approach 1:
The classification system performs self-service by automatically evaluating signal quality without human intervention. The algorithm independently assesses multiple quality parameters and makes decisions about which segments are suitable for cardiac analysis, eliminating the need for manual review while maintaining quality control.
Solution Approach 2:
The manual mechanical process of expert listening and evaluation is replaced with an automated computational system. The classification algorithm substitutes human experts by processing signals through defined mathematical criteria, achieving both speed and consistency in quality assessment.
3Ease of operation
If heuristically determined thresholds are used for quality classification, then decision-making is simplified, but portability across devices and environments deteriorates
Solution Approach 1:
The classification system achieves universality by designing thresholds and criteria that work across multiple smartphone devices, recording environments, and signal types. The system is configured to handle variations in microphone quality, ambient noise levels, and recording conditions through standardized evaluation metrics that maintain portability.
4Measurement precision
If noisy portions are rejected post-recording, then signal quality is improved, but loss of information increases
Solution Approach 1:
The system performs preliminary classification and identification of usable segments before final analysis. By pre-processing and tagging suitable portions of the recording, the system preserves maximum information while eliminating only the clearly unusable segments, rather than rejecting entire recordings or using post-hoc filtering that may discard valuable data.
Data Source
AI summary
A system and method for classifying the phonocardiogram (PCG) signal quality has been described. The system is configured to identify the quality of the PCG signal recording and accepting only diagnosable quality recordings for further cardiac analysis. The system includes the derivation of plurality features of the PCG signal from the training dataset. The extracted features are preprocessed and are then ranked using mRMR algorithm. Based on the ranking the irrelevant and redundant features are rejected if their mRMR strength is less. A training model is generated using the relevant set of features. The PCG signal of the person under test is captured using a digital stethoscope and a smartphone. The PCG signal is preprocessed and only the relevant set of features are extracted. And finally the PCG signal is classified into diagnosable or non-diagnosable using the relevant set of features and a random forest classifier.