Auscultatory Noise Detection via Cross-Correlation Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current coronary artery disease detection systems face challenges in accurately sensing auscultatory sounds from the thorax, as conventional stethoscopes may not detect cardiovascular-condition-specific sounds below audible levels, and existing sensors can be detached or debonded from the skin, leading to unreliable data acquisition.
Innovation Solution
The auscultatory coronary-artery-disease detection system employs multiple auscultatory sound sensors acoustically coupled to the thorax via hydrogel adhesive interfaces, with a data recording module and docking system that preprocess and analyze breath-held auscultatory sound signals to detect sensor decoupling and noise, ensuring reliable data acquisition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional stethoscopes are used to detect auscultatory sounds, then the device complexity is low, but the measurement precision deteriorates because cardiovascular-condition-specific sounds below audible levels cannot be detected
Solution Approach 1:
The patent replaces conventional mechanical stethoscope systems with electronic sensors that convert acoustic energy to electrical signals for processing. This substitution enables detection of sounds below audible levels by transforming the detection mechanism from purely mechanical to electro-acoustic, thereby improving measurement precision while accepting increased system complexity
Solution Approach 2:
The patent transitions from single-point acoustic detection to multi-dimensional signal processing by capturing auscultatory sounds across multiple frequencies and time domains. This dimensional expansion in signal analysis enables detection of subtle cardiovascular conditions that would be imperceptible to conventional stethoscopes
2Measurement precision
If sensors are attached to the thorax for continuous monitoring, then the measurement precision improves, but the reliability deteriorates when sensors become detached or debonded from the skin
Solution Approach 1:
The patent implements a feedback mechanism that continuously monitors sensor attachment status by analyzing signal characteristics. When detachment or debonding is detected through signal quality degradation, the system generates alerts and can trigger reattachment procedures, thereby maintaining reliability despite the inherent instability of skin-mounted sensors
Solution Approach 2:
The patent employs preliminary measures to prevent sensor detachment by using secure attachment mechanisms and adhesive interfaces designed to maintain stable contact during patient movement. This preemptive approach cushions against the reliability problem before it occurs, ensuring continuous high-quality signal acquisition
3Measurement precision
If multiple sensors are deployed to improve detection accuracy, then the measurement precision improves, but the device complexity increases
Solution Approach 1:
The patent merges multiple sensor signals through signal processing techniques that combine information from multiple sources. By integrating data from multiple sensors and processing them collectively, the system achieves improved detection accuracy while managing complexity through unified processing algorithms rather than treating each sensor independently
Data Source
AI summary
A time-series array of noise data is generated from an inverse frequency transform of the product of the frequency spectrum of an auscultatory sound signal with an associated noise filter generated responsive to a cross-correlation of frequency spectra of auscultatory sound signals from adjacent auscultatory sound-or-vibration sensors on the torso of a test subject. Noise power within at least one range of frequencies of average of frequency spectra from a plurality of windows of the time-series array of noise data is compared with a threshold to determine whether or not the associated auscultatory sound-or-vibration sensor is excessively noisy. In one embodiment, the noise filter is generated by subtracting from unity, a unity-normalized cross-correlation of frequency spectra of the auscultatory sound signals, wherein the resulting values are clipped so as to be no less than an associated noise floor.