Automatic Heart Sound Analysis Framework
Author Information
Author(s): Yuenyong Sumeth, Nishihara Akinori, Kongprawechnon Waree, Tungpimolrut Kanokvate
Primary Institution: Tokyo Institute of Technology
Hypothesis
Can heart sound analysis be performed automatically without the need for segmentation?
Conclusion
The proposed method showed promising results and high noise robustness to a wide range of heart sounds.
Supporting Evidence
- The method achieved an average performance of 0.92 in noise-free conditions.
- It maintained a performance of 0.90 under white noise at 10 dB SNR.
- The system was robust to impulse noise up to 0.3 seconds in duration.
Takeaway
This study created a new way to analyze heart sounds automatically, which can help detect heart problems without needing to break down the sounds into smaller parts.
Methodology
The method involves preprocessing heart sounds, detecting envelope signals, calculating cardiac cycle lengths, and using feature extraction and classification techniques.
Potential Biases
The sounds used for training were recorded for instructional purposes, which may not reflect real clinical recordings.
Limitations
The training set may have introduced bias due to different recording equipment and oversampling of normal sounds.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website