Non-contact heart vibration measurement using computer vision-based seismocardiography

Area of Science:

• Biomedical Engineering
• Cardiovascular Physiology
• Computer Vision

Background:

• Seismocardiography (SCG) noninvasively measures chest wall vibrations from heart activity.
• SCG shows potential for diagnosing cardiovascular diseases like heart failure and ischemia.
• Traditional SCG requires chest-worn accelerometers.

Purpose of the Study:

• To develop and evaluate a novel contactless SCG measurement method using smartphone videos.
• To extract SCG signals from chest movements captured by a smartphone camera.
• To assess the feasibility of remote cardiovascular monitoring via smartphone video analysis.

Main Methods:

• Utilized computer vision techniques, specifically Lucas-Kanade template tracking, to monitor chest displacements.
• Tracked an artificial target attached to the chest from video recordings.
• Estimated SCG signals from the tracked displacements and validated against accelerometer-based SCG.

Main Results:

• High similarity between vision-based and accelerometer-based SCG signals (average DTW index of 0.94-0.95).
• Accurate heart rate estimation from video-derived SCG signals with low bias (0.649 beats/min) compared to ECG.
• Demonstrated the potential for reliable SCG signal extraction from smartphone videos.

Conclusions:

• A novel, contactless SCG measurement method using smartphone videos has been successfully developed and validated.
• This technology offers a low-cost, accessible approach for remote monitoring of cardiovascular activity.
• Future applications may include widespread use in telemedicine and personal health tracking.