Contactless heart rate and heart rate variability estimation from neck videos

Area of Science:

• Biomedical Engineering
• Physiological Monitoring
• Signal Processing

Background:

• Video-based pulse extraction is a non-contact physiological monitoring technique.
• Traditional methods focus on facial regions, limiting applications due to privacy concerns.
• The neck region is an understudied area for pulse signal extraction.

Purpose of the Study:

• To investigate the feasibility of extracting pulse signals from the neck region using video recordings.
• To compare the performance of six different video-based pulse extraction methods on neck data.
• To validate the accuracy of extracted pulse signals against electrocardiogram (ECG) recordings.

Main Methods:

• Collected neck and chest video data from 14 healthy subjects during breath-hold.
• Tracked a region of interest on the neck using a template-based algorithm.
• Processed pixel intensity variations using six established methods: GREEN, chrominance-based, plant-orthogonal-to-skin, orthogonal matrix image transformation, independent component analysis, and local group invariance.

Main Results:

• The GREEN method showed the highest agreement with ECG-derived heart rate (HR).
• Initial HR agreement had a bias of -4.89 bpm and LoA of -29.97 to 20.17 bpm.
• After excluding noisy data, agreement improved to a bias of -1.42 bpm and LoA of -4.95 to 2.09 bpm.
• Heart rate variability (HRV) using SDNN was generally higher with the GREEN method compared to ECG.

Conclusions:

• The neck region is a viable site for video-based pulse signal extraction and HR estimation, especially with the GREEN method.
• Further refinement of video-based pulse extraction techniques is needed, particularly for HRV assessment.
• Neck-based video pulse extraction offers promising applications for remote health monitoring, addressing privacy concerns associated with facial recordings.