Research on nighttime IPPG algorithm based on ROI delay expansion and fundamental frequency constrained FastICA

Area of Science:

• Biomedical Engineering
• Signal Processing
• Remote Sensing

Background:

• Nighttime heart rate (HR) monitoring using imaging photoplethysmography (IPPG) faces challenges from motion artefacts and difficulties in identifying optimal regions of interest (ROIs).
• Existing IPPG methods struggle with accuracy and reliability during sleep due to these inherent limitations.

Purpose of the Study:

• To enhance the accuracy and reliability of nighttime HR measurements using IPPG.
• To develop an innovative approach combining FastICA with a Time-Delayed Multi-Dimensional Extended Regions of Interest Extraction (TDMDE-ROI-Ex) technique.
• To overcome challenges posed by motion artefacts and ROI identification in nighttime IPPG.

Main Methods:

• A dual-method strategy involving face detection, grayscale clustering for ROI pinpointing, and mutual information delay for multi-channel IPPG signal synthesis.
• Application of HR's Fundamental Frequency as a prior Constraint within the iterative process of FastICA (HRFFC-FastICA) to mitigate initial value fluctuations.
• Validation using the MR-NIRP dataset, followed by ablation studies and comparative evaluations against existing nighttime IPPG algorithms.

Main Results:

• The proposed HRFFC-FastICA method achieved a mean absolute error (MAE) of 4.57 bpm and a root mean squared error (RMSE) of 5.95 bpm.
• Demonstrated significant improvements over SparsePPG and PhysNet, with MAE enhanced by up to 8.39 bpm and RMSE reduced by 17.83 bpm.
• Achieved a narrower 95% Bland-Altman confidence interval (9.5 to -12.8 bpm) compared to alternative methods, indicating superior precision.

Conclusions:

• The TDMDE-ROI-Ex method significantly reduces reliance on facial motion for ROI identification.
• HRFFC-FastICA effectively counteracts motion artefacts and initial value sensitivity in FastICA.
• The integrated methodology substantially enhances the robustness and stability of nighttime IPPG monitoring, expanding its application range.