Key result
Left cheek analysis outperforms other facial regions for accurately detecting HRV from images.
Why the study?
Variations in light diffusion across facial epidermis make identifying optimal facial regions for contactless heart rate measurement challenging, and prior studies lacked valid justification for their chosen regions of interest.
Which facial region of interest is best suited for accurately detecting heart rate variability using image-based non-contact devices?
Which facial region of interest is best suited for accurately detecting heart rate variability using image-based non-contact devices?
The left cheek appears to be the optimal region of interest for accurate contactless heart rate variability detection using facial images.
Left cheek ROI may optimize contactless HRV detection; leaves open validation in clinical populations.
Recent interest in healthcare has led to increased research on image-based non-contact health diagnosis devices, particularly machine learning models that measure heart rates using facial images. Variations in light diffusion information caused by inconsistent epidermis on the face make it difficult to determine which areas of the face should be targeted during the actual application. A few recent studies have reported experimental results for their designated regions of interest (ROIs) without providing a valid justification for their selection. This study's primary objective is to provide a comprehensive analysis of the various facial regions to determine which ROI is best suited for accurately detecting heart rate variability (HRV). Three anatomically divided facial regions were analyzed using the Fast Fourier Transform (FFT) and FFT combined with Principal Component Analysis (PCA) under both laboratory and controlled illumination conditions, using sliding windows of 3 × 3 and 8 × 8 dimensions for real-time data analysis. The study revealed a high correlation between acquired results and ground truth data when detecting HRV using a particular ROI compared to the other ROIs. Under laboratory and controlled illumination conditions, the ROI on the left cheek performed remarkably well. In addition, significant results were also observed for the forehead and left facial regions of interest (ROIs) under controlled illumination.
No takes yet. Share an insight, caveat, or question.
Sharma et al. (2024) studied Heart rate variability detection. Facial regions of interest (ROI) analysis using FFT and PCA vs. Other facial ROIs was evaluated on Accuracy of detecting heart rate variability (HRV) compared to ground truth. The left cheek region of interest performed best for accurately detecting heart rate variability using facial images under both laboratory and controlled illumination conditions.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: