Pulse Harmonic Strength of facial video signal for the detection of atrial fibrillation

It is estimated that 3.2 million people have developed atrial fibrillation (AF) in the United States and 30% of AF patients are unaware of their diagnosis (silent AF). We tested a new technology for contactless detection of AF based on facial video recordings in patients undergoing electrical cardioversion for AF. The proposed technique uses the facial video signal to extract the beat-to-beat variations of the skin color reflecting the cardiac pulsatile signal. We developed the concept of Pulse Harmonic Strength (PHS) to capture AF patterns from this signal. Also, we quantified the variability of the heart rate and pulse rate using additional measurements of heart rate variability. Eleven subjects (65±6 years, 8 males) were enrolled in the study and 407 epochs of 15 sec. were acquired simultaneously with ECG and facial video signals. PHS was associated with a 20% detection error rate of AF in reference to human interpretation while the error rates of the automatic ECG-based measurements ranged between 17% and 29% across the investigated HRV parameters.