Respiratory phase detection from optical phonocardiography characteristics

This paper presents a respiratory phase prediction technique from an optical phonocardiograph (PCG) signal. The PCG acquisition was conducted using a speckle-based sensor which includes illumination of the inspected subjects by a laser beam and analyzing the temporal changes in the spatial distribution of the back scattered secondary speckle patterns. From the analysis of the 2D speckle patterns a 1D nano vibrations signal was extracted. Then, we performed an analysis of this 1D signal while relying on the PCG extracted features used in Naïve Bayes model. The performance accuracy for the respiratory phase prediction conducted over four subjects was 83%. The high accuracy made possible thanks to 9 spatial illumination spots used in our optical sensor and using a decision algorithm involving spots' combination (while each one of the 9 spots illuminating the chest of the inspected subjects was analyzed separately).