Method of synthesizing and enhancing multiple auscultatory sounds for safe and efficient robotic auscultation

Traditional auscultation requires direct skin contact to ensure optimal diagnostic accuracy. However, this often necessitates that patients remove clothing, which may lead to discomfort or embarrassment - especially in situations involving physicians and patients of different genders. For efficient, safe and gender-free auscultation, we are developing a robotic multi-acoustic sensor array system capable of acquiring auscultatory sounds in the chest and back. Since multiple acoustic sensors are placed for each auscultation position, it is necessary to post-process the acoustic data acquired by each auscultation sensor to generate diagnostic available auscultatory sounds. This study proposes a method to synthesize heart sounds from multiple sensors to enhance clarity. Noise reduction was applied, followed by delay compensation to align heart sound peaks. Weighted synthesis based on peak amplitudes improved the signal-to-noise ratio (SNR). Experiments with six subjects showed SNR improvement in all cases, with three subjects achieving over a 3 dB increase. These results confirm the effectiveness of this method in enhancing auscultation sound quality.Clinical Relevance- This method improves robotic auscultation by enhancing heart sound clarity and reducing noise. It enables non-contact, efficient diagnosis, potentially improving telemedicine, automated screenings, and accessibility for patients with mobility or privacy concerns.