Development of an Accelerometer-Based Device for Medical Percussion Analysis: A Novel Approach Using Vibration Signals

Medical percussion is a physical assessment technique that involves tapping on the body surface to evaluate a patient’s condition, serving as a diagnostic method for comprehensive examination. Although this method has been used and established over centuries, it is considered a qualitative assessment due to its high inter-examiner variability. Furthermore, there is currently no standardized instrument used in clinical practice. This study proposes a system to improve the manual percussion procedure through the analysis of percussion signals using an accelerometer. Repeatable mechanical impacts were generated using a pulse width modulation (PWM)-controlled vibration motor and a push-pull solenoid. Surface vibrations were measured with an accelerometer, and the signals were analyzed based on different media and volumes. More specifically, the vibration motor-based device generates surface vibrations through a frequency sweep in the range of 150–250 Hz and provides real-time measurements via an accelerometer. Damped vibrations from solenoid activation were classified using a convolutional neural network (CNN)-based model with images generated through time-frequency analysis. The developed prototype was experimentally validated using a positioning system based on the coordinate system of a computer numerical control (CNC) machine. The experimental reservoir for vibration measurement was prepared with air, water, and solidified gelatin in varying ratios while maintaining a constant total volume. Consequently, it was confirmed that vibration signals generated from different media and volumes could be effectively distinguished.