Automatic Puncture Timing Detection for Multi-Camera Injection Motion Analysis

Precisely detecting puncture times has long posed a challenge in medical education. This is attributable not only to the subjective nature of human evaluation but also insufficiency effective detection techniques, resulting many students lacking full proficiency injection skills upon entering clinical practice. To address this issue, we propose novel method that enables automatic during without needing wearable devices. In study, utilized hardware system and YOLOv7 algorithm detect critical features motion, including time depth parameters. We constructed sample 126 training videos students, skilled observers were employed determine accurate times. Our experimental results demonstrated mean was 2.264 s identification error 0.330 s. Moreover, confirmed there no significant difference (p = 0.25 with significance level α 0.05) between predicted value ground truth, which provides basis for validity reliability system. These show our system’s ability automatically provide approach healthcare professionals. At same time, it key technology future development skill assessment systems.