Machine Learning-Driven Robotic Surgery: Advancing Precision, Efficiency, and Accessibility

Abstract

Robotics and Artificial Intelligence (AI) are increasingly being integrated across various sectors, including healthcare. In robotic surgery, however, achieving high precision remains a critical challenge, as even minor errors can have life-threatening consequences. This study employs the YOLOv5 model developed by Ultralytics for surgical tool detection, addressing the fundamental need for accurate visual recognition. To enhance performance beyond detection, the SORT algorithm is integrated for real-time tracking of surgical instruments, significantly improving accuracy, precision, and operational adaptability in diverse environments. The proposed system is evaluated using the Cholec80 dataset, with the model undergoing iterative fine-tuning after every ten surgical procedures to optimize its weights and improve overall precision. Comparative analysis against existing approaches demonstrates that the YOLOv5-SORT integration achieves superior detection and tracking performance. The findings confirm that this approach enhances both the reliability and flexibility of AI-assisted robotic surgery systems, paving the way for safer and more precise medical interventions.

Ähnliche Arbeiten