The application of object detection in the surgical scene of thyroidectomy

Accurate identification and localization of objects during thyroidectomy surgery are crucial for ensuring surgical precision and ensuring patient safety. Although tool recognition in minimally invasive surgeries has been extensively covered in previous literature, open surgeries, including thyroidectomy, have not received comparable attention in this aspect. To tackle this challenge, We propose a comprehensive object detection framework for thyroidectomy surgical scenes using multi-angle camera imaging. Our study is centered around thyroidectomy and presents an innovative workflow for collecting surgical images using multi-angle cameras. Additionally, we introduce computer vision technology to analyze these images. By implementing the YOLOv5 target detection algorithm, we can precisely identify crucial target objects within intricate surgical scenarios. We acquired surgical images related to thyroidectomy and meticulously prepared them to ensure compatibility with neural networks. Subsequently, we trained a model using the target detection algorithm, enabling the successful identification of significant targets within complex surgical scenarios. The experimental results unequivocally demonstrate this method's outstanding performance and practicality in thyroidectomy. It enhances the efficiency and precision of surgical procedures while providing surgeons with enhanced visual assistance and decision support. As technology advances and evolves, the future prospects for applying object detection in the medical field are poised to expand even further.