Endoscopic Surgery Project

---

To improve visualization during endoscopic spine surgery (ESS), I co-developed a deep learning algorithm capable of detecting active bleeding points in real-time surgical footage. Using convolutional neural networks trained on annotated intraoperative videos, the model achieved over 95% accuracy in distinguishing bleed sites from background tissue. It localized bleeding points within a 1.1 mm margin — well within the clinical threshold for hemostatic precision — and operated with minimal latency. This tool has the potential to reduce operative time, enhance surgical safety, and lower barriers to adopting ESS in clinical practice.