Surgical tool pose estimation is important for surgical image guidance systems as near real-time tool tip position and angle estimation are needed. for example, for intraoperative optical coherence tomography tracking in retinal microsurgery. This paper presents an algorithm that improves on the current state–of- the-art algorithm for surgical tool tracking in posterior eye surgery first introduced by Alsheakhali et al. We propose better tool segmentation based on combined color space masks and set thresholds, a shadow- insensitive tool edge detector, and a more robust tool tip detector. The presented algorithms are benchmarked based on a series of manually annotated images of posterior eye surgery. Our modifications improve the algorithm’s accuracy several-fold in a posterior surgery video case study where the tool shadow has a confounding effect and spot illumination by the endoilluminator occurs at times
About the speaker
Technical Specialist at lCTER, Graduated from Warsaw University of Technology. Flve years of experience in Industrial Robotics/Computer Vision software development
