Image recognition of external auditory canal for controlling ear endoscope robot

Abstract

In transcanal endoscopic ear surgery (TEES), a locally operated surgical support robot holding an endoscope in a sterile area has been developed for a stable and clear visual field. It is important to set up the endoscope robot smoothly, and initial positioning of the endoscope lens is needed. Thus, the autonomous position setting of the endoscope robot for inserting the lens into an ear canal is required at first. Therefore, image recognition of the external auditory canal is proposed for controlling the endoscope robot. In this study, ear models are recognized by RGB images in the YOLACT++ algorithm, and the black colored external auditory canal is estimated.