FBG-based Kalman Filtering and Control of Tool Insertion Depth for Safe Robot-Assisted Vitrectomy

Vitrectomy is that portion of retinal surgery in which the vitreous gel is removed either as a definitive treatment or to provide direct tool access to the retina. This procedure should be conducted prior to several eye surgeries in order to provide better access to the eyeball posterior. It is a relatively repeatable and straight forward procedure that lends itself to robotic assistance or potentially autonomous performance if tool contact with critical structures can be avoided. One of the detrimental incidences that can occur during the robot-Assisted vitrectomy is when the robot penetrates the tool more than allowed boundaries into the eyeball toward retina. In this paper, we provide filtering and control to guide instrument insertion depth in order to avoid tool-To-retina contact. For this purpose, first the tool insertion depth measurement is improved using a Kalman filtering (KF) algorithm. This improved measurement is then used in an adaptive control strategy by which the robot reduces the tool insertion depth based on a predefined and safe trajectory for it, when safe boundaries are overstepped. The performance of the insertion depth safety control system is then compared to one in which the insertion depth is not passed through a Kalman filter prior to being fed to the control system. Our results indicate that applying KF in the adaptive control of the robot enhances procedure safety and enables the robot to always keep the tool insertion depth under the safe levels.