Vision-Guided Autonomous Robotic Electrical Bio-Impedance Scanning System for Abnormal Tissue Detection

Abstract

In cancer surgery, the correct localization of cancerous tissue is critical for the success of the surgery. Specifically, the surgical goal is to remove the malignant tissue completely while preserving the maximum amount of healthy tissue. Unfortunately, limited technologies are available for intra-operative assessment of tumor margin in Robotic Minimally Invasive Surgery (RMIS). This work presents a vision-based autonomous robotic scanning system for abnormal tissue detection. A stereo vision system, including dense 3D surface reconstruction and tissue tracking, is used to control the scanning motion to compensate for tissue motion and deformation. A high level control strategy allows to control a robot arm to scan autonomously the tissue by means of a monopolar forceps measuring the Electric Bio-Impedance (EBI) of the tissue. The acquired information is used for discriminating cancerous from healthy tissue in real-time, and this information is presented as an impedance map overlaid on the endoscopic video using Augmented Reality (AR). The proposed system performance is evaluated on ex-vivo animal tissues, through experiments simulating the identification and localization of hepatic cancer for resection surgery. Quantitative results show that the system is able to autonomously scan the area of interest and provide a tissue characterization with an accuracy over 82%.