Using Bayesian optimization to guide probing of a flexible environment for simultaneous registration and stiffness mapping

Abstract

One of the goals of computer-aided surgery is to match intraoperative data to preoperative images of the anatomy and add complementary information that can facilitate the task of surgical navigation. In this context, mechanical palpation can reveal critical anatomical features such as arteries and cancerous lumps which are stiffer that the surrounding tissue. This work uses position and force measurements obtained during mechanical palpation for registration and stiffness mapping. Prior approaches, including our own, exhaustively palpated the entire organ to achieve this goal. To overcome the costly palpation of the entire organ, a Bayesian optimization framework is introduced to guide the end effector to palpate stiff regions while simultaneously updating the registration of the end effector to an a priori geometric model of the organ, hence enabling the fusion of intraoperative data into the a priori model obtained through imaging. This new framework uses Gaussian processes to model the stiffness distribution and Bayesian optimization to direct where to sample next for maximum information gain. The proposed method was evaluated with experimental data obtained using a Cartesian robot interacting with a silicone organ model and an ex vivo porcine liver.

Concurrent nonparametric estimation of organ geometry and tissue stiffness using continuous adaptive palpation

“…In complementary work to that reported here, our collaborators at Carnegie Mellon University are exploring the use of GP in a Bayesian optimization framework to identify stiff regions by probing the organ at a smaller number of discrete locations while simultaneously registering the end effector to an a-priori geometric model of the organ [17]. One advantage of discrete probing is that it permits accurate assessment of tissue stiffness and surface location at the point probed.…”

Trajectory-Optimized Sensing for Active Search of Tissue Abnormalities in Robotic Surgery

“…In order to reduce the exploration time, Bayesian optimization-based approaches have been developed for tumor localization by directing the exploration to stiff regions [14]- [18]. These approaches model tissue stiffness as a distribution defined on the surface of the organ where each point on the surface is associated with a random variable.…”

“…The assumption is that finding the global maxima of the stiffness distribution correspond to locating the stiff inclusions. Ayvali et al [14] sequentially select the next location to probe the organ, and predict the stiffness distribution and the location of the global maximum after every measurement, while Chalasani et al [15] update after collecting several samples over finite time along a trajectory that directs the robot to the high stiffness regions. In a more recent work, Chalasani et al [18] incrementally estimate local stiffness and geometry while the organ is palpated along predefined trajectories or under telemanipulation.…”

“…Prior works commonly demonstrate results using a single stiff inclusion (single maximum) [15], [16]. When multiple inclusions are present (multiple global and local maxima) the algorithm is initialized with a coarse grid to ensure exploration of all regions [14].…”

A surgical system for automatic registration, stiffness mapping and dynamic image overlay

“…This work has been funded through the National Robotics Initiative by NSF grant IIS-1426655. While the works in literature deal with force sensing [10], [11], tumor localization [2], [4]- [6], [12] and graphical image overlays [13]- [16], there is a gap in literature when it comes to systems that deal with all these issues at the same time. For example, Yamamoto et al [16] deal with tumor localization and visual overlay, but they assume the organ is flat and place the organ on a force sensing plate, which is not representative of a surgical scenario.…”