Laparoscopic surgery has become one of the most commonly performed Minimally Invasive Surgery (MIS). Trocar insertion is the first step in any laparoscopic procedure. A majority of injuries during MIS is attributed to excessive use of force by surgeon during trocar insertion [7]. It is a difficult procedure to learn and practice because it is carried out almost entirely without any visual feedback of the organs underlying the tissues being damaged. Therefore, it a training system with haptic feedback will be very beneficial.;Challenges in developing such a feedback system are many. For example, characterizing accurate biomechanical properties of tissues from experimental data, integrating proper deformation mechanism of tissues into haptic feedback, and advanced visualization techniques are just a few of them.;In this research, to extract reliable relationship between force/torque and deformation of the abdomen walls in the thorax area, trocar insertion data (force/torque, time, displacement, etc.) was collected by inserting specially instrumented trocar into a few specimens of pig tissues. The instrumentation used included a force/torque sensor, aurora sensors, and a 6 DOF haptic device. Using this experimental data set and accurate simulation of the experimental trocar insertion process, an optimization scheme is proposed to stochastically characterize the biomechanical properties of tissues based on non-linear and large strain models. Commercially available high-level programming software, MATLAB, and finite element (FE) software, ABAQUS, are used for this purpose. The predicted material properties and deformation mechanisms have been cross-validated with a different set of experimental results. This provides sufficient confidence in reliably using the estimated material properties and deformation mechanism to further develop a virtual reality system for trocar insertion procedure with haptic feedback. A new graphic deformation system based on Artificial Neural Networks (ANN) is proposed. Our proposed ANN framework consists of two separate neural networks. The first ANN models the force (haptic) feedback of the trocar insertion procedure and synthesizes appropriate reaction force based on clinical data through a haptic device. The second ANN models the mechanism of tissue deformation. We train this second ANN model using the FE computed deformation data for real time rendering of appropriate tissue deformations. The virtual training system is finally simulated based on these two ANN models for tissue deformation and force feedback in real time. This novel method allows precise trocar insertion simulation based on prior offline FE analysis.
展开▼
