Sensing, Acquisition, and Interactive Playback of Data-based Models for Elastic Deformable Objects
International Journal of Robotics Research
| Hi-index | 0.01 |
Surgical training has traditionally followed an apprenticeship model: residents observe several procedures, then begin practicing in the operating room. And every time a resident practices his first procedure on a patient, that patient is put at some level of additional risk. Even in specialties where cadaver training is applicable, cadavers are expensive, are available only in limited number, and lack the physiology that guides surgical decision-making. Thus the effectiveness of cadavers in preparing residents for surgical practice is limited. Fortunately, computer-based simulation offers an intermediate between observation and live-patient operation. Virtual environments can allow residents to practice both basic skills and procedural logic at extremely low cost, allowing the presentation of a wide variety of operating room scenarios that cannot be duplicated in cadaver labs. Furthermore, computer-based simulation can offer even experienced surgeons a chance to practice infrequently-performed procedures, to learn new surgical techniques, and to rehearse procedures preoperatively on patient-specific anatomy. An analogy can be made to the highly successful field of flight simulation, which has been routinely used to train and re-educate pilots for decades. However, significant technical challenges stand between today's surgical simulation systems and the virtual operating room that will become a standard part of tomorrow's medical training. Simulators are still limited in rendering quality, immersiveness, intuitiveness, and simulation realism. This thesis addresses some of those challenges, specifically in the context of simulating procedures performed on the temporal bone and mandible. We present an overview of our simulation environment, specifically focusing on how this software delivers the sources of intraoperative feedback that are relevant to training surgical skills. We then discuss a project inspired by this environment, which asks whether haptic feedback can be used to teach motor skills, adding a level of training not available in traditional training labs. We then address one of the most difficult problems in surgical simulation: effectively simulating realistic deformable materials. Specifically, we address the adjustment of an interactive, low-computational-cost deformation model to behave like a more complex model. We then present a series of algorithms and data structures that emerged from this work, and conclude with a discussion on the evaluation of the realism of haptic rendering systems.