Abstract
We propose a computer integrated endoscopic simulator for training in upper endoscopy as a low-cost alternative to the traditional training methods and virtual reality simulators. The use of a real endoscope in conjunction with our simulator and computer system in an actual operating room setup makes the training environment similar to a real procedure. Endoscopic surgery is the performance of surgery through a small incision with the aid of special medical equipment called a flexible endoscope. The advantage of this technique over open surgery is that there is significantly less operative trauma, resulting in less pain and a shorter recovery time. Side effects of the surgery, such as the risk of infection, are also reduced. While endoscopy procedure has tremendous benefits, surgeons require considerable practice and time to develop competency. Traditionally, the procedure has been taught at the expense of patient comfort and safety, in other words, gastroenterology training fellows have performed the surgery under the supervision of physicians. Patients who undergo the endoscopies performed by fellows, particularly early in the training period, have been more likely to suffer more discomfort and prolonged procedures. In this study, we introduce a new type of simulator which combines the use of mechanical model and computer system as an additional or low-cost alternative for training in upper endoscopy. Our approach is to integrate a computer system with a realistic mechanical model to create a computer-based simulator for upper endoscopy training. The simulator will cover the basics of flexible endoscopy and teach a trainee the skills required to perform upper endoscopy. The mechanical training model with a sensor system that simulates a human upper gastrointestinal tract, including pathologies such as ulcers and polyps, will be built and integrated with computer software. The software offers the following functions: provides help to the trainee, provides curriculum-required learning tasks, and assesses the performance and diagnostic skills. Due to the optical nature of an endoscopic lens, the obtained image suffers from a barrel-type spatial distortion, which results in an inconsistent measurement of object size and distance. Our distortion correction system with automatic calibration, based on least squares estimation, offers a better perception of size and distance from the endoscopic images. In order to examine the endoscopic maneuvering skills of the trainee, the automatic evaluation system is created. The system uses images from the exam procedure to verify the trainee skills. We use Support Vector Machine to classify endoscopic images of different regions in upper gastrointestinal tract. The experimental results on the distortion correction and image classification are reported. Simulator validation survey result from gastroenterology surgeons and fellows is included in this dissertation. A recommendation for further study is also enclosed.