Virtual Rotator Cuff Arthroscopic Skill Trainer

• Main Author: Oumimoun, Khalil
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01.01.2021
• Subjects: Artificial intelligence; Biomedical engineering; Computer science; Surgery
• ISBN: 9798780621652

Arthroscopy is a prevailing minimally invasive surgical procedure for the diagnosis and treatment of joint-related injuries. The number of arthroscopic procedures has significantly increased in comparison to the traditional open surgeries.Among the joint-related injuries, shoulder injuries are common with people who repeatedly perform overhead movements, such as athletes or construction workers. It is estimated that more than 460,000 patients undergo some type of shoulder surgery per year in the United States alone. The arthroscopic method of performing shoulder surgery is minimally invasive, where the entire procedure is performed using an arthroscope, a small rigid fiber-optic camera with a light source. The anatomy and environment are displayed on a 2D monitor screen streamed from the fiber-optic camera. In addition, the surgeons also use the small incisions to insert surgical instruments in order to perform surgery-specific tasks (e.g., probing or removing tear tissue). Despite its popularity and many benefits, arthroscopy is one of the most challenging minimally invasive surgeries to train surgical residents. It poses several challenges such as, the constrained field of view, the unnatural hand-eye coordination, and the limited instrument control. The shoulder arthroscopy requires a great deal of coordination, precision, and practice. The subjects can acquire the skills with frequent training and practice during the residency program. However, conventional training methods such as the apprenticeship model, the use of cadavers, and plastic mannequins are risky, costly, and non-realistic. Surgeons are in need of realistic and reusable training platforms that could be used to practice their knowledge, psychomotor skills, and cognitive skills. Virtual reality based surgical simulations offer novel, risk-free, accessible, cost-effective, and high-fidelity realistic training platforms.The purpose of this study was to design and develop a virtual training platform for rotator cuff arthroscopic surgeries. This training platform is called Virtual Rotator Cuff Arthroscopic Skill Trainer (ViRCAST) and aims at creating a photorealistic VR-based training and assessment environment for the rotator cuff procedure. In ViRCAST, the subjects are able to practice the Landmark Identification tasks where they can navigate the shoulder anatomy to correctly identify all critical anatomical landmarks within the shoulder. In addition, the simulation provides a shaving/drilling task, where the subjects can practice the drilling for anchor placement. The subjects could also perform suture thread passing following the anchor deployment and perform a full repair of the tear by tying and securing the knots of the anchor suture threads.This work, as a part of the ViRCAST, presents improvements on existing surface extraction techniques and introduces a novel GPU-based technique that combines the marching cubes algorithm and adaptive mesh refinement for real-time drilling simulation and haptic force feedback. As part of ViRCAST, I also present a network of cyber-physical systems that communicates wirelessly and serve as the primary hardware control for the simulation. Finally, we discuss the general issues with cyber-physical systems used in real and simulated surgical rooms and introduce a solo-checkpointing co-recovery mechanism to minimize checkpointing overhead while maintaining the systems’ reliability. We present the results of this mechanism, where we perform random fault injections for testing and validation in our simulation. Results show that our mechanism can mask almost all injected faults with a negligible fault detection latency.