Virtual Reality Training for Ureteroscopic Skills

Currently, the most widely distributed virtual reality simulator in urology is UroMentor.k It runs on a personal computer and offers a high-level object-oriented application program interface available for use with either Microsoft Direct X or Open GL platforms (67). Simultaneous endoscopic and fluoroscopic views are displayed on the monitor and the user interacts with a mannequin with a multitude of virtual instruments including a flexible cystoscope, and rigid and flexible ureteroscopes. Tracking of the scopes with relation to the model is transmitted via a sensor on the endoscope and three sensors in the workstation. The model is volumetric in nature and created from data from digitized computer tomography and magenatic resonance imaging and the textures were obtained from endoscopic images during a real procedure. Like UW transurethral resection of the prostate, UroMentor provides a modular approach to train important subtasks for ureteroscopic and cystoscopic skills sets.

Shah and Darzi showed construct validity of a visual-spatial subtask cystoscopic module in UroMentor, which assesses the user's ability to find 10 flags in the bladder, which indicates that they have inspected the entire bladder. Seven experts outperformed 10 novices on the first trial both in the mean number of flags seen (9.57 vs. 8.0, P = 0.01) and the mean time to complete the task (2.33 vs. 4.89 minutes, P = 0.03). The experts did show improvement in speed between trials 1 and 2, but then remained the same up until trial 10. The medical students improved between the first and the 10th trial as far as speed (P = 0.0005) and approached significance with regards to seeing more flags (P = 0.05) (68). This study provided construct validity for the cystoscopy module, as experts outperformed novices presumably based on their prior experience on real cases. The authors also questioned the face validity, as finding flags did not accurately represent finding lesions in the bladder, though the fact that experts felt the simulator was realistic argues against this statement. Because the same instrument that was used for evaluation was also used for training and correlation with performance in the operating room was not examined, predictive validity was not established in this study (68).

Two other groups examined populations of novices (medical students) and UroMentor's ability to simulate and train ureteroscopic skills. Groups from the University of Texas Southwestern and Western Ontario University independently evaluated the acquisition of basic ureteroscopy skills using UroMentor using medical students (N = 21 and 20, respectively). The students were tested and randomized to training versus no training and retested on the simulator. Subjective assessment by a single subject matter expert (Western Ontario University) and two subject matter experts who were blinded to the randomization schedule was also performed. Significant improvements in performance were evident in multiple parameters within the trained groups in both studies. The authors believed the simulator to have face validity and convergent validity, as the subjective evaluation seemed to correlate with parameters tracked by the simulator. In a later study, the University of Texas Southwestern group confirmed construct validity by stratifying performance on UroMentor with clinical experience (69,70).

The group at University of Texas Southwestern also examined concurrent validity of the ureteroscopic module by examining whether training on UroMentor translated to improvement of performance on a cadaver. Sixteen medical students and 16 residents had baseline skills evaluation with UroMentor. Just the student group then underwent five hours of supervised training on the simulator and the two groups repeated the UroMentor ureteroscopic task and also performed diagnostic ureteroscopy in a male cadaver with subjective global assessment done by three faculty observers. The blinded status of these observers is unknown. Virtual reality and cadaver performances correlated closely in students but not in residents. Year of training correlated more on the cadaver performance than with the virtual reality performance. Medical students were unable to perform cadaver ureteroscopy comparable to residents despite the five hours of virtual reality training. The authors concluded that UroMentor may shorten the learning curve kSimbionix, Lod, Israel.

early in training for ureteroscopic skills, but it was unable to over-ride the impact of clinical training (71).

The University of Texas Southwestern group examined UroMentor's ability to be used as an assessment tool. The authors examined the relationship between validated basic human performance testing and performance on UroMentor. Eighteen medical students underwent 13 validated tests and these skills were compared with performance of ureteroscopy on UroMentor. These performances were then used to create a model devised to predict performance on UroMentor rated by experts in a second group of 14 medical students after they completed the basic performance tests. They found that visual-spatial short-term memory capacity was the most prevalent limiting resource in the model building group and nondominant upper extremity steadiness was the most prevalent limiting resource in the model testing group. They also noted that the model was able to predict expert evaluation performance in some but not all instances (63% within ± 10% agreement with expert rating on virtual reality performance) (72).

0 0

Post a comment