Assessment of video tracking usability for training simulators.

Objective: A simulator was developed to mimic commercial CAS systems in implementing most tasks required to carry out a surgical operation. As tracking systems are generally expensive components, an alternative solution based on low-cost video-based tracking was used. Video tracking accuracy was assessed to determine whether or not this kind of approach was suitable for use in the training domain. Ultimately, video-based tracking should enable sufficiently accurate registration between a bony model and its virtual 3D representation.

Materials And Methods: Video tracking was assessed using two types of camera. For each one, common accuracy tests were realized as a series of 10 trials at ranges of 0.5-1.0 m from the camera lens. The pointer used as a digitizer was equipped with tracked video markers. Three sizes of marker were evaluated to estimate the impact of marker size on accuracy.

Results: For the better of the two cameras tested, results were encouraging. Results are presented as rounded whole-number values in millimeters. The noise test gave accuracies of 2 mm for the 80-mm marker, 3 mm for the 60-mm marker and 5 mm for the 40-mm marker. Relative accuracies, as evaluated on a grid of equally spaced dots, were 4 mm with the 80-mm marker, 7 mm with the 60-mm marker and 12 mm with the 40-mm marker. A pivoting test around the pointer tip gave 3 mm of accuracy for the 80-mm marker, 5 mm for the 60-mm marker and 11 mm for the 40-mm marker. An additional pivoting test was completed on increasing the distance of the marker from the pointer tip, giving accuracies of 5 mm for the 80-mm marker, 6 mm for the 60-mm marker and 13 mm for the 40-mm marker. The registration test gave accuracies of 8 mm for the 80-mm marker, 9 mm for the 60-mm marker and 11 mm for the 40-mm marker.

Conclusions: The video-based approach offers sufficient accuracy to achieve registration in the domain of CAS training.