Evaluation of the factors affecting the optimal fiducial configurations calculated through a genetic-algorithm-based methodology in image-guided neurosurgery

Background Image-guided neurosurgery usually involves a point-pair registration between two spaces, associating the patient in the operating room with pre-operative image scans. The distribution and number of fiducial markers during registration are critical for the expected error at the target point. Methods A genetic algorithm has been designed to provide an optimal marker configuration. The solution, visualized on a 3D head reconstruction, is intended as a guideline for the surgeon to properly place the markers. However, deviations from ideal configurations occur during marker placement; moreover, the actual target is not a point, but a region. The consequent decrease of target accuracy is statistically investigated. Results A requirement on target minimum accuracy can be satisfied in the operating room not only by setting the number of markers for the optimized fiducial configuration, but also by considering the inevitable sources of error. Conclusions Quantifying the sources of error that affect a genetic-algorithmbased optimization shows that it is still convenient. Target point correct individuation is particularly important, as it strongly influences the optimization performance.