Asymmetrical surface scanning registration for image-guided otologic surgery: A phantom study

Abstract

ObjectiveTo develop a registration procedure to achieve a higher degree of registration accuracy in image-guided otological surgery, paying particular attention to the registration centroid. MethodsA head phantom was used to measure the target registration error (TRE) at measurement points at various depth from the surface of the head. The surface-matching registration was performed using a commercially available surgical navigation system. We registered the phantom using only one ear of either side (right 100% - left 0%, or right 0% - left 100%) or using both ears with variable ratios (right 75% - left 25%, right 50% - left 50%, or right 25% - left 75%). ResultsThe overall TRE was the smallest when registration was performed equally on both sides. However, the TRE at 20–50 mm from the surface was the smallest when the fiducial points for the registration were collected asymmetrically at a ratio of 75:25 and weighed heavier on the operating side, and this difference was statistically significant. ConclusionThe accuracy of image-guided surgery can be improved by carefully planning the registration procedure without changing the procedure itself. Accurate image-guided surgery at the middle and inner ear was achieved using 75% of the point cloud for the operating side and 25% of that for the opposite side for the registration.