Registration and fusion of CT and MRI of the temporal bone

Abstract

Objective: To present and evaluate a registration method to fuse complementary information of CT and MRI of the temporal bone. Methods: CT and MRI of the temporal bone of 26 patients were independently registered 4 times. A manual, iterative, intrinsic, rigid, and retrospective registration method was used. Mean CREm (consistency registration error) was calculated as a reproducibility measurement. Results: CREm was 0.6 mm (95% CI = 0.52-0.68 mm). T-test revealed no difference between pathologic and normal cases (t[102] = -1.71; P = 0.09). Time needed: 13 minutes. In the registered and fused datasets, important bony surgical landmarks (eg, facial nerve canal, inner ear) could be assessed in 3 dimensions relatively to tumor tissue (eg, acoustic schwannoma). Fluid distribution within partially obliterated cochleae could be assigned to either scalae. Conclusion: An accurate, reproducible registration and fusion method that improves tumor surgery and cochlea implantation planning with only minor changes to the clinical workflow was presented and described. We suggest this method in selected cases. Copyright © 2005 by Lippincott Williams & Wilkins.