OBJECTIVE: Because of the intricate anatomy of the temporal bone, we examined the feasibility and reliability of cranial surface anatomic fiducials to register computed tomographic images of the temporal bone by using a frameless image-guided system. METHODS: One-millimeter thick computed tomographic slices and the smallest possible field of view were used to register 10 dry and 10 fresh temporal bones from cadavers. The fiducials used for registration included the umbo of the tympanic membrane, emissary foramina, the asterion, various sutures, the tip of the mastoid process, and Henle's spine. RESULTS: Mean initial fiducial registration error ranged from 0.6 to 0.7 mm, and was reduced to 0.5 and 0.4 mm for the dry cranial and cadaveric studies, respectively, by eliminating or reregistering inexact fiducials. Mean target localization error ranged from 0.91 to 2.44 mm for superficial structures of the temporal bone in the dry cranial specimens and from 0.71 to 1.52 mm for deep structures such as the facial nerve, semicircular canals and ossicles in the cadaveric study. CONCLUSION: Interactive image-guided navigation in the temporal bone is possible with registration of cranial surface anatomic fiducials. It may be useful to the neurosurgeon and otologist in identifying critical anatomic structures of the temporal bone encountered during the translabyrinthine, retrolabyrinthine presigmoid, and suboccipital approaches.
OBJECTIVE: Because of the intricate anatomy of the temporal bone, we examined the feasibility and reliability of cranial surface anatomic fiducials to register computed tomographic images of the temporal bone by using a frameless image-guided system. METHODS: One-millimeter thick computed tomographic slices and the smallest possible field of view were used to register 10 dry and 10 fresh temporal bones from cadavers. The fiducials used for registration included the umbo of the tympanic membrane, emissary foramina, the asterion, various sutures, the tip of the mastoid process, and Henle's spine. RESULTS: Mean initial fiducial registration error ranged from 0.6 to 0.7 mm, and was reduced to 0.5 and 0.4 mm for the dry cranial and cadaveric studies, respectively, by eliminating or reregistering inexact fiducials. Mean target localization error ranged from 0.91 to 2.44 mm for superficial structures of the temporal bone in the dry cranial specimens and from 0.71 to 1.52 mm for deep structures such as the facial nerve, semicircular canals and ossicles in the cadaveric study. CONCLUSION: Interactive image-guided navigation in the temporal bone is possible with registration of cranial surface anatomic fiducials. It may be useful to the neurosurgeon and otologist in identifying critical anatomic structures of the temporal bone encountered during the translabyrinthine, retrolabyrinthine presigmoid, and suboccipital approaches.