Anatomical landmarks for image registration in frameless stereotactic neuronavigation

Frameless stereotactic navigation devices require preoperative application of skin markers (SM) and planning radiography, which limits their even wider use. Therefore, we prospectively studied the applicability and accuracy of anatomic "natural" markers (NM) for image registration.

The accuracy of NM was evaluated in 26 patients operated on in the supine (n=24) or sitting (n=2) position, either by comparison to our standard navigation protocol using SM and planning radiography or by the deviation of anatomic landmarks using a routine diagnostic radiograph. In 21 cases, NM were compared to SM with planning radiography (computed tomography, or CT, in nine cases and magnetic resonance imaging, or MRI, in 12). The root mean square error (RMSE) of the registered volume was calculated by the Philips EasyGuide Neuro frameless stereotactic navigation system and compared between the two registration modalities.

The mean RMSE was 3.2 mm+/-1.0 mm standard deviation using NM vs 2.9+/-1.0 mm using self-adhesive SM (P=0.13, Student's t-test). Computed tomography was slightly more accurate than MRI planning (mean RMSE 3.2 mm vs 3.3 mm). In three cases, diagnostic radiography (MRI) was used with a mean RMSE of 5.3 mm but acceptable intraoperative landmark correlation.

Our pilot study demonstrates insignificant loss of registration accuracy using NM compared to SM. Additionally, the radiologic planning investigation and accuracy loss due to SM movement may be avoided.