Article
Compensation of brain shift by registration of pre- and intraoperative image data
Ausgleich des Brain Shift durch die Registrierung prä- und intraoperativer Bilddaten
Search Medline for
Authors
Published: | April 11, 2007 |
---|
Outline
Text
Objective: To compensate for the effects of brain shift in neuronavigation intraoperative image data had to be registered to the patient in a time consuming process e.g. necessitating bone fiducials that had to be identified in the intraoperative images. The aim of this study was to establish an update method without the need for a repeated cumbersome intraoperative patient registration.
Methods: Intraoperative high-field (1.5T) MRI was applied in combination with microscope-based navigation. Intraoperative image data were rigidly registered to preoperative 3-D data by a pyramidal approach based on mutual information (iPlan2.5, Brainlab, Feldkirchen, Germany). After registering the pre- and intraoperative datasets, the initial patient registration data were applied for the intraoperative images (i.e. the navigation coordinate system). Anatomical landmarks were used to verify the intraoperative accuracy of the update procedure. In addition 6 landmarks were used in an offline analysis for a reference fusion independent of brain shifting to validate the image registration process.
Results: In a series of 818 patients investigated by intraoperative high-field MRI navigation was applied in 491. In 87 of these cases, intraoperative image data were used for an update of the navigation system resulting in further tumor removal. In the gliomas of this subgroup (n=69), the extended resection resulted in a significant (P<0.01) reduction of the remaining tumor volume where no complete resection was intended b ecause of an infiltration of eloquent areas (48 of 69), while in 21 a complete resection could be achieved. The registration error of pre- and intraoperative images ranged from 0.6 to 2.0 mm (mean: 1.4 mm). Updating the navigation system with intraoperative image data added only about 5 minutes to the whole procedure.
Conclusions: Rigid registration of pre- and intraoperative image data allows immediate updating of the navigation without the need for repeated time-consuming patient re-registration. Tumor remnants can be localized reliably and the effects of brain shift are compensated for.