MRI-directed subthalamic nucleus surgery for Parkinson's disease

The subthalamic nucleus (STN) is now regarded as the optimal surgical target for the treatment of medically refractory idiopathic Parkinson's disease. In our center, a predominantly MRI-directed method has been developed for targeting the STN. The STN is localized on T2-weighted images from a 1.5-T MRI scanner. Long acquisition, high-resolution images are acquired in both the axial and coronal planes under strict stereotactic conditions with the patient under general anesthesia. The boundary of STN is co-registered in both planes to give optimal 3-dimensional target definition. Stereotactic coordinates of the dorsolateral STN are recorded and the trajectory is planned down the axis of the nucleus in the coronal plane. Initially, per-operative macrostimulation was used for adjustment at the target prior to unilateral subthalamotomy in 26 patients. Five patients were lost to follow-up. Assessments of the lesions in post-operative images confirmed successful localisation of the lesions within the dorsolateral STN in all of the remaining 21 cases. In a subsequent series of 19 patients treated by deep brain stimulation (DBS), unilateral in 1 patient and bilateral in 18, the STN was targeted using the same MRI-directed method, guide tubes and radio-opaque stylettes were implanted, and target verification was entirely MRI-based. Following implantation of the guide tubes and stylettes, assessments of the per-operative MRI images for the 37 STN targetings confirmed a mean target error, between the stylette and the desired target in the axial plane, of 0.3 mm mediolaterally (SD = 0.4) and 0.4 mm anteroposteriorly (SD = 0.4), with median errors of 0.5 mm. This study demonstrates that MRI-directed targeting of the STN through guide tubes is accurate, and allows direct verification and corrections as necessary. Cumulative frequencies predict that the majority of DBS electrodes placed in this manner will be within 0.5 mm of the planned target. Because physiological methods are not required, the whole procedure can be performed under general anesthesia. We feel that planning with reference to a standard atlas is unreliable and not significantly helped by the addition of microelectrode recording, the accuracy of which in the axial plane is dependent upon the distance between the recording trajectories, which is typically 2 mm.