Introduction
Since the early 1990s, there has been a renewal of interest in functional neurosurgical techniques in Parkinson’s disease and other movement disorders. A better understanding of the neuroanatomical and neurophysiological basis of these disorders has fueled this renewal. Compared to the use of X-rays and ventriculograms, the advent of computed tomographic (CT)-guided stereotaxy was a godsend. With increasing need for accurate placement of instruments in smaller nuclei, magnetic resonance imaging (MRI)-guided techniques were used. The major disadvantage of using MRI alone was evident in its inherent magnetic image distortion (Walton, 1996). Magnetic resonance imaging provides a better target image, but when used alone it can lead to misplacement of an electrode or needle.
The Radionics workstation incorporates CT/MRI image fusion, Stereo-plan target planning, anatomical atlas, and the NeuroMap micro/semimi-croelectrode recording system. This system provides the surgeon with a safe, accurate, easy-to-operate system for functional techniques such as lesioning, biopsy, drug delivery, or placement of deep brain stimulators. Using early versions of Image Fusion and Stereoplan, the Oxford Movement Disorder Group demonstrated the system’s accuracy to 1 mm (Aziz, 1998). Today, with later versions of the software and the use of the NeuroMap microelec-trode recording systems, the software has become more user friendly and the frontiers of precision have been pushed a step further.
Image Fusion
In our clinical program at Neurological Associates Center for Movement Disorders, all functional neurosurgical procedures are based on image fusion using CT and MRI. The patient starts the day with an MRI using the Siemens 1.5 Tesla MP Range (3-mm axial cuts) program that provided good visualization of nuclear groups in the basal ganglia and brainstem (Fig. 1). The patient is placed under local anesthesia and the Radionics CT frame and localizer are placed with the use of temporary ear bars. The frame is placed with rough approximation to the AC-PC line from external auditory meatus to the inferior orbital rim. A noncontrast CT scan is then done (GE scanner, 3-mm cuts). The MRI and CT data are stored on optical disc and transferred to the Radionics workstation. Unlike previous software versions that fused the MRI image to the CT bone density, the newer 2.0 version rapidly fuses the two images pixel to pixel. Previous versions of Image Fusion required that the MRI and CT images be nearly coplanar. The 2.0 fuses the images rapidly, even if the scan images are not matched.
Target Planning
In the operating room, data from Image Fusion are transferred to Radionics Stereoplan program and the CT fiducials are rapidly localized. The axial fused MRI image is then brought to the screen and target planning begins. For reference during target planning, the AC-PC line is used to fuse the images with anatomical brain slices from the Schaltenbrand brain atlas (Schaltenbrand, 1977). Targets in the basal ganglia are calculated roughly using Guiot’s diagram (Guiot, 1968) and as modified by Benabid (Benabid, 1998). These targets are further refined directly on the MRI axial, coronal, and sagital images in the Stereoplan program. Once anterior-posterior, lateral, and vertical coordinates are determined, the ring and slide numbers are entered into the system to provide the safest trajectory to the target, avoiding sulci, fissures, and the ventricular system (Fig. 2). This technique is especially useful when working with a patient with some degree of atrophy and large ventricles and sulci. Once the surgeon is satisfied with the target and trajectory, the coordinates are transferred to the Radionics Cossman, Roberts, Wells (CRW) stereotactic frame.
Figure 1 Split images of magnetic resonance imaging and computed tomography in Image Fusion. Good overlap of anatomical structures is seen between the two studies. Patient has had a previous left subtha-lamic nucleus stimulator.
Neurophysiological Localization
Before placement of chronic stimulating electrodes or lesioning a target, neurophysiological localization is performed. At our center, we have shared in the development of the Radionics NeuroMap neurophysiology station.
Figure 2 Fused magnetic resonance image on the Stereoplan workstation showing trajectory for placement of subthalamic nucleus stimulating electrode.
This unit is a compact computer using a Windows NT workstation. The workstation contains a series of amplifiers and filters that allow monitoring and refinement of up to eight channels of electrophysiological data from the patient. Data are displayed in real time and can be printed or archived for later analysis. For a typical patient undergoing a basal ganglia procedure, we use one channel only for semimicroelectrode recording and mapping of the region of interest. A second channel can be used for simultaneous elec-tromyographic (EMG) data in a patient with tremor or dystonia. The electrode is mounted on the Radionics microdrive and connected to a pre-amp mounted on the CRW frame. Recording from the brain is typically started 12 mm above the target. Either microelectrodes or semimicroelectrodes can be used in this system.
For intraoperative recording, we are currently investigating the use of semimicroelectrodes. The 40-micron tip tungsten semimicroelectrode, (1 megaOhm ± 10% at 1 kHz) provides robust signals that accurately declare the margins of the target nuclei. For example, characteristic firing patterns of thalamic, subthalamic, pallidal, or nigral neurons can be identified and recorded (Fig. 3). In some procedures, such as thalamic mapping, kinesthetic response of neuronal activity can also be observed.
Figure 3 a. Semimicroelectrode recording from the subthalamic nucleus showing ”pauser cells.” b. Semimicroelectrode recording from the substantia nigra reticulata (SNr) showing fast activity.
Figure 4 Real-time fast Fourier transformation spectral analysis of neuronal activity in the subthalamic nucleus. Predominant activity is noted at 4 Hz and 75 Hz with harmonics in the far spectrum.
This technique has the advantage of allowing more precise mapping of the target’s nuclear perimeter. With these data, one can then tailor the placement of the chronic stimulating electrode. The greatest limitation of semimicroelectrode recording is that one typically records from more than one neuron at a time. Semimicroelectrode recording provides a strong and reliable neuronal signal but makes determination of cell firing frequency difficult. To quantify frequencies of neuronal groups, we are currently in- vestigating the use of real-time fast Fourier transformation (FFT) analysis of semimicroelectrode signals (Fig. 4).
Figure 5 A. Patient with bilateral subthalamic nucleus stimulators (axial image).
Once the nucleus of interest is mapped, the Medtronic Activa stimulating electrode can be placed and tested in the awake patient. The Radionics radiofrequency generator is used if a lesion such as a pallidotomy or thal-amotomy is planned.
Macrostimulation, impedance calculations, and physical responses are evaluated before placement of a permanent lesion. To evaluate the placement of electrodes or lesions, all patients undergo postoperative magnetic resonance imaging (MRI) (Fig. 5).
Figure 5 B. Patient with a Gpi stimulator and a pallidotomy (coronal image).
DISCUSSION
Safety is the most important part of any neurosurgical procedure. Functional stereotactic techniques demand a high level of forethought, caution, and preparation. A well-planned, successful procedure can modify the function of the brain to the patient’s advantage. The surgeon should always work toward making the procedure as precise and noninvasive to the patient as is possible.
Second only to patient selection, the most critical part of stereotactic surgery is initial acquisition of targets and accurate placement of electrodes.
Technical advances in the field, such as image fusion, neuronavigation, and neurophysiological analysis provide better clinical outcome. It is our opinion that no stereotactic functional procedure should be based on MRI alone. The inherent distortion in the MR image can lead the surgeon astray. As a fisherman trying to snare a fish in the stream knows the prey is not exactly where it appears to be. To be successful, the thoughtful neurosurgeon must learn how to deal with distortion. Fusion of MRI on a CT image corrects this distortion and allows precise targeting.
Further confirmation of target is obtained through micro- or semimi-croelectrode recording. The use of microelectrode recording is often debated, but we feel it adds another degree of accuracy and allows mapping the perimeters of the nucleus. Better placement of the chronic electrode is obtained by this technique. In our center, we are finding that semimicroelec-trodes are dependable and provide adequate confirmation of target. Further advances with this technique, such as fast Fourier analysis, are promising and should provide more information about the nucleus of interest.
One shortcoming of the Radionics system is the fusion of the patient’s anatomy with the Scaltenbrand atlas. Because every patient’s anatomy is different, the atlas provides only a rough approximation and should never be used to plan target coordinates. Using the patient’s own brain anatomy provided by MRI/CT fusion is reliable and accurate.
Some clear advantages of this workstation are accuracy, dependability, and expediency. Image Fusion and Stereoplan allow "first pass” target acquisition instead of multiple passes in an attempt to make up for MRI distortion.
Single-cell recording is important for research investigations to define functionality and cell firing frequency, but is a technical challenge because of the nature of the tiny electrode tip. Semimicroelectrodes are easy to use and give the surgeon a better map of the target nuclear boundaries. In addition, the developing Fourier analysis technique may help define the signature of the target nucleus and its predominant cell-firing frequencies.
