Operating Room Results

We have used the described image-guided neurosurgery system on 70 patients. These cases included: 44 Supratentorial - 22 high grade, 22 low grade; 6 Meningiomas; 3 Metastases; 2 Posterior Fossa; 1 Meningioangiomatosis; 7 Intractable Epilepsy; 4 Vascular; 2 Biopsies; and 1 Demyelinating lesion.

In all cases the system effectively supported the surgery:

• by providing guidance in planning bone cap removal - this was done through the augmented reality visualization in which the surgeon could visualize paths to the critical tissue and plan an appropriate entry point.

• identifying margins of tumor - this was done by tracing the boundaries of tissue with the trackable probe.

• localizing key blood vessels.

• and orienting the surgeon's frame of reference.

Selected examples are shown in Figure 5.

  

Figure: Examples of neurosurgical cases. The last example includes a fusion of fMR data overlaid on top of the structural model.

To qualitatively validate the system's performance, the surgeon placed the pointer on several known landmarks: skull marks from previous surgeries, ventricle tip, inner skull bones such as eye orbits, sagittal sinus, and small cysts or necrotic tissues. He then estimated their position in the MRI scan, and we compared the distance between the expected position and the system's tracked position. In all cases, this error was less than two voxels (MRI resolution was $0.9375mm \times 0.9375mm \times 1.5mm$).

One example of the effectiveness of the system is illustrated by the following study. Twenty patients with low-grade gliomas underwent surgery with the system. The pathologies included 10 low grade astrocytomas (grades I, II out of IV), 7 oligoastrocytomas (without anaplastic features) and 3 oligodendrogliomas. Thirteen patients underwent cortical mapping including 7 who underwent speech and motor mapping, 2 motor alone, 1 speech alone and 3 motor and sensory. 31% had a subtotal resection, the remainder had total resection. One patient exhibited temporary left-sided weakness. Cortical mapping had represented the sensory cortex diffusely behind this patient's gross tumor. The post-operative weakness was temporary and was thought to be due to swelling. One patient showed a mild, left upper extremity proprioreceptive deficit which was due to a vascular accident on post-operative day one. The remaining patients were neurological intact following the procedure.

In addition to the tumor resection cases, we have also used the system in 10 pediatric epilepsy cases [3]. In the first stage of this two stage surgery, the patient's cortex is exposed and a grid of electrical pickups is placed on the cortical surface. A lead from each pickup is threaded out through the skin for future monitoring. In addition to registering the MRI model of the patient to his/her position, the location of each electrical contact is recorded and transformed to MRI coordinates. The patient is then closed up and monitored for several days. During any seizure event, the activity from each cortical probe is monitored, and transformed to the MRI model. This enables the surgeon to isolate potential foci in MRI coordinates. During a second surgical procedure, the augmented MRI model is reregistered to the patient and the locations of the hypothesized foci are presented to the surgeon for navigational guidance. An example of this is shown in Figure 6.

  

Figure: Grid of electrodes placed on cortical surface. Location of grid points overlaid onto MR reconstruction, with focal area highlighted.

To see the range of cases handled by our system, we encourage readers to visit the web site:

   http://splweb.bwh.harvard.edu:8000/pages/comonth.html

which shows selected cases with descriptions of the use and impact of the navigation system on the case.