J Neurol Surg B Skull Base 2016; 77 - A128
DOI: 10.1055/s-0036-1579915

The Role of Diffusion Tensor Imaging in Skull Base Surgery

Amin Kassam 1, Melanie B. Fukui 1, Martin J. Corsten 1, Richard A. Rovin 1, John Jennings 1
  • 1Aurora Neuroscience Innovation Institute, Milwaukee, Wisconsin, United States

Introduction: There has been increasing interest in and clinical utilization of diffusion tensor imaging (DTI) in Neurosurgery to define white matter tracts and cranial nerves to reduce surgical morbidity. Barriers to the use of DTI in skull base surgery include: 1. geometric inaccuracy with “co-registration” of DTI onto anatomic T1 imaging because of errors in warp correction, 2. inability to use multimodality imaging, 3. lack of 3D rendering that can be manipulated in realtime for surgical trajectory planning, and 4. lengthy processing times.

Methods: We report our experience with 50 cases of varied skull base procedures with a recently FDA approved 3D DTI planning system Bright Matter Plan (Synaptive Medical, Toronto, Canada). Tractography data are generated by first performing brain extraction (via template initialization and level set evolution) and skin surface detection (via image statistics). Next the raw diffusion dataset is then corrected for motion and eddy current artifacts. Then this diffusion data are co-registered to an anatomical target (e.g., T1 or T2 contrast volume) using a mutual information metric to find a rigid co-registration matrix. Now the raw diffusion data are used to estimate diffusion tensors (via a dual basis solver) in the anatomical space.

Noise in the tensor estimate outside the cranium is attenuated using the surface mask, while the brain mask is used to define a tractography region. Whole brain seeding is then used to generate tracts using a deterministic 4th order Runge-Kutta interpolator. Finally, a cleanup pass is performed to remove noise in the tract data as well as perform geometric simplification of the tract objects.

Results: In all cases we were able to obtain accurate 3D geometric fits with multimodality rendering. Processing times of 15min from image acquisition in a automated process improved the ergonomics and minimized the resources needed for processing. The rendering platform accurately merged multiple imaging modalities simultaneously and was transferred to the navigation and robotic optical positioning systems. As a representative example the case of a large olfactory groove meningioma is shown. The perimeter location of the bilateral inferior frontal occipital fibers and the superior longitudinal fasciculus was helpful in selecting an endonasal corridor for this lesion as opposed to a transcranial route.

Conclusions: As skull base surgery evolves and recognizes neurocognitive morbidity, the identification of critical fibers that subserve these eloquent processes need to be considered during preoperative planning. BrightMatter Plan addresses the aforementioned barriers:

• The ability to plan in multiple orthogonal axes simultaneously and simulate the procedure taking into account the critical position of cranial nerves and fiber tracts was found most useful in large anterior and posterolateral skull base pathologies.

• The emergence of an integrated DTI planning system that is quick (15 minute processing time), geometrically accurate, and includes real time 3D rendering that can be fused with multiple imaging modalities represents a valuable surgical tool.

• Increasing experience will be needed to correlate postoperative tract preservation with outcome to understand the optimal corridors for skull base surgery based on previously “invisible” anatomy.

Fig. 1