Neuropediatrics 2011; 42 - P044
DOI: 10.1055/s-0031-1274016

Comparison of different tractography algorithms and validation by intraoperative stimulation in a child with a brain tumor

S Gröschel 1, V Kumar 2, MU Schuhmann 3, M Alber 4, I Krägeloh-Mann 4, M Wilke 1
  • 1Universitätskinderklinik, Experimentelle Pädiatrische Neurobildgebung und Abteilung für Neuropädiatrie, Tübingen, Germany
  • 2Universitätsklinik, Sektion Experimentelle Kernspinresonanz des ZNS, Abteilung für Diagnostische und Interventionelle Neuroradiologie, Tübingen, Germany
  • 3Universitätsklinik, Pädiatrische Neurochirurgie, Tübingen, Germany
  • 4Universitätskinderklinik, Abteilung für Neuropädiatrie, Tübingen, Germany

Aims: Advanced MRI modalities like functional MRI (fMRI) and MR diffusion tractography, implemented in modern neuronavigation software, are increasingly used for neurosurgery planning in children. However, mainly only a basic deterministic diffusion tensor (dDTI) tractography algorithm is used in these software packages.

Aim: Evaluation of different tractography algorithms and comparison with intraoperative subcortical electrical stimulation results in a child with a brain tumor.

Method: Pre-surgical fMRI and MR diffusion tractography was done in a 6 year-old child with seizures, but no motor symptoms, due to a neuroectodermal tumor in the left central region.

MR diffusion acquisition was done using a b-value of 3000s/mm2 and 60 diffusion directions. Four different tractography algorithms were used: dDTI, probabilistic DTI-tracking (pDTI, FSL's Markov chain Monte carlo), deterministic and probabilistic CSD-tracking (dCSD/pCSD; constrained spherical deconvolution, Tournier et al. 2007).

Results: All four tractography algorithms were able to localize the cortico-spinal tract anterior to the tumor with good agreement. More fibres could be visualized using the pDTI- and the d/pCSD-tracking algorithms than using dDTI. A proportion of the fibres traversed a solid part of the tumor, close to the internal capsule, which was best detected using pCSD and was not seen using the dDTI method. Intraoperative stimulation confirmed the visualized fibres, in particular those traversing the tumor. Consequently, only a subtotal tumor resection was performed, leaving the patient was with no sensorimotor deficit postoperatively. The spared region containing cortico-spinal tract fibres were identified in the postsurgical MRI 4 months later.

Conclusion: Although all tractography algorithms successfully identified the cortico-spinal pathway, deterministic DTI-tractography, as widely used in clinical neuronavigation software, only insufficiently visualized critical fibres in this case. Further research is necessary to validate different tractography algorithms systematically.