Technical and functional validations of High Angular Resolution Diffusion Imaging (HARDI)-MR tractography in paediatric epilepsy patients

• ¹ Murdoch Childrens Research Institute, Developmental Imaging, Australia
• ² The Royal Children’s Hospital, Department of Neurosurgery, Australia
• ³ The Royal Children’s Hospital, Department of Neurology, Australia
• ⁴ The University of Melbourne, Department of Paediatrics, Australia
• ⁵ Murdoch Childrens Research Institute, Child Neuropsychology, Australia
• ⁶ The University of Melbourne, School of Psychological Sciences, Australia

Background: HARDI-MR tractography is a neuroimaging technique that allows modelling of white matter tracts in the brain. The ability to visualise white matter tracts pre-operatively has the potential to improve neurosurgery accuracy and outcome. However despite its growing popularity, questions remain regarding its technical reproducibility and validity in routine clinical use. Here we examine the reliability of a locally practised HARDI-MR tractography method in paediatric epilepsy patients and in a novel demonstration present the functional validation of the technique against multimodal clinical investigations.

Methods: Eight patients were identified retrospectively from our Children’s Epilepsy Program database. All cases had pre-operative diffusion MRI sequences and motor and/or language functional MRI (fMRI). In select cases, intra- or extra-operative motor and/or language cortical mapping results were available. Corticospinal (CST) and superior longitudinal fasciculus (SLF) tracts were generated by two different operators. Coefficients of variation (CV), intra-class correlation coefficients (ICC) and dice similarity coefficients (DSC) were calculated to assess reproducibility of tract metrics and degree of tract overlap between the operators. Tractography results were compared to fMRI activation, cortical mapping results and the patient’s clinical assessment findings.

Results: All white matter tracts generated were visually comparable between operators. The tract measures showed low CV values, high ICC values and high DSC scores, indicating there was a high degree of reproducibility and overlap within and between the two operators. Cortical origin and terminations of all CST and SLF tracts matched with the patient’s areas of motor or language fMRI activation. In all but one case, tractography results matched cortical stimulation findings. Tracts were also compatible with the clinical assessment findings.

Discussion: In a group of paediatric epilepsy patients, tractography results could be reliably generated and reproduced. In a novel demonstration, there was correspondence between the anatomical course of the tracts with areas of fMRI activation and, in all but one case, cortical stimulation findings. This suggests that in experienced hands, HARDI-MR tractography can be a reliable and valid technique in clinical settings. HARDI-MR tractography has the potential to make valuable contributions to neurosurgery planning to improve outcome for patients.