Authors :
Presenting Author: Juan Bottan, MD, MSc – Western University
Renee Ragguett, MESc – Western University; Lingkai Tang, MSc – Western University; greydon Gilmore, BSc, MSc, PhD – Western University; Matea Zuljevic, MSc – Western University; Andrea Andrade, MD – Paediatric Neurology – Western University; Michael Jurkiewicz, MD – Medical Imaging – Western University; Emma Duerden, PhD – Western University; Roy Eagleson, PhD, PEng – Western University; Sandrine de Ribaupierre, MD, MSc – Clinical Neurological Sciences – Western University; Maryam Nabavi Nouri, MD, MSc – Pediatric Neurology – Western University
Rationale:
Task-based fMRI (tb-fMRI) is the only clinically available tool for non-invasive preoperative language mapping in children with drug-resistant epilepsy (DRE). Children’s cooperation during tb-fMRI is low, motion and network immaturity reduces reliability. Resting-state fMRI (rs-fMRI) emerges as a valid alternative, but clinical application remains challenging due to discrepancies in results and lack of standardized methodology, particularly in younger children with atypical language networks.
Addition of functional Near-Infrared Spectroscopy (fNIRS), a lower-complexity technique, shows promise to overcome some of the limitations of rs-fMRI and enhance reliability. Our objective is to develop a combined approach utilizing fNIRS and rs-fMRI to reliably determine language lateralization and localization in children with DRE that are candidates for surgery.
Methods:
We are conducting a pilot feasibility study, currently including 40 children with DRE candidates for epilepsy surgery (Table 1) with tb-fMRI (2 tasks, verb generation and object naming in block design) and rs-fMRI (short video projection) and fNIRS with the same task protocol. Additionally, a structural T1w with the fNIRS cap and vitamin E markers was performed in a subgroup of 12 patients. For resting-state, an independent component analysis (ICA) approach with a template-matching procedure was used to identify language networks. For task-based analysis, a traditional general linear model approach was used. Based on the block averages obtained, contrast-to-noise ratio was calculated as a measurement of activation level for selected channels on HbO, Hbr and Hbt signals, respectively. Laterality index was obtained. A 3D-morphometric modelling pipeline of the vitamin E markers was performed to co-register fNIRS to structural and functional scans. An fNIRS-guided seed-based analysis will be carried out and compared to both, tb-fMRI and ICA rs-fMRI. Dice coefficient between the language networks derived from the different methods will be used to determine degree of overlap and concordance.
Results:
An expected high incidence of atypical language lateralization was observed in tb-fMRI and rs-fMRI with acceptable concordance. ICA rs-fMRI tends to retrieve more extensive language networks than tb-fMRI. Preliminary results on 7 children in the pilot group suggest lateralization with fNIRS is overall concordant with tb-fMRI and well tolerated by children (Table 2). Discrepancies in task-specific lateralization were observed in some participants. We expect that the fNIRS-guided SCA results will be comparable to those of other fMRI techniques for language lateralization.
Conclusions:
The high incidence of atypical language development in children with DRE remains a challenge for non-invasive neuroimaging analysis. The combined approach using fNIRS and fMRI offers a novel alternative to localize language networks in children with DRE with potential to increase accessibility and reliability of non-invasive techniques. Larger cohorts and invasive confirmation is needed to further validate these method.
Funding: Self-funded