Authors :
Presenting Author: Maksim Parfyonov, MD – Pediatric Epilepsy Section, Epilepsy Center at the Cleveland Clinic Neurological Institute, Cleveland, OH, USA
Spencer Morris, MS – Cleveland Clinic
Ting-Yu Su, PhD – Cleveland Clinic
Joon Choi, PhD – Yonsei University
Ingmar Blümcke, MD – University Hospitals Erlangen
Ken Sakaie, PhD – Cleveland Clinic
Andreas Alexopoulos, MD, MPH – Cleveland Clinic
Imad Najm, MD – Cleveland Clinic
Dan ma, PhD – Duke University
Stephen Jones, MD – Cleveland Clinic
Zhong Irene Wang, PhD – Cleveland Clinic
Rationale:
Magnetic resonance fingerprinting (MRF) is a novel MRI method that generates multiple tissue property maps from a single scan by pseudorandomly varying acquisition parameters, creating tissue-specific signal evolutions, and matching these to a pre-generated dictionary1. Compared to conventional MRI, MRF is robust, quantitative, repeatable across scanners and patients, and less susceptible to external factors. These advantages make MRF well-suited for identifying tissue-level biomarkers of epileptogenic pathology and linking these to electrophysiology. In the present study, we aimed to assess differences in MRF signal across brain regions with varying degrees of involvement in the epileptogenic network as defined by intracranial recordings.
Methods: We conducted a pilot study of 12 patients who underwent SEEG at the Cleveland Clinic between 2013–2024 and achieved seizure freedom after surgery (resective: N=10; LITT: N=2). Whole-brain T1 and T2 maps were generated from 3D MRF scans on a Siemens 3T Prisma scanner, with matching to a dictionary as described previously2. Separately, MRF data from 52 healthy controls was used to compute normative T1 and T2 values for GM and WM within each anatomical label in their VEP parcellations3 (Figure 1). To account for partial volume effects, only voxels with probability of >0.5 for GM or WM were used and voxels with probability of CSF >0.1 were excluded. Individual patient voxel-wise Z-score maps were derived by comparing observed T1/T2 values to the normative reference. Z-score values were extracted at electrode contact coordinates within a 2 mm radius. Each SEEG contact was classified into one of four zones based on clinical interpretation: seizure-onset zone (SOZ), propagation zone, irritative zone, or non-involved zone. Statistical analyses were performed in R.
Results: Twelve patients were included (5 male, 7 female; 9 right-handed, 3 left-handed; mean age 26 years, range 13–44). The mean number of electrodes was 17 (range 14–23). Implantation was unilateral in 9 patients (6 right hemisphere, 3 left) and bilateral in 3. We highlight one example patient: a 43 year old patient with MOGHE and non-lesional MRI. Compared to non-involved regions, contacts in the seizure-onset zone show elevated T1 gray matter Z-scores (Figure 2). There was a trend for progressively increased Z-scores from non-involved to irritative, propagation, and SOZ contacts. This case demonstrates the feasibility of our pipeline and suggests that GM Z-scores may reflect epileptogenicity. Additional patient analyses and group comparisons are ongoing.
Conclusions: This pilot study highlights the potential of MRF to quantify tissue-level microstructural differences within the epileptogenic network. To validate its clinical utility, larger studies are needed to rigorously evaluate the correlation between MRF data, invasive electrophysiology, and histopathological findings.
Funding: NIH 2R01 NS109439