Authors :
Presenting Author: Jia Ying, MS – Emory University School of Medicine
Zeyu zhou, PhD – Emory University School of Medicine
Jie Luo, PhD – School of Biomedical Engineering, Shanghai Jiao Tong University
Chuan Huang, PhD – Emory University School of Medicine
Rationale:
Mesial Temporal Lobe Epilepsy (MTLE) is a common neurological disorder. While the majority of patients present with radiographic evidence of hippocampal sclerosis (MRI-visible HS), approximately 30% of MTLE patients exhibit no visible lesions on standard MRI (MRI-negative), making diagnosis, seizure localization, and surgical planning more difficult. Improved imaging approaches are needed to better characterize network dysfunction in these challenging cases. 18F-FDG PET is a sensitive tool for detecting metabolic disturbances. Diffusion MRI enables the assessment of white matter structural connectivity and provides anatomically grounded network information. In this study, we applied diffusion MRI-derived structural masks to PET-based metabolic networks to compare connectivity alterations between MR-visible HS patients, MR-negative patients, and healthy controls.
Methods:
Data were acquired on a Siemens 3T PET/MR scanner (Biograph mMR). FDG-PET was performed ~45 minutes post-injection. Simultaneous MR acquisition included high-resolution T1-weighted MPRAGE (1 mm³ isotropic) and diffusion MRI (2 mm² in-plane, 30 directions, and b = 1000 s/mm²). Clinical data such as age, sex, and radiological diagnosis were also collected.
We evaluated metabolic networks using diffusion MRI-guided structural masks. These masks were derived from the average diffusion count connectivity matrix across all MTLE patients (Fig 1A), with connections above a given threshold retained at each level (Fig 1B). The resulting binary masks were applied to metabolic connectivity matrices computed using the Kullback-Leibler Divergence (DKL) method from FDG-PET scans (Fig 1C). Network metrics assessed included weighted global efficiency (GE) and weighted characteristic path length (CPL). Between-group comparisons were performed after regressing out the effects of age and gender.
Results:
The study included 28 MTLE patients with MRI-visible HS, 27 MRI-negative MTLE patients, and 15 healthy controls. For the weighted CPL (Figure 2A), patients with MRI-visible HS demonstrated consistent differences compared to healthy controls at multiple thresholds (highlighted regions, p < 0.05). MRI-negative MTLE patients also exhibited differences relative to healthy controls, although these were less widespread and appeared predominantly at selective threshold ranges. For weighted GE (Figure 2B), MRI-visible HS patients showed robust differences in GE relative to healthy controls across several thresholds, reflecting compromised network efficiency. In contrast, MR-negative patients exhibited fewer significant differences, suggesting subtler network-level disruptions.