Authors :
Presenting Author: Jing Xiang, MD, PhD – Cincinnati Children's Hospital Medical Center
Hisako Fujiwara, PhD – Cincinnati Children's Hospital Medical Center
Kelly Kremer, MD – Cincinnati Children's Hospital Medical Center
Hansel Greiner, MD – University of Cincinnati College of Medicine; Cincinnati Children's Hospital Medical Center
Francesco Mangano, DO – University of Cincinnati College of Medicine; Cincinnati Children's Hospital Medical Center
Jeffrey Tenney, MD – Cincinnati Children's Hospital Medical Center
Rationale:
Epileptogenic zones (EZs) lack specific biomarkers, and magnetoencephalography (MEG) is hindered by the “ill-posed inverse” problem. High-frequency oscillations (HFOs; >80 Hz) have emerged as promising epilepsy biomarkers, and MEG provides a noninvasive platform for their detection. While HFOs are spatially focal and low in amplitude, low-frequency epileptic discharges tend to be stronger and more diffuse. In this study, we introduce frequency-encoded source imaging (FESI), a novel method that integrates multi-frequency biomarker analysis with advanced source reconstruction to overcome both of these fundamental challenges.
Methods:
To address the challenge of reconstructing neuromagnetic sources from MEG signals (which reflect activity from ~50,000 neurons), we developed a cellular assembling technique (CAT). In this framework, each source is modeled as a grid point within a beamformer and decomposed into over six sub-sources, each occupying a small cubic volume and generating distinct signals (Fig. 1). FESI extracts frequency-specific neural signatures from MEG recordings across both low- and high-frequency bands (Fig. 2). These signatures are assembled into candidate sources, and only those comprising two or more synchronized sub-sources are retained. When spatial frequency maps overlap, higher-frequency components take precedence. The assembled sources are then used to reconstruct EZ volumes, which are quantified using kurtosis, skewness, and connectomic metrics across all frequencies. MEG data from a phantom, 11 patients with epilepsy (ages 6–18 years; 6 females, 5 males), and age‑ and gender‑matched healthy controls were used to test and validate the methods. Group analyses employed magnetic resonance imaging (MRI) templates (clinicaltrials.gov/study/NCT00600717). FESI results were compared with clinical epileptogenic zones (EZs) identified by invasive recordings.
Results:
FESI demonstrated significantly improved localization accuracy (p < 0.002) compared to conventional beamforming techniques. It revealed unique frequency-encoded activation patterns not accessible with traditional mathematic modelling. EZs identified by FESI exhibited elevated frequency signatures, kurtosis, and connectivity metrics, distinguishing them from broader irritative zones and noise sources.