Authors :
Honghao Han, PhD – University of Electronic Science and Technology of China
Xiyue Sun, M.E – University of Electronic Science and Technology of China
Zhi Fang, M.E – University of Electronic Science and Technology of China
Ting Zou, PhD – University of Electronic Science and Technology of China
Huafu Chen, PhD – University of Electronic Science and Technology of China
Qiangqiang Liu, MD – Ruijin Hospital, Shanghai Jiao Tong University School of Medicine
Presenting Author: Rong Li, PhD – University of Electronic Science and Technology of China
Rationale:
Aberrant dynamic transitions between intrinsic brain states are thought to underlie the pathophysiology of epileptic seizures, particularly in focal drug-resistant epilepsy (DRE). Stereo-electroencephalography-guided radiofrequency thermocoagulation (SEEG-guided RF-TC) has emerged as a minimally invasive alternative to traditional resection surgery, demonstrating promising efficacy in reducing seizure frequency through targeted thermocoagulative lesions. However, the mechanisms by which RF-TC modulates brain network dynamics remain poorly understood. Elucidation of these mechanisms may shed light on how RF-TC achieves seizure control by modulating pathological brain dynamics, thereby informing more precise and personalized therapeutic strategies.
Methods:
The study comprised 43 patients diagnosed with DRE, including 22 cases with temporal lobe epilepsy, 15 with frontal lobe epilepsy, and 6 with occipital lobe epilepsy. These patients were enrolled at two time points: namely preoperatively and a minimum of six months following RF-TC, alongside 22 healthy controls. All participants underwent functional MRI scans. The therapeutic response was defined as a >50% reduction in seizure frequency sustained for at least six months following RF-TC. Using lesion network mapping (LNM), a network was identified that exhibited significant functional connectivity with to all patients’ thermocoagulative targets. This network is henceforth referred to as the RF-TC network. Subsequently, a data-driven modelling approach, the Hidden Markov Model (HMM), was applied to identify recurring dynamic activation states within the RF-TC network based on blood-oxygen-level-dependent signals. The fractional coverage of these states was analyzed across groups and time points and further examination was conducted into how changes in state coverage were related to the functional disconnection between surgical targets and the RF-TC network.
Results:
The RF-TC network was primarily distributed across the temporo-limbic circuit, fronto-parietal networks, cerebellum, and brainstem. HMM identified four distinct dynamic states that reflected epileptic propagation patterns within this network. Fractional coverage analysis revealed significant normalization in responders following RF-TC (P < 0.05), with values recovering to those of healthy controls, whereas non-responders showed no meaningful change (P > 0.05). Moreover, changes in fractional coverage were significantly associated with the estimated functional disconnection between patients’ surgical targets and the RF-TC network (P < 0.05).