Physiology-guided model improves analysis of intracranial EEG signals
PHASE: A Physiology-Guided Hierarchical Foundation Model for Intracranial EEG
Machine LearningArtificial Intelligence
Summary
Analyzing brain signals recorded directly from inside the head is important for understanding conditions like epilepsy. Existing AI models try to learn these signals by guessing missing parts, but they don’t explicitly focus on important physiological features doctors use. The authors built a new model, PHASE, that learns these key physiological traits directly and can better identify seizures and other brain functions. This model works well even on data from hospitals it wasn’t trained on, showing its potential for broad clinical use.
What this means in practice
- •For hospital data teams: Use PHASE to improve detection of seizure onset zones and other clinical EEG tasks across diverse hospital data.
- •For medical device engineers: Incorporate PHASE’s physiology-guided model to enhance intracranial EEG monitoring devices with better seizure and brain function decoding.$Commercial implications: Enables advanced EEG analysis features in medical devices for epilepsy diagnosis and brain monitoring, improving clinical decision tools sold to hospitals.
Authors
Yipeng Zhang, Chenda Duan, Yuanyi Ding, Tianyi Wang, Atsuro Daida, Masaki Izumi, Yuta Tanoue, Naoto Kuroda, Shaun A. Hussain, Nishant Sinha, Eishi Asano. Hiroki Nariai, Vwani Roychowdhury
Abstract
Clinicians and neuroscientists have long analyzed intracranial electroencephalography (iEEG) through directly measurable physiological characteristics, which carry much of the information that downstream tasks depend on. Recent iEEG foundation models learn by reconstructing or predicting their inputs, which leaves the retention of these characteristics implicit. They are also evaluated mainly on cognitive decoding and a narrow clinical task, i.e., seizure detection. On a broad, clinically relevant benchmark such as Omni-iEEG, they remain below task-specific models when used frozen. We introduce PHASE, a physiology-guided foundation model that makes these characteristics explicit learning targets, pairing them with masked latent prediction in a temporal stage (PHASE-T) within each channel and a spatiotemporal stage (PHASE-ST) across synchronized channels. PHASE is pretrained on heterogeneous recordings from 222 participants at nine clinical sites. On all five Omni-iEEG clinical tasks, frozen PHASE-T outperforms every evaluated foundation model by up to 31\%, and fine-tuned PHASE-T surpasses the task-specific models, setting a new state of the art. PHASE-T benefits from physiological supervision, outperforming variants trained with latent prediction alone or auxiliary waveform reconstruction on every task in matched ablations. PHASE-T generalizes to unseen institutions, outperforming the compared models with few or no local labels. PHASE-ST further improves seizure-onset-zone identification over PHASE-T and, when frozen, decodes sound volume and pitch on BrainTreebank better than published models. Beyond task performance, PHASE learns to encapsulate the physiological characteristics clinicians recognize, from seizure onset and its propagation to anatomical region identity, even though its pretraining contains no ictal recordings or anatomical labels.