iEEG Can Help EEG: Improving Clinical Scalp EEG Analysis through Intracranial Priors
Authors: Xianglong Bao, Yi Ding, Zengxiang Li, Nan Liu, Kewen Wang, Zhe Wang, Ming Jin
Organizations: Griffith University School of Information and Communication Technology Brisbane, Australia · Nanyang Technological University College of Computing and Data Science Singapore · SingHealth Duke-NUS AI in Medicine Institute Singapore · National University of Singapore Duke-NUS Medical School Singapore
Scalp electroencephalography (EEG) is a cornerstone of clinical brain monitoring, making the quality of its signal fundamental to downstream tasks and models. However, that signal is heavily contaminated by non-neural interference and blurred by volume conduction. Existing methods operate almost exclusively within the scalp modality, so they inherit the contamination they would need to overcome. In practice, intracranial EEG (iEEG) observes the neural activity with far higher fidelity, yet it is invasive and therefore scarce, and the two modalities are rarely recorded as pairs at scale, share no common channel space, and offer no clean ground truth to directly supervise a mapping. Motivated by the neuronal synchronization mechanism shared by the two modalities, we propose a reprogramming approach that fits a signal-to-signal model on iEEG alone and freezes it as a cross-modal prior. We introduce ReBrain, a framework to transfer an iEEG-learned prior to scalp EEG, which re-expresses any scalp signal through that prior and can be inserted upstream of any downstream model with no pairing or fine-tuning. ReBrain consistently improves host models across recordings from multiple hospitals and across architecture families, and our analysis traces this gain to genuine neural structure carried by iEEG rather than to the architecture itself. This work provides evidence that a scarce clean modality can serve as a transferable prior for an abundant noisy one, offering a general recipe for cross-modal transfer among biosignals. Our code is available at https://github.com/Xianglong-Bao/ReBrain.
Figures & tables
Figure 1. Reprogramming scalp EEG through the frozen iEEG prior improves downstream performance on recordings from multiple hospitals that retain their native clinical class imbalance.
Figure 2. Overview of ReBrain . Stage A fits the neural prior on iEEG alone. Stage B freezes this prior and reprograms scalp EEG through it, re-expressing the raw signal in the iEEG-learned representation. Stage C feeds the reprogrammed signal to a downstream model of any architecture.
Figure 3. Effect of reprogramming on downstream seizure detection. The three bars per metric are the host model trained on raw scalp EEG, on EEG reprogrammed by a prior fitted to scalp EEG itself, and on EEG reprogrammed by the iEEG-fitted prior. Error bars are the cross-patient standard error and brackets mark per-patient Wilcoxon significance ( ∗p<0.05 , ∗∗p<0.01 , ∗∗∗p<0.001 ), both defined only where a per-patient distribution exists. TUSZ uses a subject-independent evaluation split and therefore carries neither error bars nor significance markers. Beirut has only five patients, for which a two-sided signed-rank test cannot reach p<0.05 even if every patient improves, so a consistent effect there carries no marker.
Figure 4. Per-feature ictal–interictal separability before and after reprogramming. Each panel is one morphology descriptor and each dot a per-subject standardized mean difference between the two classes, with a subject’s raw and reprogrammed values joined.
Figure 5. Reprogramming behaves as a denoiser rather than a pattern amplifier. Top: within-class scatter falls sharply under reprogramming. Bottom: the between-class distance separating the ictal and interictal centroids is unchanged. The separability gain therefore comes entirely from suppressing within-class variability, not from pushing the classes apart.
Figure 6. Effect of the corpus on which the prior is fitted. Each row is the density curve of the per-subject changes relative to raw. Vertical ticks mark the mean and stars mark a per-subject Wilcoxon signed-rank test against raw.
Appendix figures & tables5 assets
Supplementary material from the paper’s appendix.
Appendix
Front-end
BF1
Sn
Pr
BAcc
MCC
AUROC
Raw
0.681
0.717
0.706
0.852
0.688
0.955
Wavelet thresholding
0.657
0.733
0.654
0.852
0.662
0.955
ASR
0.667
0.732
0.690
0.848
0.674
0.953
ICA
0.661
0.697
0.705
0.841
0.671
0.956
EEGdenoiseNet (EOG)
0.604
0.651
0.630
0.817
0.611
0.945
EEGdenoiseNet (EMG)
0.626
0.661
0.650
0.823
0.631
0.939
Appendix
Table 1. Downstream seizure detection on CHB-MIT with EEGWaveNet as the host, comparing ReBrain against EEG denoising front-ends.
Figure 7. Representation-manifold view of reprogramming. (a) Individual windows in three dimensions and (b) class-density contours in two dimensions, for raw scalp EEG, reprogrammed scalp EEG, and iEEG, in the ReBrain latent representation embedded with t-SNE.
Fitting corpus
BF1
Sn
Pr
BAcc
MCC
AUROC
Raw
0.681
0.717
0.706
0.852
0.688
0.955
ReBrain (MC)
0.721
0.770
0.742
0.874
0.728
0.969
Appendix
Table 2. Downstream seizure detection on CHB-MIT with EEGWaveNet as the host, for a prior fitted on the Multi-Center (MC) iEEG corpus.
Component
Parameters
Encoder
1,055,432 ( 1.06 M)
Decoder
1,055,369 ( 1.06 M)
Codebook ( M=4 , K=256 , d=64 )
132,100 ( 0.13 M)
Deployed front-end
2,242,901 ( 2.24 M )
Discriminator (discarded)
1,004,298 ( 1.00 M )
Appendix
Table 3. Size of the deployed ReBrain front-end. The discriminator is used only while the prior is fitted and is not part of what is deployed.
Quantity
Value
Per 18 -channel 4 s window, one CPU thread
Latency
175 ms ( 148 – 234 )
Real-time factor
0.044 ( 0.037 – 0.059 )
Time to reprogram one hour of recording
157 s
Appendix
Table 4. Compute cost of reprogramming. Latency is the median over six runs on a single CPU thread of an Intel Xeon w3-2423, with the observed range in parentheses. The real-time factor is compute time divided by signal duration, so a value below one is faster than real time.
Electroencephalography (EEG) studies brain-generated electrical activity to understand cognition and support diagnosis, monitoring, and brain-computer interfaces.
Intracranial EEG (iEEG) provides high-fidelity neural recordings essential for clinical and brain-computer interface applications, but acquiring these signals requires invasive surgery. While recent studies have attempted to estimate iEEG from non-invasive scalp EEG, most rely on patient-specific models, creating a circular dependency: if surgery is required to collect training data, the non-invasive model offers limited practical benefit. In this study, we address the challenge of cross-subject iEEG reconstruction by predicting intracranial signals for unseen patients using models trained on other individuals. We propose CAST (Cross-Attention Spatial-Temporal Transformer), a machine learning framework that translates scalp EEG into multi-channel iEEG waveforms through a two-stage transfer learning strategy. First, a temporal encoder extracts multi-scale neural representations at three different resolutions. Then, because electrode placements vary substantially across patients, a channel-aware decoder is calibrated using only a few minutes of data from the target subject. We evaluated the proposed method using leave-one-subject-out cross-validation on two public datasets comprising 1,282 iEEG channels. Experimental results demonstrate that CAST reconstructs cortical signals located near the scalp surface substantially better than deep subcortical activity. In highly observable sensorimotor regions, the model achieved peak correlations of up to r=0.864 in the precentral gyrus. Furthermore, with a channel selection strategy, CAST obtained a mean correlation of r=0.545 on viable subjects, outperforming previous within-subject baselines. These findings indicate that cortical iEEG signals can be reconstructed for unseen subjects from scalp EEG without extensive patient-specific training, and that only a brief calibration phase is sufficient to adapt the model to new hardware configurations.
Tien-Dat Pham, Xuan-The Tran
HAI-Smartlink Research Lab, Anchi STE Company, Haiphong, Vietnam · School of Mechanical Engineering, Vietnam Maritime University, Haiphong, Vietnam
Electroencephalography (EEG) has become one of the key modalities underpinning brain-computer interfaces (BCIs) due to its high temporal resolution, rapid responsiveness, non-invasiveness, low cost, and portability. However, EEG signals are substantially inferior to intracranial EEG (iEEG) in signal-to-noise ratio and local spatial resolution, whereas iEEG suffers from extremely limited clinical accessibility owing to its invasive nature, hindering widespread application. To address this challenge, this study proposes a unified data-and prior knowledge-driven framework for EEG-iEEG representational enhancement. Guided by the principle that "geometric structure dictates function", the framework maps static cortical anatomy onto dynamic constraints governing neural signal propagation and integrates general-purpose neural representations extracted by a pre-trained large EEG model to explicitly model signal transmission through the brain. Enhanced EEG signals are then synthesized via a multidimensional representation diffusion process. Numerous experimental results demonstrate that the generated enhanced EEG signals effectively recover the neural activity patterns lost during propagation through the brain. This finding indicates that the performance ceiling of BCIs is constrained not only by acquisition hardware but also by the depth to which the generative model resolves the mechanisms of neural signal propagation. Collectively, the proposed framework provides a viable pathway toward acquiring high-fidelity neural signals at low cost.
Yihang Dong, Changhong Jing, Shuqiang Wang
Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China. · University of Chinese Academy of Sciences, Beijing, China.
Although obtaining deep brain activity from non-invasive scalp electroencephalography (sEEG) is crucial for neuroscience and clinical diagnosis, directly generating high-fidelity intracranial electroencephalography (iEEG) signals remains a largely unexplored field, limiting our understanding of deep brain dynamics. Current research primarily focuses on traditional signal processing or source localization methods, which struggle to capture the complex waveforms and random characteristics of iEEG. To address this critical challenge, this paper introduces NeuroFlowNet, a novel cross-modal generative framework whose core contribution lies in the first-ever reconstruction of iEEG signals from the entire deep temporal lobe region using sEEG signals. NeuroFlowNet is built on Conditional Normalizing Flow (CNF), which directly models complex conditional probability distributions through reversible transformations, thereby explicitly capturing the randomness of brain signals and fundamentally avoiding the pattern collapse issues common in existing generative models. Additionally, the model integrates a multi-scale architecture and self-attention mechanisms to robustly capture fine-grained temporal details and long-range dependencies. Validation results on a publicly available synchronized sEEG-iEEG dataset demonstrate NeuroFlowNet's effectiveness in terms of temporal waveform fidelity, spectral feature reproduction, and functional connectivity restoration. This study establishes a more reliable and scalable new paradigm for non-invasive analysis of deep brain dynamics. The code of this study is available in https://github.com/hdy6438/NeuroFlowNet
Dongyi He, Bin Jiang, Kecheng Feng +5
School of Artificial Intelligence, Chongqing University of Technology, Chongqing, 400054, China · School of Smart Health, Chongqing Polytechnic University of Electronic Technology, Chongqing, 401331, China