- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper proposes E2fMoE for reconstructing regional BOLD signals from EEG. The model combines a shared latent-state module, an ROI-conditioned mixture-of-experts decoder, and FC-template calibration. It is evaluated on four simultaneous EEG-fMRI datasets for temporal reconstruction and FC preservation, with ablations and an expert routing analysis.
Critical strengths
-
The model is tested on four datasets, including in a cross-subject setting (Table A2). On NeuroBOLT, it reaches a correlation of 0.575 and an MSE of 0.238 (Table 1).
-
The ablations test each component separately, and the authors not only report overall scores but also analyze how ROIs are routed to experts.
Critical weaknesses
-
The method description doesn’t match the released code. Section 3.2 describes recurrent state updates over EEG tokens, but the code pools Transformer-encoded temporal patches and applies HemoSSMBlock across scale vectors. The routing statistics also differ: the Heschl–E2 weight is 0.46 in Table A3 but about 0.282 in the plotting script. Without clarification, the claims about the architecture and expert specialization can’t be verified.
-
The contribution of EEG to FC prediction is unclear. Removing the FC template lowers the FC score from 0.772 to 0.612 (Table A9), so the template plays a large role. A baseline that uses only the training-set template is needed to show how much scan-specific connectivity the model actually recovers from EEG.
Questions for authors
-
Which code version and configuration produced the main results and ablations? How do they relate to Section 3.2 and to the routing statistics in Figure 4 and Table A3?
-
Could you report a baseline that uses only the training-set FC template?
4 comments
[Part 2/2] Response to Reviewer gpRP
Q2: What evidence distinguishes the contribution of EEG-derived scan-specific information from that of the training-set FC template?
A2: We thank the reviewer for this important question. We agree that the fixed FC template provides a population-level connectivity prior. To distinguish its contribution from EEG-derived information, we conducted three complementary analyses.
(1) Direct template-only baseline comparison.
Following the reviewer's suggestion, we directly evaluated the training-set FC template without EEG input.
| Evaluation | Template-only FC ↑ | Full-model FC ↑ | Relative gain |
|---|---|---|---|
| Intra-scan | 0.407 | 0.772 ± 0.076 | 89.8% |
| Cross-subject | 0.345 | 0.583 ± 0.096 | 69.1% |
Relative gains were calculated using unrounded template-only values.
The full model outperforms the template-only baseline by 89.8% and 69.1% in the intra-scan and cross-subject settings, respectively. This directly demonstrates that the fixed FC template alone cannot explain the full model's connectivity prediction performance.
(2) Retraining ablations.
We further retrained the model after removing the FC template or randomly shuffling EEG–fMRI training pairs.
| Evaluation | Method | BOLD R ↑ | BOLD MSE ↓ | FC correlation ↑ |
|---|---|---|---|---|
| Intra-scan | Full model | 0.575 ± 0.170 | 0.238 ± 0.139 | 0.772 ± 0.076 |
| Intra-scan | Without FC template | 0.249 ± 0.019 | 1.028 ± 0.018 | 0.157 ± 0.022 |
| Intra-scan | Shuffled training pairs | -0.001 ± 0.017 | 1.099 ± 0.002 | 0.092 ± 0.071 |
| Cross-subject | Full model | 0.345 ± 0.040 | 0.219 ± 0.029 | 0.583 ± 0.096 |
| Cross-subject | Without FC template | 0.127 ± 0.007 | 1.037 ± 0.031 | 0.031 ± 0.014 |
| Cross-subject | Shuffled training pairs | 0.004 ± 0.025 | 0.999 ± 0.003 | 0.140 ± 0.133 |
Removing the FC template substantially degrades FC prediction, confirming its role as a structural prior. However, shuffling EEG–fMRI pairs also markedly reduces performance despite retaining the template, demonstrating the importance of meaningful EEG–fMRI correspondence.
(3) Test-time EEG replacement.
We kept the trained model, FC template, and target fMRI unchanged while replacing the correctly paired EEG with EEG from another subject or another scan of the same subject.
| Evaluation | EEG input | BOLD R ↑ | BOLD MSE ↓ | FC correlation ↑ |
|---|---|---|---|---|
| Intra-scan | Correct EEG, all scans | 0.575 ± 0.170 | 0.238 ± 0.139 | 0.772 ± 0.076 |
| Intra-scan | Other-subject EEG | -0.012 ± 0.008 | 1.111 ± 0.005 | 0.175 ± 0.031 |
| Intra-scan | Correct EEG, matched subset | 0.229 ± 0.007 | 0.994 ± 0.007 | 0.089 ± 0.017 |
| Intra-scan | Same-subject, other-scan EEG | 0.045 ± 0.014 | 1.152 ± 0.013 | 0.087 ± 0.015 |
| Cross-subject | Correct EEG, all scans | 0.345 ± 0.040 | 0.219 ± 0.029 | 0.583 ± 0.096 |
| Cross-subject | Other-subject EEG | 0.003 ± 0.009 | 1.103 ± 0.024 | 0.047 ± 0.017 |
| Cross-subject | Correct EEG, matched subset | 0.029 ± 0.006 | 1.095 ± 0.063 | -0.346 ± 0.038 |
| Cross-subject | Same-subject, other-scan EEG | 0.013 ± 0.016 | 1.106 ± 0.063 | -0.348 ± 0.036 |
Replacing correctly paired EEG with other-subject EEG reduces FC correlation from 0.772 to 0.175 (77.4% reduction) in the intra-scan setting and from 0.583 to 0.047 (92.0% reduction) in the cross-subject setting, despite the unchanged FC template.
For same-subject, different-scan replacement, BOLD correlation decreases on the matched subsets, whereas FC correlation changes only slightly. Thus, these results support EEG-dependent connectivity prediction, although scan-to-scan FC specificity remains less conclusive.
Revision location (highlighted in blue): Appendix J.4 (pp. 27–28), including the template-only baseline comparison (Table A15), retraining ablations (Table A16), and test-time EEG replacement experiments (Table A17). The corresponding intra-scan EEG replacement results are also updated in Section 4.7, Table 4 (p. 9).
Thank you for the further clarifications and additional results. Most of my concerns have been addressed, and I have increased my score accordingly.
Response to Reviewer gpRP
We sincerely thank the reviewer for thoughtful feedback and for taking the time to reconsider our work. We greatly appreciate your constructive suggestions, which have helped us improve the clarity and rigor of the paper.
[Part 1/2] Response to Reviewer gpRP
We sincerely thank the reviewer for the careful and constructive evaluation, and for acknowledging our extensive experiments and ablation studies. We address the reviewer's concerns point by point below:
Q1: Which code version and configurations produced the reported results, and how do they align with the temporal-state formulation and routing statistics?
A1: We apologize that an incorrect code version was inadvertently included in the original supplementary material. We have now re-uploaded the version used to produce the reported results.
The experiment configurations are organized as follows:
configs/neurobolt_intra_scan.yamlandconfigs/neurobolt_inter_scan.yaml.configs/minf_intra_scan.yaml.configs/neurobolt_global_inter_scan.yaml,configs/cpd_global_inter_scan.yaml, andconfigs/sef_sleep_global_inter_scan.yaml.The temporal-state implementation is provided in
src/eeg2fmri/models/inter_scan.py. Specifically,TemporalStateBlockexplicitly computes sk=Aˉksk−1+Bˉkxk and uk=Cˉksk+Dˉkxk. In the same file,BOLDSequenceSupervisorproduces the matching BOLD-derived latent sequence, andstate_supervision_lossapplies sequence-level squared-error and cosine alignment, corresponding to Section 3.2 and Eq. (2).ROI routing and the reported statistics. In src/eeg2fmri/models/multiscale_roi_moe.py, ROIMoEDecoder.forward computes routing logits from the input latent representation and ROI embeddings, then applies a softmax over experts. These per-input, per-ROI weights are the quantities to aggregate when reporting routing distributions.
Revision location (highlighted in blue): Appendix J.1 (p. 25), which clarifies the recurrent state formulation, BOLD-state alignment, and evaluation settings.