- Rating
- 2
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 1
Read review
Paper summary
This paper proposes TopoScanPE to represent each ROI through the topology of its local neighborhood, encodes the resulting Topo-Scan sequences, and concatenates them with connectivity profiles in BrainNetTF.
Critical strengths
- The ROI-level encoding provides a sensible way to incorporate local topology into attention-based brain network models.
- Results on ABIDE are encouraging, with improvements over the reported baselines across all four metrics.
Critical weaknesses
- The description for some parts of the method is incomplete. The Forman–Ricci node function and heat kernel signature are not explicitly defined. Their formulas, Laplacian convention, and computation within each neighborhood are needed for reproducibility. Distinct quantile values can produce variable-length sequences, whereas the encoder assumes a fixed . Please explain how duplicate thresholds, constant functions, and padding are handled.
- The efficiency analysis is missing. Report the preprocessing cost of ROI-wise topology computation, additional parameters, and training/inference time to substantiate the description of the encoder as lightweight.
Questions for authors
- What specific pattern has been captured by the proposed PE? How does it align with existing findings for a specific disease?
- Does the topology encoding remain beneficial when added to a second transformer backbone?
- Existing work explicitly explores positional and structural encodings for brain networks [1-3]. Why these works are not included for discussion and comparison?
[1] Brain-JEPA: Brain Dynamics Foundation Model with Gradient Positioning and Spatiotemporal Masking. NeurIPS 2024 [2] Contrasformer: A Brain Network Contrastive Transformer for Neurodegenerative Condition Identification. CIKM 2024 [3] A Brain Graph Foundation Model: Pre-Training and Prompt-Tuning across Broad Atlases and Disorders. ICLR 2026
4 comments
Response to Reviewer WCKQ (2/4)
W2. Efficiency
The efficiency analysis is missing. Report the preprocessing cost of ROI-wise topology computation, additional parameters, and training/inference time to substantiate the description of the encoder as lightweight.
Reply: The signature encoder adds 106,960 parameters (2.4% of the model). Topology is computed once per subject before training, in about 3.6 s per subject (about 1 h for all of ABIDE), and cached. Training and inference are therefore faster than with RWSE or LapPE, which are recomputed in every forward pass.
Cost (ABIDE, BrainNetTF backbone, 200 epochs per run, mean over the five splits; all on one NVIDIA RTX PRO 5000 Blackwell GPU)
| Encoding | Pre-processing (once) | Training (min per run) | Inference (ms per subject) |
|---|---|---|---|
| RWSE (ALTER's encoding) | — (computed in every forward pass) | 4.77 | 1.09 |
| LapPE | — (computed in every forward pass) | 16.24 | 4.49 |
| TopoScanPE | ≈ 3.6 s per subject (≈ 1 h for 1,009 subjects × 200 ROI neighbourhoods) | 4.35 | 0.40 |
Q1. Which pattern does the encoding capture, and how does it relate to disease findings?
What specific pattern has been captured by the proposed PE? How does it align with existing findings for a specific disease?
Reply: The encoding describes how each ROI's local functional neighbourhood is organised, not how strongly the ROI is connected. Along the Forman–Ricci order of the edges (from edges between hubs to peripheral edges) and the heat-kernel order of the vertices, Topo-Scan records whether the neighbourhood stays connected or splits (β₀), how many cycles are not filled by triangles (β₁), and the slice sizes. It thus distinguishes an ROI whose neighbours form a tight module from one whose neighbours are loosely connected or split into groups, a local measure of segregation versus integration. On the trained models, the learned code predicts clustering (R² = 0.50), betweenness (0.38) and participation (0.37) far better than RWSE (0.07, 0.17, 0.07), and carries little information about connection strength.
Both disorders are commonly described at this level: autism by an atypical balance between local and long-range functional connectivity (Belmonte et al., 2004; Courchesne & Pierce, 2005), and Alzheimer's disease by reduced local clustering and a loss of small-world organisation (Supekar et al., 2008). On ABIDE, 39 of 200 ROIs have signatures that differ between ASD and controls (FDR q < 0.05), mostly in the somatomotor and subcortical/cerebellar networks. In 29 of them the signature is lower in ASD, i.e. fewer vertices, edges, components and cycles per slice; 10, mostly default-mode ROIs, are higher. Half of the 20 ROIs the trained model relies on most are among these 39 (hypergeometric p = 0.001; new Fig. 6, Appendix D). On ADNI (124 participants) no ROI survives FDR correction, so we draw no disease-level conclusion there. We have added this interpretation to Appendix D (Figs. 5 and 6).
Response to Reviewer WCKQ (3/4)
Q2. A second transformer backbone
Does the topology encoding remain beneficial when added to a second transformer backbone?
Reply: Yes, for some backbones. We added TopoScanPE, unchanged, to Graphormer, GraphTrans and SAN on ABIDE and ADNI.
Other backbones (test metrics, mean ± std over five splits; base: scores from Table 1; + TopoScanPE: layers and dropout chosen per split on validation loss; Table 3 of the revised paper)
| Dataset | Backbone | Variant | AUC | ACC | SEN | SPE |
|---|---|---|---|---|---|---|
| ABIDE | Graphormer | base (Table 1) | 63.91 ± 4.05 | 61.88 ± 6.85 | 66.30 ± 9.98 | 55.74 ± 11.00 |
| + TopoScanPE | 69.81 ± 2.72 | 64.63 ± 3.78 | 67.78 ± 5.35 | 61.27 ± 5.11 | ||
| GraphTrans | base (Table 1) | 60.13 ± 6.73 | 57.83 ± 4.71 | 65.70 ± 10.30 | 49.77 ± 11.52 | |
| + TopoScanPE | 57.19 ± 4.01 | 53.79 ± 3.26 | 44.72 ± 11.96 | 64.48 ± 8.27 | ||
| SAN | base (Table 1) | 71.35 ± 2.18 | 65.34 ± 2.91 | 55.41 ± 9.29 | 68.39 ± 7.50 | |
| + TopoScanPE | 61.59 ± 6.49 | 57.83 ± 5.49 | 59.02 ± 10.33 | 57.09 ± 8.50 | ||
| ADNI | Graphormer | base (Table 1) | 60.69 ± 5.26 | 55.75 ± 3.18 | 60.18 ± 11.36 | 47.75 ± 13.53 |
| + TopoScanPE | 64.36 ± 5.32 | 62.31 ± 8.57 | 41.82 ± 12.33 | 77.33 ± 17.18 | ||
| GraphTrans | base (Table 1) | 61.41 ± 3.65 | 58.60 ± 5.41 | 65.57 ± 6.05 | 54.37 ± 3.42 | |
| + TopoScanPE | 49.45 ± 10.76 | 51.54 ± 7.54 | 25.45 ± 23.98 | 70.67 ± 24.07 | ||
| SAN | base (Table 1) | 66.11 ± 3.41 | 61.78 ± 4.22 | 53.94 ± 7.56 | 63.63 ± 8.51 | |
| + TopoScanPE | 61.82 ± 7.84 | 61.54 ± 7.69 | 27.27 ± 19.07 | 86.67 ± 17.38 |
Against their Table 1 scores, TopoScanPE improves Graphormer on both datasets (AUC +5.90 on ABIDE and +3.67 on ADNI, with higher accuracy). It does not improve GraphTrans (AUC −2.94 on ABIDE, −11.96 on ADNI) or SAN (−9.76 and −4.29), so we describe the transfer to other backbones as partial.
Response to Reviewer WCKQ (4/4)
Q3. Existing positional and structural encodings for brain networks
Existing work explicitly explores positional and structural encodings for brain networks [1-3]. Why these works are not included for discussion and comparison?
Reply: Thank you for these references; we have added them to the related work. Brain-JEPA (Dong et al., NeurIPS 2024) positions each ROI by its coordinates on population-level connectivity gradients, and Contrasformer (Xu et al., CIKM 2024) adds a learnable ROI identity embedding; both are the same for every subject. BrainGFM (Wei et al., ICLR 2026) uses RWSE as its structural encoding. TopoScanPE is instead computed per subject from the topology of each ROI's own neighbourhood and can be added to an existing backbone. Under the same backbone and protocol, RWSE reaches 79.92 AUC on ABIDE against 81.96 for TopoScanPE. We also trained the three models themselves on our five ABIDE splits and the five ADNI splits, with their published architectures and learning rates (Brain-JEPA and BrainGFM fine-tuned from their released pre-trained checkpoints) and the same model selection as TopoScanPE:
ABIDE; dropout of each baseline chosen per split on validation loss; TopoScanPE as in Table 1 (test metrics, mean ± std)
| Model | AUC | ACC | SEN | SPE |
|---|---|---|---|---|
| Brain-JEPA | 56.82 ± 2.31 | 52.81 ± 3.60 | 55.20 ± 29.78 | 51.96 ± 26.75 |
| Contrasformer | 69.87 ± 2.96 | 64.24 ± 2.43 | 71.22 ± 5.35 | 56.95 ± 10.78 |
| BrainGFM | 72.85 ± 2.98 | 67.68 ± 1.70 | 63.54 ± 5.41 | 72.17 ± 5.59 |
| TopoScanPE | 81.96 ± 1.04 | 75.47 ± 0.81 | 74.99 ± 6.58 | 76.34 ± 5.58 |
TopoScanPE has the highest AUC, accuracy, sensitivity and specificity.
ADNI; dropout of each baseline chosen per split on validation loss; TopoScanPE as in Table 1 (test metrics, mean ± std)
| Model | AUC | ACC | SEN | SPE |
|---|---|---|---|---|
| Brain-JEPA | 51.45 ± 9.69 | 64.62 ± 8.21 | 32.73 ± 9.27 | 88.00 ± 9.80 |
| Contrasformer | 65.58 ± 4.75 | 61.54 ± 8.43 | 60.00 ± 14.77 | 62.67 ± 20.04 |
| BrainGFM | 63.64 ± 7.79 | 60.00 ± 10.20 | 50.91 ± 9.27 | 66.67 ± 18.38 |
| TopoScanPE | 71.15 ± 2.93 | 72.31 ± 1.72 | 58.18 ± 13.79 | 82.67 ± 12.11 |
On ADNI, TopoScanPE also has the highest AUC and accuracy.
Response to Reviewer WCKQ (1/4)
Dear Reviewer WCKQ, thank you for the thoughtful review and the questions about interpretation and related work. We added the missing definitions, a cost analysis, an analysis of the patterns the encoding captures and their relation to ASD and AD findings, results on other backbones, and a comparison with Brain-JEPA, Contrasformer and BrainGFM. We hope these additions resolve your remaining doubts and encourage you to reconsider your assessment.
W1. Definitions, duplicate thresholds, constant functions and padding
Reply: Thank you for pointing this out; these details are indeed needed for reproducibility. We have added the definitions to Appendix A, and Sec. 4.2 and Appendix A now explain the duplicate thresholds and the padding:
All quantities are computed on the induced subgraph Gᵢ = (Vᵢ, Eᵢ) of ROI i's bounded two-hop neighbourhood in the unweighted, symmetrised k-nearest-neighbour graph (k = 8); degrees are taken within Gᵢ.
Forman–Ricci curvature (edge function). We use the combinatorial Forman curvature of an unweighted graph without higher-order cells, F(a, b) = 4 − deg(a) − deg(b) for an edge (a, b) ∈ Eᵢ. Topo-Scan uses F directly on the edges: the thresholds are quantiles of {F(e)} over Eᵢ, and each slice is the subgraph formed by the edges whose curvature falls in the window, together with their endpoints.
Heat kernel signature (node function). We use the symmetric normalised Laplacian of Gᵢ, ℒ = I − D^(−1/2) A D^(−1/2) (rows of vertices without edges equal those of I). With the eigendecomposition ℒ = Σₖ λₖ φₖ φₖᵀ, HKSₜ(v) = Σₖ exp(−t λₖ) φₖ(v)², at t = 1, computed by a full eigendecomposition of each neighbourhood (at most 81 vertices on ABIDE and 37 on ADNI at ρ = 0.4).
Duplicate thresholds and padding. For each filtration function f we take Q = L + m = 22 quantile thresholds of its values (over Eᵢ for Forman–Ricci, over Vᵢ for HKS) and remove duplicates, leaving Q′ ≤ Q distinct thresholds. The windows α_ℓ ≤ f ≤ α_{ℓ+m} give min(Q′ − m, L) slices. Thresholds are never repeated to fill the sequence: the remaining positions are zero columns, (0, 0, 0, 0), so every sequence has length L = 20. Sequences are shorter than L for 7.54% (Forman–Ricci) and 0.26% (HKS) of the ROIs on ABIDE, and for 13.43% and 0.11% on ADNI. We have corrected Eq. (4), which incorrectly implied L = Q − m.
Constant functions. The thresholds are taken from the values of f within each ROI neighbourhood, so f is constant only if it has the same value at every edge (Forman–Ricci) or vertex (HKS) of Gᵢ, e.g. when Gᵢ is a complete graph. This is rare (0.21% of the ROIs on ABIDE, none on ADNI) and needs no separate rule: no threshold is repeated, so all L slices are zero columns. Appendix A now states this.