- Rating
- 3
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
HYPERION fine-tunes TITAN in Lorentz space using caption-conditioned Einstein midpoints, contrastive and entailment losses, confidence-filtered patch pseudo-labels, and intermediate-norm propagation. It greatly improves zero-shot tumor localization, roughly preserves slide-level prediction, and gives small report-generation gains with qualitatively better MLLM lesion grounding.
Critical strengths
The paper tests whether the gain harms the original representation, using 36 slide-level tasks and an MLLM. This is much broader than segmentation alone.
Critical weaknesses
The key mechanistic claim is under-isolated: patch pseudo-supervision is the largest contributor, the matched Euclidean ablation is under-specified, and no radius/cone-activity diagnostics establish that the model is genuinely exploiting hyperbolic geometry rather than mainly improved supervision and radial rescaling. In addition, cross-model segmentation Dice uses different native tile grids and excludes negative slides, so the absolute “best on five benchmarks” comparison is less conclusive than the strong same-grid HYPERION-vs-TITAN result.
Questions for authors
-
Could the authors precisely define “w/o hyperbolic space”? Report matched Euclidean-vs-hyperbolic controls with radius, cone-activation, and hierarchy-violation statistics, plus variance across seeds for Tab. 4.
-
Can all methods be evaluated on a common grid, with false-positive rates on tumor-negative slides?
-
Are mutation captions used in L_pat? How accurate are the pseudo-abnormal and pseudo-normal labels against ground truth, given that low similarity to one abnormal caption need not imply normal tissue?
6 comments
Rebuttal for Reviewer MEY1 [2/3]
3) Radius and cone diagnostics support the intended ordering. We added HoroPCA projections of held-out TCGA slides to Appendix C.4 (Figure A3), comparing TITAN, HYPERION, and ablations without hyperbolic space or norm propagation. Original-space radii confirm on all slides (0.76, 1.23, 2.03), with pseudo-tumor/normal radii of 1.55/1.27. Before fine-tuning, the midpoint lies inside the caption radius (0.36 vs. 0.78), and groups overlap in projection. Without hyperbolic space, the midpoint radius collapses to 0.0055 (caption: 0.38; slide: 0.88).
Table R3 reports cone-apex angles () for the three learned containment relations on TCGA slides across five seeds. Smaller angles indicate tighter alignment of a child with its parent’s cone axis. Fine-tuning reduces TITAN’s caption-to-slide and caption-to-midpoint angles from 132.6° and 151.8° to 35.5° and 39.7° for HYPERION. The ablations further show how the hierarchy changes when model components are removed.
Table R3. Entailment cone angle statistics on held-out TCGA.
| Method | |||
|---|---|---|---|
| TITAN (before fine-tuning) | 132.6 | 151.8 | 88.9 |
| HYPERION | 35.5±1.3 | 39.7±0.9 | 40.4±2.6 |
| w/o hyperbolic space | 73.8±4.6 | 165.7±0.9 | 38.1±2.3 |
| w/o entailment | 65.0±5.1 | 63.3±6.3 | 72.3±3.1 |
| w/o midpoint contrastive | 46.8±2.3 | 51.0±2.8 | 56.8±1.2 |
| w/o | 68.9±4.1 | 87.7±7.6 | 38.2±2.0 |
| w/o norm propagation | 42.3±2.1 | 47.9±1.8 | 53.5±1.3 |
We included the HoroPCA projections and radius statistics in Appendix C.4 (L. 1295–1305) and Figure A3, and the cone-angle statistics in Appendix C.4 (L. 1306–1313) and Table A16, referenced in Sec. 5.5 (L. 469).
4) The variance across seeds for Table 4. Table R4 shows the standard deviations for Table 4. We included the standard deviations in Appendix C.3 (L. 1259–1262) and Table A15.
Table R4. Component ablation of Table 4 over five seeds (Dice, mean ± std).
| Setting | CM16 | CM17 | PAIP Prostate | PAIP Colon | Overall |
|---|---|---|---|---|---|
| TITAN (no fine-tuning) | 0.143 | 0.098 | 0.248 | 0.563 | 0.263 |
| HYPERION | 0.463±0.023 | 0.404±0.028 | 0.773±0.004 | 0.783±0.034 | 0.606±0.022 |
| w/o hyperbolic space | 0.271±0.049 | 0.286±0.015 | 0.700±0.029 | 0.746±0.020 | 0.501±0.016 |
| w/o entailment | 0.364±0.045 | 0.306±0.021 | 0.734±0.020 | 0.701±0.009 | 0.526±0.012 |
| w/o midpoint contrastive | 0.408±0.031 | 0.375±0.030 | 0.689±0.019 | 0.688±0.029 | 0.540±0.021 |
| w/o | 0.296±0.018 | 0.238±0.013 | 0.421±0.035 | 0.641±0.024 | 0.399±0.018 |
| w/o norm propagation | 0.330±0.020 | 0.256±0.024 | 0.613±0.034 | 0.756±0.015 | 0.489±0.016 |
[W2] In addition, cross-model segmentation Dice uses different native tile grids and excludes negative slides, so the absolute “best on five benchmarks” comparison is less conclusive than the strong same-grid HYPERION-vs-TITAN result.
We evaluated all methods on a common grid in both directions: on the 224 and 512 grid (Table R5). Consequently, HYPERION remains the best on every common grid, so the ranking does not depend on the grid. Every method uses the identical grid and ground-truth rasterization. We exclude PRISM because it is the only method without per-tile outputs. We already reported PRISM in its native block setting in Table 1. From Table R5, HYPERION achieves the best Dice in all eight settings. Its margin over the strongest baseline is 0.13–0.19 on CAMELYON16, CAMELYON17, and PAIP Colon, and about 0.35 on PAIP Prostate. The ranking does not depend on the choice of grid. We address the negative slides in our response to Q2.
Table R5. Dice on common grids (224 grid / 512 grid).
| Dice (224 grid / 512 grid) | CM16 | CM17 | PAIP Prostate | PAIP Colon |
|---|---|---|---|---|
| HYPERION (Ours) | 0.416 / 0.463 | 0.381 / 0.404 | 0.759 / 0.773 | 0.778 / 0.783 |
| PathGen-CLIP | 0.265 / 0.282 | 0.250 / 0.251 | 0.263 / 0.283 | 0.586 / 0.607 |
| CHIEF | 0.146 / 0.173 | 0.085 / 0.096 | 0.402 / 0.435 | 0.571 / 0.600 |
| CONCH | 0.155 / 0.183 | 0.112 / 0.138 | 0.291 / 0.303 | 0.503 / 0.604 |
| TITAN | 0.128 / 0.143 | 0.090 / 0.098 | 0.235 / 0.248 | 0.564 / 0.563 |
We clarified the grid settings in Appendix A.2 (L. 862–863), and included the common-grid comparisons in Appendix D.1 (L. 1403, 1428–1433) and Table A20.
Rebuttal for Reviewer MEY1 [3/3]
[Q2] Can all methods be evaluated on a common grid, with false-positive rates on tumor-negative slides?
We measured tile-level false positives on tumor-free CM16 and CM17 slides, reporting the median and zero-FP slide fraction alongside the outlier-sensitive mean (Table R6). HYPERION keeps false positives near zero without sacrificing tumor detection. CONCH, CHIEF, TITAN, and PRISM flag 23–52% of normal tissue as tumor, exceeding 1% on 95–100% of slides. HYPERION maintains near-zero FPR on both grids (mean 0.6% on CM16, 0.06% on CM17), with the most zero-FP slides (32.1% and 53.5%).
Table R6. False-positive-rates (FPR) on tumor-negative slides (224 grid / 512 grid). † PRISM has no per-tile output, so it is reported only on its native 224 grid.
| Method | Mean FPR ↓ | Median FPR ↓ | FP=0 slides ↑ | FP>1% slides ↓ |
|---|---|---|---|---|
| CAMELYON16 normal | ||||
| HYPERION | 0.0058/0.0055 | 0.00047/0.00046 | 32.1%/32.1% | 1.2%/1.2% |
| PathGen-CLIP | 0.0263/0.0049 | 0.01764/0.00126 | 0.0%/17.3% | 86.4%/12.3% |
| CONCH | 0.3397/0.3235 | 0.32678/0.29122 | 0.0%/0.0% | 100%/100% |
| CHIEF | 0.4761/0.4503 | 0.47193/0.44866 | 0.0%/0.0% | 100%/100% |
| TITAN | 0.4926/0.4881 | 0.50413/0.49675 | 0.0%/0.0% | 100%/100% |
| PRISM† | 0.2991/– | 0.27276/– | 0.0%/– | 100%/– |
| CAMELYON17 negative | ||||
| HYPERION | 0.0006/0.0006 | 0.00000/0.00000 | 53.5%/53.5% | 0.3%/0.3% |
| PathGen-CLIP | 0.0152/0.0020 | 0.01254/0.00111 | 0.3%/21.1% | 61.6%/2.2% |
| CONCH | 0.2988/0.2580 | 0.26730/0.22257 | 0.0%/0.0% | 100%/100% |
| CHIEF | 0.5158/0.4911 | 0.52408/0.49568 | 0.0%/0.3% | 99.7%/99.7% |
| TITAN | 0.3994/0.3961 | 0.39900/0.39426 | 0.0%/0.0% | 99.7%/99.7% |
| PRISM† | 0.2273/– | 0.14609/– | 0.6%/– | 95.3%/– |
We clarified the evaluation on tumor-negative slides in Appendix A.1 (L. 782–783), and included the false-positive analysis in Appendix D.1 (L. 1444–1453) and Table A21.
[Q3] Are mutation captions used in L_pat? How accurate are the pseudo-abnormal and pseudo-normal labels against ground truth, given that low similarity to one abnormal caption need not imply normal tissue?
Mutation captions. Yes, each slide carries one unified caption pool formed as the union of its report captions and its mutation captions. We clarified the unified pool of report and mutation captions in Appendix A.4 (L. 951–953).
Pseudo-label precision and tumor burden. We found pseudo-normal precision of at least 98.9% across tumor-burden groups and high pseudo-abnormal precision in ranges relevant to training (Tables R7–R9). As TCGA lacks patch captions, we used tumor annotations from CAMELYON16, CAMELYON17, PAIP Prostate, and PAIP Colon. We pooled patches by slide tumor-overlap fraction. denotes patches paired with abnormal captions in , and the caption-specific pseudo-normal set.
Tumor contamination in was low across all groups (Table R7), despite the ambiguity of inferring normality from low caption similarity. For tumor burden ≥1%, average precision was 90.0%, 94.1%, 79.2%, and 72.3%, respectively (Table R8). Precision was limited below 1%, occurring only in CAMELYON16 and CAMELYON17. However, such cases represented only 0.4% of training slides, while 99.2% had ≥25% tumor cells (Table R9), suggesting limited training impact. TCGA tumor-cell percentages came from the portal metadata field percent_tumor_cells.
Table R7. Pseudo-normal precision stratified by tumor burden.
| Benchmark / Tumor burden | <1% | 1–5% | 5–10% | 10–25% | 25–50% | ≥50% |
|---|---|---|---|---|---|---|
| CAMELYON16 | 100% (2,258/2,258) | 100% (1,143/1,143) | 100% (1,264/1,264) | 100% (717/717) | 100% (93/93) | 100% (27/27) |
| CAMELYON17 | 99.93% (1,472/1,473) | 100% (978/978) | 100% (644/644) | 100% (298/298) | 100% (172/172) | — |
| PAIP Prostate | — | 99.93% (4,536/4,539) | 99.92% (3,639/3,642) | 100% (5,182/5,182) | 99.35% (1,676/1,687) | 98.90% (450/455) |
| PAIP Colon | — | 100% (407/407) | 100% (1,624/1,624) | 100% (4,226/4,226) | 99.97% (5,903/5,905) | 99.23% (2,196/2,213) |
Table R8. Pseudo-abnormal precision stratified by tumor burden.
| Benchmark / Tumor burden | <1% | 1–5% | 5–10% | 10–25% | 25–50% | ≥50% |
|---|---|---|---|---|---|---|
| CAMELYON16 | 10.3% (73/711) | 65.0% (169/260) | 99.6% (557/559) | 91.4% (469/513) | 89.7% (87/97) | 93.7% (118/126) |
| CAMELYON17 | 6.0% (42/702) | 98.6% (273/277) | 92.2% (178/193) | 100% (120/120) | 71.8% (51/71) | — |
| PAIP Prostate | — | 79.0% (828/1,048) | 86.6% (1,248/1,441) | 82.5% (2,123/2,572) | 65.3% (611/936) | 53.5% (131/245) |
| PAIP Colon | — | 13.2% (14/106) | 44.9% (241/537) | 68.8% (1,826/2,653) | 78.3% (2,591/3,307) | 78.7% (1,221/1,551) |
Table R9. Tumor-cell percentage of the TCGA slides used for training.
| Metric/Tumor cells | <1% | 1–5% | 5–10% | 10–25% | 25–50% | ≥50% | 50–75% | 75–90% | ≥90% |
|---|---|---|---|---|---|---|---|---|---|
| Slides | 26 | 0 | 2 | 25 | 408 | 6,115 | 1,607 | 1,844 | 2,664 |
| Share | 0.4% | 0.0% | <0.1% | 0.4% | 6.2% | 93.0% | 24.4% | 28.0% | 40.5% |
We added the pseudo-label precision and tumor-burden analyses to Appendix C.5 (L. 1377–1393) and Tables A18–A19.
Rebuttal for Reviewer MEY1 [1/3]
We thank the reviewer for recognizing HYPERION’s improved zero-shot tumor localization and MLLM lesion grounding, and our evaluation across 36 slide-level tasks and an MLLM to assess whether grounding gains preserve the original representations. We address the reviewer’s comments below.
We address four points: 1) the contribution beyond patch alignment, 2) clarification of the “w/o hyperbolic space” and matched comparisons with unnormalized Euclidean learning, 3) the interpretation of the radius and cone diagnostics, and 4) variance across seeds for Table 4.
1) Patch-level alignment does not account for HYPERION’s full improvement. Table R1 shows that hyperbolic learning provides a clearly favorable structure for grounding, further strengthened by patch alignment, by comparing diverse fine-tuning configurations under matched data and budget. Specifically, Euclidean slide alignment achieves overall Dice of 0.219, below TITAN’s 0.263. Adding pseudo-caption-based patch alignment raises Dice to 0.458, still below HYPERION (0.606). Without patch alignment, HYPERION reaches 0.399, substantially outperforming TITAN. Along the Lorentz sequence, Dice increases to 0.345 with entailment, 0.366 with midpoint learning, and 0.399 with norm propagation, without patch supervision.
Table R1. Fine-tuning TITAN with the same training data and budget across four datasets (Dice). Line in bold is HYPERION (ours)
We included the matched fine-tuning controls in Sec. 5.5 (L. 430–431, 460–462) and the final rows of Table 4 (L. 456–458), with the incremental analysis and per-dataset results in Appendix C.3 (L. 1239–1246) and Table A14, and the single-component Lorentz controls in Table 5.
2) Clarification of the “w/o hyperbolic space” ablation and matched comparison to unnormalized Euclidean controls. The “w/o hyperbolic space” in Table 4 means replacing Lorentz representations with unnormalized Euclidean ones while retaining all other proposed components. We note that Table R1 uses standard (L2-normalized) contrastive fine-tuning as in TITAN, whereas Table R2 uses unnormalized Euclidean controls; hence their Euclidean entries differ (0.219 vs. 0.266 with slide alignment; 0.458 vs. 0.455 with patch alignment). By extending this control to matched ablations in both spaces, we found that Lorentz-space learning consistently outperforms unnormalized Euclidean learning (Table R2). Specifically, Lorentz learning outperforms its Euclidean counterpart across all evaluated configurations and datasets. Notably, adding the remaining components to patch alignment improves average Dice from 0.455 to 0.501 in Euclidean space (+0.046), versus 0.499 to 0.606 in Lorentz space (+0.107). This larger gain supports the benefit of hyperbolic hierarchical modeling beyond norm preservation alone.
Table R2. Component ablations in unnormalized Euclidean (Euc.) and Lorentz (Lor.) spaces.
We clarified the definition of “w/o hyperbolic space” in Sec. 5.5 (L. 421–422, 466–467) and Appendix A.7 (L. 998–1003), and added the matched Euclidean–Lorentz comparison to Sec. 5.5 (L. 464–470) and Table 5.