Pith. sign in

REVIEW 3 major objections 6 minor 23 references

The paper claims that replacing surface normals with TRELLIS's 1024-dimensional latent surface features improves aneurysm classification and segmentation on the Intra3D benchmark and reduces blood-flow simulation error on AnXplore by about

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

TRELLIS-derived surface features improve aneurysm classification, segmentation, and hemodynamic simulation, including a 15% lower blood-flow prediction error.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection Frozen TRELLIS features nearly solve aneurysm classification, but the paper's cross-domain transfer claim needs a split clarification and a high-dimensional control. the 3 major comments →

arxiv 2509.03095 v1 pith:KUGSL4XM submitted 2025-09-03 cs.CV cs.LG

TRELLIS-Enhanced Surface Features for Comprehensive Intracranial Aneurysm Analysis

Cl\'ement Herv\'e , Paul Garnier , Jonathan Viquerat , Elie Hachem This is my paper
classification cs.CV cs.LG
keywords intracranial aneurysm3D point cloud classification3D mesh segmentationhemodynamics simulationfeature transferTRELLISgraph neural networkmedical imaging
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper sets out to show that latent surface features from TRELLIS, a 3D generative model trained on ordinary non-medical objects, can be transplanted into medical tasks where annotated aneurysm data are scarce. The authors replace the standard point-cloud inputs—3D coordinates plus surface normals—with TRELLIS's 1024-dimensional per-point features, and feed the same features into a graph neural network that predicts blood flow. On the Intra3D benchmark, this lifts classification accuracy on vessels to about 99.9% and on aneurysms to 100% (F1 = 0.999) for PointNet++ at 512 points, and it improves segmentation IoU and Dice scores across the board. On the AnXplore blood-flow benchmark, adding the features cuts the GNN's all-rollout RMSE by roughly 15%. A plain MLP trained on the feature vectors alone, with no spatial coordinates, already separates aneurysms from vessels, which the authors read as evidence that the pretrained encoder has captured reusable shape knowledge that transfers across domains.

Core claim

The central claim is that the geometric representation learned by a general-purpose 3D generative encoder transfers to a specialized medical domain without any fine-tuning. Concretely, the paper shows that PointNet and PointNet++ fed with TRELLIS surface features instead of point normals achieve near-perfect classification and segmentation on the Intra3D aneurysm benchmark, surpassing the previous baselines from the dataset paper and from 3DMedPT. The same features, concatenated with existing mesh node features in a GNN with transformer blocks, reduce the all-rollout RMSE of blood-flow simulation on AnXplore by about 15%, averaged over the S/1 and L/1 model sizes. The paper also shows that t

What carries the argument

The load-bearing object is the TRELLIS surface feature: a 1024-dimensional latent vector attached to each point of a mesh, produced by rendering the object from multiple views, voxelizing it into a 64x64x64 grid, and encoding the voxel-feature pairs with a transformer-based sparse variational autoencoder that consumes DINOv2 visual features. In the classification and segmentation experiments, these vectors replace surface normals as the point-wise input to PointNet and PointNet++; in the simulation experiments, they are concatenated with the mesh node features of a GNN with masked multi-head self-attention. The feature vector is what carries the cross-domain knowledge, and it is the only var

Load-bearing premise

The paper's load-bearing premise is that the improved results come from geometric knowledge transferred by TRELLIS's encoder, rather than from the sheer dimensionality of the 1024-dimensional feature vectors or from global cues like object orientation and overall shape.

What would settle it

Train the same PointNet++, MLP, and GNN pipelines with a matched 1024-dimensional per-point descriptor that carries no pretrained geometric knowledge—for example, a frozen random linear projection of coordinates and normals, or features from an untrained TRELLIS encoder—and compare accuracy and RMSE. If the gains persist, the claim that pretrained generative-model knowledge causes the improvement is refuted; if they vanish, the transfer interpretation is supported.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Pretrained 3D generative encoders can act as ready-made feature extractors for small medical point-cloud datasets, reducing the need to train 3D encoders from scratch.
  • With these features, simple architectures reach near-ceiling accuracy on the Intra3D classification benchmark, so further progress on this dataset would have to come from harder or more varied data.
  • Adding the same fixed features to a hemodynamics GNN improves rollout fidelity by about 15% without changing the solver architecture, suggesting geometry-conditioning is a cheap way to improve physics proxies.
  • Because the features separate aneurysm and vessel regions in latent space, they could support weakly supervised detection and labeling of vascular structures with far fewer manual annotations.
  • The one-time encoding cost (about five minutes per object) is amortized across the multiple downstream tasks that reuse the same features.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The paper does not compare against an equally high-dimensional descriptor that carries no learned geometric knowledge, so a reader should treat the cross-domain transfer interpretation as open until a random-feature or untrained-encoder control is run.
  • The clustering by size and morphology suggests the same features could feed rupture-risk stratification models, but the paper only shows correlation with hemodynamic metrics, not a predictive risk model.
  • If the transfer effect is real, the strategy may extend to other 3D medical structures with scarce annotations—bone, cardiac, or pulmonary meshes—though only cerebrovascular cases are demonstrated here.
  • The feature extraction pipeline's use of rendered views means the features may partly encode viewing direction and global object pose; ablating orientation alignment or testing cropped local patches would clarify what the downstream networks actually exploit.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper proposes using latent features from the pretrained TRELLIS 3D generative model as additional point/mesh descriptors for three downstream tasks: classification of aneurysms versus healthy vessels on Intra3D, segmentation of aneurysm/vessel regions on the same dataset, and prediction of time-evolving hemodynamics on the AnXplore dataset. The authors replace point normals (or augment mesh features) with 1024-dimensional TRELLIS features, and report large improvements in classification accuracy/F1, segmentation IoU/DSC, and a roughly 15% reduction in simulation RMSE. The paper also includes feature-space visualizations, a clustering analysis, and an MLP-only ablation, and makes the code publicly available.

Significance. The paper targets a timely and practically relevant question: whether features from a large-scale 3D generative model can compensate for scarce medical data. It evaluates the idea on three tasks and reports consistent improvements; the public code and multi-seed reporting are strengths. I agree there is no circularity: the TRELLIS encoder is fixed and was not trained on the target labels. However, the specific contribution claimed—cross-domain transfer of geometric knowledge—is not yet demonstrated because input dimensionality and global object cues are confounded with pretraining. If the requested controls confirm the transfer effect, the paper would be a useful addition to the medical 3D deep-learning literature.

major comments (3)
  1. [Section 4.2, Table 1] The central claim is that pretrained TRELLIS geometry transfers to medical 3D tasks, but the experiments confound transfer with input dimensionality and global context. The baseline uses coordinates + normals (3D per point), while the TRELLIS condition uses 1024-dimensional per-point features. The MLP-only ablation (Section 4.2) excludes coordinates but still feeds 1024D features, so it does not rule out memorization of object-level statistics. Moreover, the TRELLIS encoder is a global, multi-view model; its per-voxel tokens may encode orientation, bounding-box shape, or global vessel context rather than reusable local surface geometry. Please add at least one non-transfer high-dimensional control, e.g., random Fourier features of coordinates/normals projected to 1024D, a random/untrained TRELLIS encoder, or PCA/random projections of normals to 1024D. The same control is needed for the G
  2. [Section 6.1, Table 1] The classification protocol is under-specified. The text says 'training 5 times the models over 5 different seeds' but does not describe the train/validation/test split. If evaluation is on the training set, the 99.88/100 accuracy values are uninterpretable. Please state the split, whether it is stratified by class, and whether all models share identical point sampling. In addition, the authors' PointNet++ baseline without TRELLIS features reaches F1 ≈ 0.81 and aneurysm accuracy ≈ 52%, whereas the published PointNet++ results from Yang et al. (2020) on the same benchmark report F1 ≈ 0.89 and aneurysm accuracy ≈ 86.7% at 512 points. This large gap suggests protocol or implementation differences; the claimed 'state-of-the-art' comparison must be made under an identical protocol, including the same training setup and preprocessing.
  3. [Section 2.1] The description of the classification dataset is ambiguous: '1,694 healthy vessels and 215 aneurysms, which can be combined with the 116 annotated aneurysms used to train segmentation models, resulting in a total of 331 aneurysms for classification.' In the original Intra3D dataset, the 116 annotated segmentation cases are a subset of the 215 classification cases. If that is the case, combining them double-counts 116 cases; if they are disjoint, that must be stated explicitly. This affects the composition of every classification result in Table 1.
minor comments (6)
  1. [Table 2 caption] Caption says 'results are ... F1-score' but the table reports IoU and DSC; correct the caption or the table.
  2. [Table 1 caption] Caption says 'vessels segment accuracy' and 'aneurysms segment accuracy' for a classification task; rename to 'vessel accuracy' and 'aneurysm accuracy'.
  3. [Section 3.2] The statement 'the optimal number of clusters for the 101 aneurysms was 15' gives no method (silhouette, elbow, etc.) for choosing k; please specify.
  4. [Section 6.4] The discussion of reducing rendering views from 200 to 100/50 is speculative and not backed by experiments; either add an ablation or clearly label it as future work.
  5. [References] The citation 'A et al., 2024' should be expanded to a proper author list (Goetz, A. et al.) in the text, and the reference formatting should be consistent.
  6. [Abstract / Section 5.2] The abstract says 'replacing conventional point normals or mesh descriptors', but Section 5.2 states that TRELLIS features are 'combined with the original mesh features' for the GNN experiments. Please make the description consistent.

Circularity Check

0 steps flagged

No significant circularity: TRELLIS features are fixed external inputs and downstream gains are measured empirically.

full rationale

The paper's central claims are empirical performance comparisons, not analytic derivations that reduce to their inputs. TRELLIS features are extracted from a fixed, pretrained generative model trained on non-medical 3D data; the encoder is not fit to Intra3D/AnXplore labels or to the downstream predictions. Classification, segmentation, and simulation models are trained on these fixed features, and the reported accuracy/F1/RMSE improvements are measured outcomes, not quantities defined in terms of fitted constants or self-citations. The GNN architecture and AnXplore dataset come from prior work by the same group, but they serve only as baseline tools and data; the with-versus-without TRELLIS feature comparison is conducted within the paper and does not depend on accepting the prior work's conclusions. No uniqueness theorem, ansatz, or fitted parameter is imported from self-citations, and no prediction is constructed from the target labels. The potential confound of 1024-dimensional TRELLIS features versus 3D normals is a threat to the transfer interpretation, but it is a question of experimental control, not circularity: the feature values themselves are not defined in terms of the claimed outcomes.

Axiom & Free-Parameter Ledger

2 free parameters · 3 axioms · 0 invented entities

The central claim does not rest on fitted constants, but it assumes the TRELLIS feature mapping is meaningful at the point level and that the medical ground truths (labels and CFD simulations) are reliable. The only hand-tuned parameters are the rendering view count and the exploratory cluster count.

free parameters (2)
  • Number of rendering views = 200
    Hand-chosen for TRELLIS encoding; authors note it accounts for ~80% of compute and suggest reducing to 50-100.
  • Number of clusters (AnXplore) = 15
    Chosen as 'optimal' for the clustering analysis in Section 3.2; not part of the main task results but an exploratory free parameter.
axioms (3)
  • domain assumption TRELLIS encoder features provide meaningful point-level geometric information for medical meshes despite being trained on non-medical data.
    Central hypothesis tested in the paper; assumed true when features are mapped to sampled points in Section 4.1.
  • domain assumption AnXplore CFD simulations are a valid ground truth for hemodynamic prediction.
    The simulation error is measured against these simulations (Section 5.2); if the CFD data is unreliable, the 15% improvement is on a flawed target.
  • domain assumption Intra3D labels (healthy vs aneurysm, point-level annotations) are correct.
    Used for classification and segmentation training/evaluation (Section 2.1).

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of TRELLIS-Enhanced Surface Features for Comprehensive Intracranial Aneurysm Analysis." pith.science (2026). https://pith.science/paper/KUGSL4XM

@misc{pith2026250903095,
  author       = {Pith},
  title        = {Pith review of: TRELLIS-Enhanced Surface Features for Comprehensive Intracranial Aneurysm Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KUGSL4XM}},
  note         = {Machine review of arXiv:2509.03095}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Intracranial aneurysms pose a significant clinical risk yet are difficult to detect, delineate and model due to limited annotated 3D data. We propose a cross-domain feature-transfer approach that leverages the latent geometric embeddings learned by TRELLIS, a generative model trained on large-scale non-medical 3D datasets, to augment neural networks for aneurysm analysis. By replacing conventional point normals or mesh descriptors with TRELLIS surface features, we systematically enhance three downstream tasks: (i) classifying aneurysms versus healthy vessels in the Intra3D dataset, (ii) segmenting aneurysm and vessel regions on 3D meshes, and (iii) predicting time-evolving blood-flow fields using a graph neural network on the AnXplore dataset. Our experiments show that the inclusion of these features yields strong gains in accuracy, F1-score and segmentation quality over state-of-the-art baselines, and reduces simulation error by 15\%. These results illustrate the broader potential of transferring 3D representations from general-purpose generative models to specialized medical tasks.

Figures

Figures reproduced from arXiv: 2509.03095 by Cl\'ement Herv\'e, Elie Hachem, Jonathan Viquerat, Paul Garnier.

Figure 1
Figure 1. Figure 1: Examples of intracranial aneurysms and vessels from the Intra3D dataset. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Examples of aneurysms from the AnXplore dataset [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Illustrations of the entire pipeline based on TRELLIS to extract features for [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Results of PCA on the Intra3D dataset for classification and AnXplore dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Result of t-SNE on the Intra3D dataset for classification and AnXplore dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Results of t-SNE on annotated aneurysms from Intra3D (Yang et al., 2020) [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Results of PCA and t-SNE on all the features from an annotated aneurysm from [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Results of clustering on the AnXplore dataset using PCA on the mean features. [PITH_FULL_IMAGE:figures/full_fig_p010_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Correlation heatmap between the 3 main features from TRELLIS and several [PITH_FULL_IMAGE:figures/full_fig_p010_9.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

23 extracted references · 2 canonical work pages

  1. [3]

    Amirhossein Arzani, Jian-Xun Wang, and Roshan M

    doi: 10.1109/TMI.2023.3265000. Amirhossein Arzani, Jian-Xun Wang, and Roshan M. D’Souza. Uncover- ing near-wall blood flow from sparse data with physics-informed neural networks. Physics of Fluids , 33(7), July

    work page arXiv 2023

  2. [4]

    doi: 10.1063/5.0055600

    ISSN 1089-7666. doi: 10.1063/5.0055600. URL http://dx.doi.org/10.1063/5.0055600. Julian Suk, Pim de Haan, Phillip Lippe, Christoph Brune, and Jelmer M. Wolterink. Mesh neural networks for se(3)-equivariant hemodynamics esti- mation on the artery wall.Computers in Biology and Medicine, 173:108328, May

    doi:10.1063/5.0055600

  3. [7]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N

    URL https://arxiv.org/abs/2010.03409. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. Attention is all you need. CoRR, abs/1706.03762,

    Pith/arXiv arXiv 2010

  4. [9]

    Hanxiao Jiang, Hao-Yu Hsu, Kaifeng Zhang, Hsin-Ni Yu, Shenlong Wang, and Yunzhu Li

    URL https://arxiv.org/ abs/2503.10110. Hanxiao Jiang, Hao-Yu Hsu, Kaifeng Zhang, Hsin-Ni Yu, Shenlong Wang, and Yunzhu Li. Phystwin: Physics-informed reconstruction and simulation of deformable objects from videos,

    arXiv

  5. [10]

    URL https://arxiv.org/abs/ 2503.17973. Matt Deitke, Ruoshi Liu, Matthew Wallingford, Huong Ngo, Oscar Michel, Aditya Kusupati, Alan Fan, Christian Laforte, Vikram Voleti, Samir Yitzhak Gadre, Eli VanderBilt, Aniruddha Kembhavi, Carl Von- drick, Georgia Gkioxari, Kiana Ehsani, Ludwig Schmidt, and Ali Farhadi. Objaverse-xl: A universe of 10m+ 3d objects,

    Pith/arXiv arXiv

  6. [11]

    org/abs/2307.05663

    URL https://arxiv. org/abs/2307.05663. Jasmine Collins, Shubham Goel, Kenan Deng, Achleshwar Luthra, Leon Xu, Erhan Gundogdu, Xi Zhang, Tomas F. Yago Vicente, Thomas Diderik- sen, Himanshu Arora, Matthieu Guillaumin, and Jitendra Malik. Abo: Dataset and benchmarks for real-world 3d object understanding,

    Pith/arXiv arXiv

  7. [14]

    Charles Ruizhongtai Qi, Hao Su, Kaichun Mo, and Leonidas J

    URL https://arxiv.org/abs/2306.11290. Charles Ruizhongtai Qi, Hao Su, Kaichun Mo, and Leonidas J. Guibas. Point- net: Deep learning on point sets for 3d classification and segmentation. CoRR, abs/1612.00593,

    Pith/arXiv arXiv

  8. [16]

    Yangyan Li, Rui Bu, Mingchao Sun, and Baoquan Chen

    URL http://arxiv.org/abs/1706.02413. Yangyan Li, Rui Bu, Mingchao Sun, and Baoquan Chen. Pointcnn. CoRR, abs/1801.07791, 2018a. URL http://arxiv.org/abs/1801.07791. 22 Jiaxin Li, Ben M. Chen, and Gim Hee Lee. So-net: Self-organizing network for point cloud analysis. CoRR, abs/1803.04249, 2018b. URL http:// arxiv.org/abs/1803.04249. Wenxuan Wu, Zhongang Qi...

    Pith/arXiv arXiv

  9. [18]

    URL https://arxiv.org/abs/2304.07193. Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Min- derer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at scale,

    Pith/arXiv arXiv

  10. [19]

    Zhirong Wu, Shuran Song, Aditya Khosla, Fisher Yu, Linguang Zhang, Xiaoou Tang, and Jianxiong Xiao

    URL https://arxiv.org/abs/2010.11929. Zhirong Wu, Shuran Song, Aditya Khosla, Fisher Yu, Linguang Zhang, Xiaoou Tang, and Jianxiong Xiao. 3d shapenets: A deep represen- tation for volumetric shapes. In 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , pages 1912–1920,

    Pith/arXiv arXiv 2010

  11. [22]

    Biao Zhang and Rico Sennrich

    URL https://arxiv.org/abs/1606.08415. Biao Zhang and Rico Sennrich. Root mean square layer normalization,

    Pith/arXiv arXiv

  12. [23]

    23 Appendix A

    URL https://arxiv.org/abs/1910.07467. 23 Appendix A. Supplementary figures on the analysis of the TREL- LIS features Figure A.10: Results of PCA on the Intra3D dataset combined with the AnXplore dataset. Figures show the mean, standard deviation, minimum, and maximum of the PCA compo- nents. Figure A.11: Results of PCA on the Intra3D dataset. Figures show...

    Pith/arXiv arXiv 1910

  13. [2015]

    Ilya Loshchilov and Frank Hutter

    doi: 10.1109/CVPR.2015.7298801. Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization,

    arXiv 2015

  14. [2016]

    Charles Ruizhongtai Qi, Li Yi, Hao Su, and Leonidas J

    URL http://arxiv.org/abs/1612.00593. Charles Ruizhongtai Qi, Li Yi, Hao Su, and Leonidas J. Guibas. Pointnet++: Deep hierarchical feature learning on point sets in a metric space. CoRR, abs/1706.02413,

    Pith/arXiv arXiv

  15. [2017]

    URL http://arxiv.org/abs/1706. 03762. Jianfeng Xiang, Zelong Lv, Sicheng Xu, Yu Deng, Ruicheng Wang, Bowen Zhang, Dong Chen, Xin Tong, and Jiaolong Yang. Structured 3d latents for scalable and versatile 3d generation. arXiv preprint arXiv:2412.01506 ,

    Pith/arXiv arXiv

  16. [2018]

    URL http: //arxiv.org/abs/1811.07246. Maxime Oquab, Timoth´ ee Darcet, Th´ eo Moutakanni, Huy Vo, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, Mahmoud Assran, Nicolas Ballas, Wojciech Galuba, Russell Howes, Po-Yao Huang, Shang-Wen Li, Ishan Misra, Michael Rabbat, Vasu Sharma, Gabriel Synnaeve, Hu X...

    Pith/arXiv arXiv

  17. [2019]

    Dan Hendrycks and Kevin Gimpel

    URL https://arxiv.org/abs/1711.05101. Dan Hendrycks and Kevin Gimpel. Gaussian error linear units (gelus),

    Pith/arXiv arXiv

  18. [2020]

    Mukul Khanna, Yongsen Mao, Hanxiao Jiang, Sanjay Haresh, Brennan Shacklett, Dhruv Batra, Alexander Clegg, Eric Undersander, Angel X

    URL https://arxiv.org/abs/2009.09633. Mukul Khanna, Yongsen Mao, Hanxiao Jiang, Sanjay Haresh, Brennan Shacklett, Dhruv Batra, Alexander Clegg, Eric Undersander, Angel X. Chang, and Manolis Savva. Habitat synthetic scenes dataset (hssd-200): An analysis of 3d scene scale and realism tradeoffs for objectgoal naviga- tion,

    Pith/arXiv arXiv 2009

  19. [2021]

    20 Yifan Liu, Wuyang Li, Jie Liu, Hui Chen, and Yixuan Yuan

    URL https://arxiv.org/abs/2112.04863. 20 Yifan Liu, Wuyang Li, Jie Liu, Hui Chen, and Yixuan Yuan. Grab-net: Graph-based boundary-aware network for medical point cloud segmenta- tion. IEEE Transactions on Medical Imaging , 42(9):2776–2786,

    Pith/arXiv arXiv

  20. [2022]

    Huan Fu, Rongfei Jia, Lin Gao, Mingming Gong, Binqiang Zhao, Steve May- bank, and Dacheng Tao

    URL https://arxiv.org/abs/2110.06199. Huan Fu, Rongfei Jia, Lin Gao, Mingming Gong, Binqiang Zhao, Steve May- bank, and Dacheng Tao. 3d-future: 3d furniture shape with texture,

    Pith/arXiv arXiv

  21. [2023]

    doi: 10.1038/s41597-022-01721-8

    ISSN 2052-4463. doi: 10.1038/s41597-022-01721-8. URL http://dx.doi.org/10.1038/ s41597-022-01721-8 . Jianhui Yu, Chaoyi Zhang, Heng Wang, Dingxin Zhang, Yang Song, Tiange Xiang, Dongnan Liu, and Weidong Cai. 3d medical point transformer: In- troducing convolution to attention networks for medical point cloud anal- ysis,

    doi:10.1038/s41597-022-01721-8 2052

  22. [2024]

    doi: 10.1016/j.compbiomed.2024.108328

    ISSN 0010-4825. doi: 10.1016/j.compbiomed.2024.108328. URL http://dx.doi.org/10.1016/j.compbiomed.2024.108328. Paul Garnier, Vincent Lannelongue, Jonathan Viquerat, and Elie Hachem. Training transformers for mesh-based simulations,

    work page arXiv 2024

  23. [2025]

    Goetz A, Jeken-Rico P, Pelissier U, Chau Y, S´ edat J, and Hachem E

    URL https:// arxiv.org/abs/2508.18051. Goetz A, Jeken-Rico P, Pelissier U, Chau Y, S´ edat J, and Hachem E. Anx- plore: a comprehensive fluid-structure interaction study of 101 intracranial aneurysms

    Pith/arXiv arXiv

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.