The DOI matches its Crossref record; 14 of the 18 references it cites were found in a registry or an index; 2 of the 13 sources its claims cite were checked for support.
Hetero-Modal Variational Encoder-Decoder for Joint Modality Completion and Segmentation
Open accessAbstract
We propose a new deep learning method for tumour segmentation when dealing with missing imaging modalities. Instead of producing one network for each possible subset of observed modalities or using arithmetic operations to combine feature maps, our hetero-modal variational 3D encoder-decoder independently embeds all observed modalities into a shared latent representation. Missing data and tumour segmentation can be then generated from this embedding. In our scenario, the input is a random subset of modalities. We demonstrate that the optimisation problem can be seen as a mixture sampling. In addition to this, we introduce a new network architecture building upon both the 3D U-Net and the Multi-Modal Variational Auto-Encoder (MVAE). Finally, we evaluate our method on BraTS2018 using subsets of the imaging modalities as input. Our model outperforms the current state-of-the-art method for dealing with missing modalities and achieves similar performance to the subset-specific equivalent networks.
Overview
CoChat Summary
TL;DR
The paper proposes U-HVED, a hetero-modal variational 3D encoder-decoder that jointly completes missing MRI modalities and segments brain tumours from arbitrary input modality subsets. On BraTS2018, it outperforms the prior missing-modality state of the art HeMIS and approaches subset-specific U-Net models.
Key findings
- U-HVED achieved mean Dice scores of 80.1% for complete tumour, 64.0% for core tumour, and 50.0% for enhancing tumour, compared with U-HeMIS at 78.6%, 59.7%, and 48.1%, and HeMIS at 50.7%, 28.7%, and 27.4%, respectively.
- U-HVED significantly outperformed U-HeMIS in 13 out of 15 cases for complete tumour, 10 out of 15 for core tumour, and 11 out of 15 for enhancing tumour, with significance assessed by Wilcoxon test (\(p < 0.05\)).
- U-HVED performed similarly to the 15 subset-specific Single U-Net models, whose mean Dice scores were 81.6% for complete tumour, 66.2% for core tumour, and 52.7% for enhancing tumour.
- Modality completion was evaluated qualitatively; the authors report good-quality T1 and FLAIR reconstructions and comparable reconstructions using 3 versus 4 modalities.
Methods
- Design: a 3D variational encoder-decoder (U-HVED) combining a 3D U-Net with a Multi-Modal Variational Auto-Encoder, using 4 modality-specific encoders and 5 decoders for the 4 imaging modalities plus segmentation.
- Data: BraTS18 training set with 285 patients (210 high-grade glioma, 75 low-grade glioma), each scanned with T1, T1c, T2, and FLAIR; 3-fold validation by random split into training (70%), validation (10%), and testing (20%).
- Analysis: segmentation performance measured by Dice Similarity across all possible input modality subsets; training used mixture sampling over modality subsets, \(112 \times 112 \times 112\) patches, Adam optimizer, and early stopping.
Limitations
- The paper states that modality completion was evaluated only qualitatively because image completion is “a means rather than an end.”
- The paper states future work is needed to analyze the learned embedding, which is challenging due to the multi-scale representation of the hidden variable.
- Evident from the text: evaluation is limited to the BraTS18 training set with a 3-fold random split, with no external validation or quantitative completion metrics reported.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
1 Valid · 3 Not valid · 10 Not checked · 24 Own results and interpretations
None of the sources cited for this claim have been checked yet.
In the paper
“Most of them require the full set of n modalities as input [4,9], while a scenario of missing modalities is common in practice.”
What the sources say
- Not checked yetcites [4] Isensee, F., Kickingereder, P., Wick, W., Bendszus, M., Maier-Hein, K.H.: No new-net. In: Crimi, A., Bakas, S., Kuijf, H., Keyvan, F., Reyes, M., van Walsum, T. (eds.) BrainLes 2018. LNCS, vol. 11384, pp. 234–244. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-11726-9_21 source
- Could not verifycites [9] Myronenko, A.: 3D MRI brain tumor segmentation using autoencoder regularization. In: Crimi, A., Bakas, S., Kuijf, H., Keyvan, F., Reyes, M., van Walsum, T. (eds.) BrainLes 2018. LNCS, vol. 11384, pp. 311–320. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-11726-9_28 sourceExperimental — accuracy not yet measured
Our copy of Myronenko et al. never characterizes what other methods require as input: the Related work (pp. 2-3) reviews EMMA, Wang et al., Isensee et al., McKinley et al. and Zhou et al. without stating their modality-input requirements, and the Introduction (p. 1) only notes that several complementary modalities (T1, T1c, T2, FLAIR) are usually acquired and that BraTS 2018 provides four co-registered modalities. The only statements about full-modality input concern this paper's own network (Results, p. 6: 'We concatenated 4 available 3D MRI modalities into the 4 channel image as an input'; Fig. 1 caption, p. 3), which is an instance of the practice rather than the claimed survey-level generalization about 'most' methods. Searched for: missing/incomplete modalities, 'all four modalities as input', 'most methods require'; no passage in our copy addresses the claim, and nothing contrary is stated either. (LLM, conf=0.60)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Unlike these three approaches, U-HVED ( Ours ) generates missing modalities.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Interestingly, our model tries to reconstruct the tumour information even when the tumour information is missing or not clear, such as in T1 scans.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Moreover, comparable reconstructions are performed using 3 modalities and 4 modalities.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“First, the U-Net architecture in U-HeMIS always achieves better performance than the original 2D fully-convolutionnal HeMIS .”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
17 more claims
Create a free account to see all 38 claims and to verify the 11 references still open.
- Probably not a substantial claim. 12
- It may state two separate claims as one. 7
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Performance of these methods depend on the generation model that imputes the missing modalities.”
From Robust Multimodal Learning with Missing Modalities via Parameter-Efficient Adaptation (2025) · cites this paper as [26]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 2.1 Multi-modal Variational Auto-Encoders (MVAE) · p. 3
“To make sampling from z | x amenable to back-propagation, reparametrisation is used [5]: µ φ ( x ) + Σ φ ( x ) × ϵ where ϵ ∼ N (0 , I ).”
§ 2.1 Multi-modal Variational Auto-Encoders (MVAE) · p. 3
“h φ ( x ) = ( µ φ ( x ) , Σ φ ( x ) )”
Citations
Cites
U-Net: Convolutional Networks for Biomedical Image Segmentation
Ronneberger, O., Fischer, P., & Brox, T. (2015) · Lecture notes in computer science · doi:10.1007/978-3-319-24574-4_28 · Cited by 99,404
Auto-Encoding Variational Bayes
Kingma, D. P., & Welling, M. (2013) · arXiv · doi:10.48550/arxiv.1312.6114 · Cited by 20,067 · cited 2×
“[5] proposed to optimise, with respect to (φ, θ), the evidence lower-bound (ELBO):”
BackgroundV-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation
Milletari, F., Navab, N., & Ahmadi, S.-A. (2016) · 2016 Fourth International Conference on 3D Vision (3DV) · doi:10.1109/3dv.2016.79 · Cited by 11,931
The multimodal brain tumor image segmentation benchmark (BRATS)
Menze, B. H., Jakab, A., Bauer, S., Kalpathy-Cramer, J., Farahani, K., Kirby, J., Burren, Y., Porz, N., Slotboom, J., Wiest, R., et al. (2015) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2014.2377694 · Cited by 6,781
3D MRI Brain Tumor Segmentation Using Autoencoder Regularization
Myronenko, A. (2019) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-11726-9_28 · Cited by 1,283
NiftyNet: a deep-learning platform for medical imaging
Gibson, E., Li, W., Sudre, C., Fidon, L., Shakir, D. I., Wang, G., Eaton-Rosen, Z., Gray, R., Doel, T., Hu, Y., et al. (2018) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2018.01.025 · Cited by 552
Self-citationDeep Learning Based Imaging Data Completion for Improved Brain Disease Diagnosis
Li, R., Zhang, W., Suk, H.-I., Wang, L., Li, J., Shen, D., & Ji, S. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10443-0_39 · Cited by 469
Isensee, F., Kickingereder, P., Wick, W., Bendszus, M., & Maier-Hein, K. H. (2019) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-11726-9_21 · Cited by 341
HeMIS: Hetero-Modal Image Segmentation
Havaei, M., Guizard, N., Chapados, N., & Bengio, Y. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46723-8_54 · Cited by 332
Ladder Variational Autoencoders
Sønderby, C. K., Raiko, T., Maaløe, L., Sønderby, S. K., & Winther, O. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1602.02282 · Cited by 237 · cited 1×
“Unlike the existing hierarchical VAE models [11,14], we propose a fully convolutional network.”
Method
Cited by
Zhang, Y., He, N., Yang, J., Li, Y., Wei, D., Huang, Y., Zhang, Y., He, Z., & Zheng, Y. (2022) · Lecture Notes in Computer Science · doi:10.1007/978-3-031-16443-9_11 · Cited by 236
Multi-Modal Learning with Missing Modality via Shared-Specific Feature Modelling
Wang, H., Chen, Y., Ma, C., Avery, J., Hull, L., & Carneiro, G. (2023) · 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52729.2023.01524 · Cited by 216
“1, which compares ShaSpec with current state-of-the-art (SOTA) methods, including U-HeMIS [13], U-HVED [8], Robust-MSeg (RbSeg) [5] and mmFormer (mmFm) [35].”
BackgroundMethodInfluentialRFNet: Region-aware Fusion Network for Incomplete Multi-modal Brain Tumor Segmentation
Ding, Y., Yu, X., & Yang, Y. (2021) · 2021 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00394 · Cited by 175
“Comparisons with the state-of-the-art methods, including HeMIS [14], U-HVED [10] and RobustSeg [3], on BRATS2020.”
BackgroundMethodInfluentialKnowledge Distillation from Multi-modal to Mono-modal Segmentation Networks
Hu, M., Maillard, M., Zhang, Y., et al. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-59710-8_75 · Cited by 154
“[3], inspired by HeMIS, proposed U-HVED where they introduced skip-connections by considering intermediate layers, before each down-sampling step, as a feature map.”
MethodResultModality-Aware Mutual Learning for Multi-modal Medical Image Segmentation
Zhang, Y., Yang, J., Tian, J., Shi, Z., Zhong, C., Zhang, Y., & He, Z. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-87193-2_56 · Cited by 160
“The strategies proposed to handle missing modalities include synthesizing missing modalities by a generative model [16] or learn a modality-invariant feature space [4,6].”
Background
Metadata
- Authors
- Dorent, R., Joutard, S., Modat, M., Ourselin, S., & Vercauteren, T.
- Affiliations
- King's College London (GB)
- Corresponding author
- Vercauteren, Tom
- Venue
- Lecture Notes in Computer Science
- Publisher
- Springer International Publishing
- Published
- 2019
- Pages
- 74-82
- ISSN
- 0302-9743 (ISSN-L)
- 1611-3349
- OpenAlex
- W2962678076
- Cited by
- 160
- References
- 18
- Citations by year
- 2026: 19 · 2025: 38 · 2024: 35 · 2023: 33 · 2022: 8 · 2021: 18 · 2020: 7 · 2019: 2
- FWCI
- 2.85
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- www.springernature.com/gp/researchers/text-and-data-mining (tdm)
- www.springernature.com/gp/researchers/text-and-data-mining (vor)
- Locations
- Lecture notes in computer science (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Medical Image Segmentation Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Brain Tumor Detection and Classification
- Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Radiomics and Machine Learning in Medical Imaging
- Keywords
- tumor segmentation, shared latent representation, 3D U-Net, BraTS 2018, encoder-decoder architecture
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Hetero-Modal Variational Encoder-Decoder for Joint Modality Completion and Segmentation. https://app.cochat.ai/papers/10.1007/978-3-030-32245-8_9
- The paper
- Dorent, R., Joutard, S., Modat, M., Ourselin, S., & Vercauteren, T. (2019). Hetero-Modal Variational Encoder-Decoder for Joint Modality Completion and Segmentation. Lecture Notes in Computer Science. https://doi.org/10.1007/978-3-030-32245-8_9
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.