UteroVAE: A Shape-Informed Variational Autoencoder for Uterine MRI Encoding in Adenomyosis, Fibroids, and Healthy Uteri

In: Lecture Notes in Computer Science · 2025 · pp. 71–81 · doi:10.1007/978-3-032-05825-6_7 · W4414448912
book-chapter OA: closed CC0
Full text JSON View on OpenAlex View at publisher
AI-generated summary by claude@2026-06+body, 2026-06-07

This study adapted a VAE to jointly encode uterine MRI scans and segmented cavities, creating shape-informed embeddings that better capture disease patterns and improve downstream clustering.

One-sentence paraphrase of the abstract; not a substitute for reading it. No clinical advice. How this works

AI-generated deep summary by claude@2026-06, 2026-06-07 · read from full text

The paper studied how to encode uterine MRI shapes and morphologic configurations for characterizing uterine disorders, focusing on adenomyosis and also considering fibroids and healthy uteri, using a variational autoencoder approach. The authors adapted a fastMRI pre-trained VAE and designed a hybrid model that jointly encodes MRI scans and segmented uterine cavity masks with an anatomy-informed prior, then evaluated whether fine-tuning and this joint encoding improve alignment with clinically relevant disease patterns and downstream clustering. They report that both fine-tuning and hybrid encoding produced more clinically relevant embeddings and improved clustering performance. The main limitation stated in the provided text is that results are described at a high level without detailing dataset composition, evaluation metrics, or external validation. This paper is centrally about endometriosis—specifically adenomyosis—by using uterine MRI shape-informed latent representations to capture diagnostic contour patterns such as the “question mark sign.”

Read from the paper's body, not the abstract. Not a substitute for reading the paper. No clinical advice. How this works

Full text 7,033 characters · extracted from oa-doi-fallback · 2 sections · click to expand

Abstract

Uterine disorders, such as adenomyosis and fibroids, are major contributors to pelvic pain, abnormal uterine bleeding, and infertility. Morphologic configurations and geometric alterations of the uterine cavity serve as critical imaging biomarkers in clinical diagnosis. One well established example is the question mark sign, a highly specific indicator of adenomyosis, characterized by distinctive uterine contour distortions. However, beyond this singular marker, a broader range of shape variations may hold diagnostic significance. To systematically capture these morphologic and geometric patterns, we adapted a Variational Autoencoder (VAE) pre-trained on fastMRI datasets. Instead of encoding MRI images alone, we designed the model to jointly incorporate both the segmented uterine cavity and MRI scans. By embedding an anatomy-informed prior, the model is better equipped to characterize structural anatomy relevant to uterine pathology. Our results indicate that both fine-tuning the VAE and using the hybrid encoding approach produce embeddings that align more closely with clinically relevant disease patterns and improve downstream clustering performance. By refining the joint representation of segmentation and MRI data, our method could enhance the potential of latent diffusion models for extracting imaging biomarkers in female pelvic disorders. Access this chapter Tax calculation will be finalised at checkout Purchases are for personal use only Similar content being viewed by others

References

Bourdon, M., et al.: Adenomyosis: an update regarding its diagnosis and clinical features. J. Gynecol. Obstet. Hum. Reprod. 50, 102228 (2021) Freytag, D., Günther, V., Maass, N., Alkatout, I.: Uterine fibroids and infertility. Diagnostics 11, 1455 (2021) Andres, M.P., Borrelli, G.M., Ribeiro, J., Baracat, E.C., Abrão, M.S., Kho, R.M.: Transvaginal ultrasound for the diagnosis of adenomyosis: systematic review and meta-analysis. J. Minim. Invasive Gynecol. 25, 257–264 (2018) Zannoni, L., et al.: Question mark sign and transvaginal ultrasound uterine tenderness for the diagnosis of adenomyosis: a prospective validation. J. Ultrasound Med. 39, 1405–1412 (2020) Atlason, H.E., Love, A., Sigurdsson, S., Gudnason, V., Ellingsen, L.M.: SegAE: unsupervised white matter lesion segmentation from brain MRIs using a CNN autoencoder. NeuroImage: Clin. 24, 102085 (2019) Muhammad, H., et al.: Unsupervised subtyping of cholangiocarcinoma using a deep clustering convolutional autoencoder. In: Shen, D., et al. (eds.) MICCAI 2019. LNCS, vol. 11764, pp. 604–612. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-32239-7_67 Wasserthal, J., et al.: TotalSegmentator: robust segmentation of 104 anatomic structures in CT images. Radiol Artif Intell 5, e230024 (2023) Lindholz, M., et al.: Analyzing the TotalSegmentator for facial feature removal in head CT scans. Radiography 31, 372–378 (2025) Liu, C., Yu, X., Wang, D., Jiang, T.: ACLNet: a deep learning model for ACL rupture classification combined with bone morphology. In: International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 57-67. Springer, Cham (2024) Pan, H., et al.: Large-scale uterine myoma MRI dataset covering all FIGO types with pixel-level annotations. Sci. Data 11, 410 (2024) Isensee, F., Jaeger, P.F., Kohl, S.A., Petersen, J., Maier-Hein, K.H.: nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat. Methods 18, 203–211 (2021) Microsoft, MRI Autoencoder v0.1, Hugging Face Model Hub. https://huggingface.co/microsoft/mri-autoencoder-v0.1. Accessed 27 Feb 2025 Zbontar, J., et al.: fastMRI: an open dataset and benchmarks for accelerated MRI. arXiv preprint arXiv:1811.08839 (2018) Podell, D., et al.: Sdxl: improving latent diffusion models for high-resolution image synthesis. arXiv preprint arXiv:2307.01952 (2023) Kingma, D.P., Welling, M.: Auto-encoding variational bayes. Banff, Canada (2013) Kingma, D.P., Salimans, T., Welling, M.: Variational dropout and the local reparameterization trick. In: Advances in Neural Information Processing Systems, vol. 28 (2015) Ho, J., Jain, A., Abbeel, P.: Denoising diffusion probabilistic models. Adv. Neural. Inf. Process. Syst. 33, 6840–6851 (2020) Akiba, T., Sano, S., Yanase, T., Ohta, T., Koyama, M.: Optuna: a next-generation hyperparameter optimization framework. In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp. 2623–2631 (2019) McInnes, L., Healy, J., Melville, J.: UMAP: uniform manifold approximation and projection for dimension reduction. arXiv preprint arXiv:1802.03426 (2020) Acknowledgments This study was funded by “NUM 2.0” (FKZ: 01KX2121). Author information Authors and Affiliations Corresponding author Editor information Editors and Affiliations Ethics declarations Disclosure of Interests M.L. receives funding from the Berlin Institute of Health (Junior Digital Clinician Scientist Grant). C.A.H. receives funding from the Berlin Institute of Health (Digital Clinician Scientist Grant) and Siemens Healthineers and previously received seed funding from the University Medicine Greifswald and speaker fees from Siemens Healthineers, Bayer AG, Koelis AG, and Bracco Imaging Deutschland outside the submitted work. T.P. also receives funding from the Berlin Institute of Health (Advanced Clinician Scientist Grant, Platform Grant), the Ministry of Education and Research (BMBF, 01KX2121 (“NUM 2.0”, RACOON), 68GX21001A, 01ZZ2315D), the German Research Foundation (DFG, SFB 1340/2), and the European Union (H2020, CHAIMELEON: 952172, DIGITAL, EUCAIM:101100633). Furthermore, T.P. declares relationships with the following companies: research agreements (no personal payments) with AGO, Aprea AB, AR-CAGY-GINECO, Astellas Pharma Global Inc. (APGD), Astra Zeneca, Clovis Oncology, Inc., Holaira, Incyte Corporation, Karyopharm, Lion Biotechnologies, Inc., MedImmune, Merck Sharp & Dohme Corp, Millennium Pharmaceuticals, Inc., Morphotec Inc., NovoCure Ltd., PharmaMar S.A. and PharmaMar USA, Inc., Roche, Siemens Healthineers, and TESARO Inc.; fees for a book translation (Elsevier B.V.); and fees for speaking engagements (Bayer Healthcare). Rights and permissions Copyright information © 2026 The Author(s), under exclusive license to Springer Nature Switzerland AG About this paper Cite this paper Ruppel, R. et al. (2026). UteroVAE: A Shape-Informed Variational Autoencoder for Uterine MRI Encoding in Adenomyosis, Fibroids, and Healthy Uteri. In: Celebi, M.E., et al. Skin Image Analysis, and Computer-Aided Pelvic Imaging for Female Health. DGM4MICCAI 2025. Lecture Notes in Computer Science, vol 16149. Springer, Cham. https://doi.org/10.1007/978-3-032-05825-6_7 Download citation DOI: https://doi.org/10.1007/978-3-032-05825-6_7 Published: Publisher Name: Springer, Cham Print ISBN: 978-3-032-05824-9 Online ISBN: 978-3-032-05825-6 eBook Packages: Computer ScienceComputer Science (R0)Springer Nature Proceedings Computer Science

Text is read by the "Ask this paper" AI Q&A widget below. Extraction quality varies by source — PMC NXML preserves structure cleanly, OA-HTML may include some navigation residue, and OA-PDF can have broken hyphenation. The publisher copy (via DOI) is the canonical version.

My notes (saved in your browser only)

Ask this paper AI returns verbatim quotes from the full text · source: oa-doi-fallback

Answers must be backed by verbatim quotes from this paper's full text. Hallucinated quotes are dropped automatically; if no verbatim passage answers the question, we say so. How this works

Condition tags

adenomyosis

Citation neighborhood (sparse)

Too few in-corpus citations on either side for a chart; here are the lists.

Cites (2)

References (13)

Source provenance

openalex
last seen: 2026-06-04T00:00:01.174412+00:00
License: CC0 · commercial use OK