{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,29]],"date-time":"2026-07-29T14:58:55Z","timestamp":1785337135184,"version":"3.55.0"},"reference-count":34,"publisher":"MDPI AG","issue":"15","license":[{"start":{"date-parts":[[2026,7,29]],"date-time":"2026-07-29T00:00:00Z","timestamp":1785283200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Diagnostics"],"abstract":"<jats:p>Background: Medical imaging datasets often suffer from class imbalance and limited availability of pathology-rich cases, which limit the performance of machine learning models for segmentation, classification, and vision\u2013language tasks. We propose POWDR, a pathology-preserving outpainting framework for 3D MRI that uses real pathological regions as conditioning evidence while generating anatomically plausible surrounding tissue. Methods: Our approach leverages wavelet-domain conditioning to enhance high-frequency detail and mitigate blurring common in latent diffusion models. We introduce a random connected mask training strategy to reduce conditioning-induced collapse and improve diversity outside the lesion. POWDR is evaluated on brain MRI using BraTS datasets and extended to knee MRI to assess applicability beyond brain imaging. Results: Quantitative metrics (FID, MS-SSIM, LPIPS) were used to assess image realism. Random connected mask training improved diversity, reducing cosine similarity from 0.9947 to 0.9580 and increasing KL divergence from 0.00026 to 0.01494. To validate pathology preservation, we compared lesion overlap, volume, intensity, and morphology. For downstream segmentation, nnU-Net performance improved from 0.6992 to 0.7137 Dice after augmentation with 50 synthetic cases, representing a modest but statistically significant improvement (paired t-test, p = 0.016). Tissue volume analysis showed no significant differences for CSF and GM compared to real images, while WM volume was lower in synthetic images. Conclusions: POWDR provides a framework for generating diverse, pathology-preserving synthetic MRI data. The results suggest potential utility for data augmentation while maintaining clinically relevant lesion characteristics.<\/jats:p>","DOI":"10.3390\/diagnostics16152385","type":"journal-article","created":{"date-parts":[[2026,7,29]],"date-time":"2026-07-29T13:53:31Z","timestamp":1785333211000},"page":"2385","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["POWDR: Pathology-Preserving Outpainting with Wavelet Diffusion for 3D MRI"],"prefix":"10.3390","volume":"16","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-2423-631X","authenticated-orcid":false,"given":"Fei","family":"Tan","sequence":"first","affiliation":[{"name":"GE HealthCare, San Ramon, CA 94583, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ashok Vardhan","family":"Addala","sequence":"additional","affiliation":[{"name":"GE HealthCare, San Ramon, CA 94583, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8890-7963","authenticated-orcid":false,"given":"Bruno Astuto Arouche","family":"Nunes","sequence":"additional","affiliation":[{"name":"GE HealthCare, San Ramon, CA 94583, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5537-5542","authenticated-orcid":false,"given":"Xucheng","family":"Zhu","sequence":"additional","affiliation":[{"name":"GE HealthCare, Menlo Park, CA 94025, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ravi","family":"Soni","sequence":"additional","affiliation":[{"name":"GE HealthCare, San Ramon, CA 94583, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,7,29]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","unstructured":"Khader, F., Mueller-Franzes, G., Arasteh, S.T., Han, T., Haarburger, C., Schulze-Hagen, M., Schad, P., Engelhardt, S., Baessler, B., and Foersch, S. (2023). Medical Diffusion: Denoising Diffusion Probabilistic Models for 3D Medical Image Generation. arXiv.","DOI":"10.1038\/s41598-023-34341-2"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"458","DOI":"10.1038\/s41551-025-01468-8","article-title":"Unconditional latent diffusion models memorize patient imaging data","volume":"10","author":"Dar","year":"2025","journal-title":"Nat. Biomed. Eng."},{"key":"ref_3","unstructured":"Kofler, F., Meissen, F., Steinbauer, F., Graf, R., Ehrlich, S.K., Reinke, A., Oswald, E., Waldmannstetter, D., Hoelzl, F., and Horvath, I. (2024). The Brain Tumor Segmentation (BraTS) Challenge: Local Synthesis of Healthy Brain Tissue via Inpainting. arXiv."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"1673","DOI":"10.1038\/s41598-020-80930-w","article-title":"Deep learning-Based 3D inpainting of brain MR images","volume":"11","author":"Kang","year":"2021","journal-title":"Sci. Rep."},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"Rombach, R., Blattmann, A., Lorenz, D., Esser, P., and Ommer, B. (2022). High-Resolution Image Synthesis with Latent Diffusion Models. arXiv.","DOI":"10.1109\/CVPR52688.2022.01042"},{"key":"ref_6","doi-asserted-by":"crossref","unstructured":"Lozupone, G., Bria, A., Fontanella, F., Meijer, F.J.A., Stefano, C.D., and Huisman, H. (2025). Latent Diffusion Autoencoders: Toward Efficient and Meaningful Unsupervised Representation Learning in Medical Imaging. arXiv.","DOI":"10.1016\/j.media.2026.103932"},{"key":"ref_7","unstructured":"Falck, F., Pandeva, T., Zahirnia, K., Lawrence, R., Turner, R., Meeds, E., Zazo, J., and Karmalkar, S. (2025). A Fourier Space Perspective on Diffusion Models. arXiv."},{"key":"ref_8","unstructured":"Gerdes, M., Welling, M., and Cheng, M.C.N. (2024). GUD: Generation with Unified Diffusion. arXiv."},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Friedrich, P., Wolleb, J., Bieder, F., Durrer, A., and Cattin, P.C. (2024, January 9). WDM: 3D Wavelet Diffusion Models for High-Resolution Medical Image Synthesis. Proceedings of the DGM4MICCAI 2024, Marrakesh, Morocco.","DOI":"10.1007\/978-3-031-72744-3_2"},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Durrer, A., Wolleb, J., Bieder, F., Friedrich, P., Melie-Garcia, L., Ocampo-Pineda, M., Bercea, C.I., Hamamci, I.E., Wiestler, B., and Piraud, M. (2024). Denoising Diffusion Models for 3D Healthy Brain Tissue Inpainting. arXiv.","DOI":"10.1007\/978-3-031-72744-3_9"},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Durrer, A., Bieder, F., Friedrich, P., Menze, B., Cattin, P.C., and Kofler, F. (2025). fastWDM3D: Fast and Accurate 3D Healthy Tissue Inpainting. arXiv.","DOI":"10.1007\/978-3-032-05472-2_17"},{"key":"ref_12","unstructured":"Ferreira, A., Luijten, G., Puladi, B., Kleesiek, J., Alves, V., and Egger, J. (2024). Brain Tumour Removing and Missing Modality Generation using 3D WDM. arXiv."},{"key":"ref_13","doi-asserted-by":"crossref","unstructured":"Friedrich, P., Durrer, A., Wolleb, J., and Cattin, P.C. (2024). cWDM: Conditional Wavelet Diffusion Models for Cross-Modality 3D Medical Image Synthesis. arXiv.","DOI":"10.1007\/978-3-031-72744-3_2"},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"25","DOI":"10.1016\/S0045-7906(01)00011-8","article-title":"The Haar wavelet transform: Its status and achievements","volume":"29","author":"Falkowski","year":"2003","journal-title":"Comput. Electr. Eng."},{"key":"ref_15","doi-asserted-by":"crossref","unstructured":"Jha, D., Smedsrud, P.H., Riegler, M.A., Johansen, D., Lange, T.D., Halvorsen, P., and D. Johansen, H. (2019, January 9\u201311). ResUNet++: An Advanced Architecture for Medical Image Segmentation. Proceedings of the 2019 IEEE International Symposium on Multimedia (ISM), San Diego, CA, USA.","DOI":"10.1109\/ISM46123.2019.00049"},{"key":"ref_16","unstructured":"Loshchilov, I., and Hutter, F. (2019). Decoupled Weight Decay Regularization. arXiv."},{"key":"ref_17","unstructured":"Ho, J., Jain, A., and Abbeel, P. (2020). Denoising Diffusion Probabilistic Models. arXiv."},{"key":"ref_18","unstructured":"Kingma, D.P., and Welling, M. (2022). Auto-Encoding Variational Bayes. arXiv."},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Esser, P., Rombach, R., and Ommer, B. (2021). Taming Transformers for High-Resolution Image Synthesis. arXiv.","DOI":"10.1109\/CVPR46437.2021.01268"},{"key":"ref_20","doi-asserted-by":"crossref","unstructured":"Isola, P., Zhu, J.Y., Zhou, T., and Efros, A.A. (2018). Image-to-Image Translation with Conditional Adversarial Networks. arXiv.","DOI":"10.1109\/CVPR.2017.632"},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Ronneberger, O., Fischer, P., and Brox, T. (2015). U-Net: Convolutional Networks for Biomedical Image Segmentation. arXiv.","DOI":"10.1007\/978-3-319-24574-4_28"},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"799","DOI":"10.1038\/s42256-023-00652-2","article-title":"Federated benchmarking of medical artificial intelligence with MedPerf","volume":"5","author":"Karargyris","year":"2023","journal-title":"Nat. Mach. Intell."},{"key":"ref_23","unstructured":"Baid, U., Ghodasara, S., Mohan, S., Bilello, M., Calabrese, E., Colak, E., Farahani, K., Kalpathy-Cramer, J., Kitamura, F.C., and Pati, S. (2021). The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification. arXiv."},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"170117","DOI":"10.1038\/sdata.2017.117","article-title":"Advancing The Cancer Genome Atlas glioma MRI collections with expert segmentation labels and radiomic features","volume":"4","author":"Bakas","year":"2017","journal-title":"Sci. Data"},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"1993","DOI":"10.1109\/TMI.2014.2377694","article-title":"The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS)","volume":"34","author":"Menze","year":"2015","journal-title":"IEEE Trans. Med. Imaging"},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"R202","DOI":"10.1186\/ar3174","article-title":"Comparison of 3T MR scanners in regional cartilage-thickness analysis in osteoarthritis: A cross-sectional multicenter, multivendor study","volume":"12","author":"Balamoody","year":"2010","journal-title":"Arthritis Res. Ther."},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"e200078","DOI":"10.1148\/ryai.2021200078","article-title":"The International Workshop on Osteoarthritis Imaging Knee MRI Segmentation Challenge: A Multi-Institute Evaluation and Analysis Framework on a Standardized Dataset","volume":"3","author":"Desai","year":"2021","journal-title":"Radiol. Artif. Intell."},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"e200165","DOI":"10.1148\/ryai.2021200165","article-title":"Automatic Deep Learning\u2013assisted Detection and Grading of Abnormalities in Knee MRI Studies","volume":"3","author":"Astuto","year":"2021","journal-title":"Radiol. Artif. Intell."},{"key":"ref_29","unstructured":"Heusel, M., Ramsauer, H., Unterthiner, T., Nessler, B., and Hochreiter, S. (2018). GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium. arXiv."},{"key":"ref_30","unstructured":"Wang, Z., Simoncelli, E., and Bovik, A. (2003, January 9\u201312). Multiscale structural similarity for image quality assessment. Proceedings of the The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003, Pacific Grove, CA, USA."},{"key":"ref_31","doi-asserted-by":"crossref","unstructured":"Zhang, R., Isola, P., Efros, A.A., Shechtman, E., and Wang, O. (2018). The Unreasonable Effectiveness of Deep Features as a Perceptual Metric. arXiv.","DOI":"10.1109\/CVPR.2018.00068"},{"key":"ref_32","unstructured":"Chen, S., Ma, K., and Zheng, Y. (2019). Med3D: Transfer Learning for 3D Medical Image Analysis. arXiv."},{"key":"ref_33","doi-asserted-by":"crossref","first-page":"203","DOI":"10.1038\/s41592-020-01008-z","article-title":"nnU-Net: A self-configuring method for deep learning-based biomedical image segmentation","volume":"18","author":"Isensee","year":"2021","journal-title":"Nat. Methods"},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"45","DOI":"10.1109\/42.906424","article-title":"Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithm","volume":"20","author":"Zhang","year":"2001","journal-title":"IEEE Trans. Med. Imaging"}],"container-title":["Diagnostics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2075-4418\/16\/15\/2385\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,29]],"date-time":"2026-07-29T14:00:29Z","timestamp":1785333629000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2075-4418\/16\/15\/2385"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,29]]},"references-count":34,"journal-issue":{"issue":"15","published-online":{"date-parts":[[2026,8]]}},"alternative-id":["diagnostics16152385"],"URL":"https:\/\/doi.org\/10.3390\/diagnostics16152385","relation":{},"ISSN":["2075-4418"],"issn-type":[{"value":"2075-4418","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,29]]}}}