{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,6]],"date-time":"2026-06-06T16:48:55Z","timestamp":1780764535662,"version":"3.54.1"},"reference-count":22,"publisher":"Springer Science and Business Media LLC","issue":"9","license":[{"start":{"date-parts":[[2021,7,12]],"date-time":"2021-07-12T00:00:00Z","timestamp":1626048000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2021,7,12]],"date-time":"2021-07-12T00:00:00Z","timestamp":1626048000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"DOI":"10.13039\/501100006360","name":"Bundesministerium f\u00fcr Wirtschaft und Energie","doi-asserted-by":"publisher","award":["ZF4026303TS9"],"award-info":[{"award-number":["ZF4026303TS9"]}],"id":[{"id":"10.13039\/501100006360","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Int J CARS"],"published-print":{"date-parts":[[2021,9]]},"abstract":"<jats:title>Abstract<\/jats:title><jats:sec>\n                <jats:title>\n                           <jats:italic>Purpose<\/jats:italic>\n                        <\/jats:title>\n                <jats:p>Brain Magnetic Resonance Images (MRIs) are essential for the diagnosis of neurological diseases. Recently, deep learning methods for unsupervised anomaly detection (UAD) have been proposed for the analysis of brain MRI. These methods rely on healthy brain MRIs and eliminate the requirement of pixel-wise annotated data compared to supervised deep learning. While a wide range of methods for UAD have been proposed, these methods are mostly 2D and only learn from MRI slices, disregarding that brain lesions are inherently 3D and the spatial context of MRI volumes remains unexploited.\n<\/jats:p>\n              <\/jats:sec><jats:sec>\n                <jats:title>\n                           <jats:italic>Methods<\/jats:italic>\n                        <\/jats:title>\n                <jats:p>We investigate whether using increased spatial context by using MRI volumes combined with spatial erasing leads to improved unsupervised anomaly segmentation performance compared to learning from slices. We evaluate and compare 2D variational autoencoder (VAE) to their 3D counterpart, propose 3D input erasing, and systemically study the impact of the data set size on the performance.<\/jats:p>\n              <\/jats:sec><jats:sec>\n                <jats:title>\n                           <jats:italic>Results<\/jats:italic>\n                        <\/jats:title>\n                <jats:p>Using two publicly available segmentation data sets for evaluation, 3D VAEs outperform their 2D counterpart, highlighting the advantage of volumetric context. Also, our 3D erasing methods allow for further performance improvements. Our best performing 3D VAE with input erasing leads to an average DICE score of 31.40% compared to 25.76% for the 2D VAE.<\/jats:p>\n              <\/jats:sec><jats:sec>\n                <jats:title>\n                           <jats:italic>Conclusions<\/jats:italic>\n                        <\/jats:title>\n                <jats:p>We propose 3D deep learning methods for UAD in brain MRI combined with 3D erasing and demonstrate that 3D methods clearly outperform their 2D counterpart for anomaly segmentation. Also, our spatial erasing method allows for further performance improvements and reduces the requirement for large data sets.<\/jats:p>\n              <\/jats:sec>","DOI":"10.1007\/s11548-021-02451-9","type":"journal-article","created":{"date-parts":[[2021,7,12]],"date-time":"2021-07-12T06:02:26Z","timestamp":1626069746000},"page":"1413-1423","update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":25,"title":["Three-dimensional deep learning with spatial erasing for unsupervised anomaly segmentation in brain MRI"],"prefix":"10.1007","volume":"16","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-2229-9547","authenticated-orcid":false,"given":"Marcel","family":"Bengs","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Finn","family":"Behrendt","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Julia","family":"Kr\u00fcger","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Roland","family":"Opfer","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Alexander","family":"Schlaefer","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2021,7,12]]},"reference":[{"issue":"4","key":"2451_CR1","doi-asserted-by":"crossref","first-page":"449","DOI":"10.1007\/s10278-017-9983-4","volume":"30","author":"Z Akkus","year":"2017","unstructured":"Akkus Z, Galimzianova A, Hoogi A, Rubin DL, Erickson BJ (2017) Deep learning for brain MRI segmentation: state of the art and future directions. J Digit Imaging 30(4):449\u2013459","journal-title":"J Digit Imaging"},{"key":"2451_CR2","doi-asserted-by":"publisher","first-page":"170117","DOI":"10.1038\/sdata.2017.117","volume":"4","author":"S Bakas","year":"2017","unstructured":"Bakas S, Akbari H, Sotiras A, Bilello M, Rozycki M, Kirby JS, Freymann JB, Farahani K, Davatzikos C (2017) Advancing the cancer genome atlas glioma MRI collections with expert segmentation labels and radiomic features. Sci Data 4:170117","journal-title":"Sci Data"},{"key":"2451_CR3","unstructured":"Bakas S, Reyes M, Jakab A, Bauer S, Rempfler M, Crimi A et al (2018) Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the BRATS challenge. arXiv preprint arXiv:1811.02629"},{"key":"2451_CR4","doi-asserted-by":"crossref","unstructured":"Baur C, Denner S, Wiestler B, Navab N, Albarqouni S (2021) Autoencoders for unsupervised anomaly segmentation in brain MR images: a comparative study. Med Image Anal 101952","DOI":"10.1016\/j.media.2020.101952"},{"key":"2451_CR5","doi-asserted-by":"crossref","unstructured":"Baur C, Wiestler B, Albarqouni S, Navab N (2018) Deep autoencoding models for unsupervised anomaly segmentation in brain MR images. In: International MICCAI brainlesion workshop, pp 161\u2013169. Springer","DOI":"10.1007\/978-3-030-11723-8_16"},{"issue":"6","key":"2451_CR6","doi-asserted-by":"publisher","first-page":"1668","DOI":"10.1148\/rg.2015150023","volume":"35","author":"MA Bruno","year":"2015","unstructured":"Bruno MA, Walker EA, Abujudeh HH (2015) Understanding and confronting our mistakes: the epidemiology of error in radiology and strategies for error reduction. Radiographics 35(6):1668\u20131676","journal-title":"Radiographics"},{"key":"2451_CR7","unstructured":"Chen X, Konukoglu E (2018) Unsupervised detection of lesions in brain MRI using constrained adversarial auto-encoders. In: International conference on medical imaging with deep learning"},{"key":"2451_CR8","doi-asserted-by":"publisher","first-page":"101713","DOI":"10.1016\/j.media.2020.101713","volume":"64","author":"X Chen","year":"2020","unstructured":"Chen X, You S, Tezcan KC, Konukoglu E (2020) Unsupervised lesion detection via image restoration with a normative prior. Med Image Anal 64:101713","journal-title":"Med Image Anal"},{"key":"2451_CR9","unstructured":"De Vries T, Taylor GW (2017) Improved regularization of convolutional neural networks with cutout. arXiv preprint arXiv:1708.04552"},{"issue":"16","key":"2451_CR10","doi-asserted-by":"publisher","first-page":"4669","DOI":"10.1002\/hbm.24729","volume":"40","author":"KL Ito","year":"2019","unstructured":"Ito KL, Kim H, Liew SL (2019) A comparison of automated lesion segmentation approaches for chronic stroke T1-weighted MRI data. Hum Brain Mapp 40(16):4669\u20134685","journal-title":"Hum Brain Mapp"},{"key":"2451_CR11","doi-asserted-by":"crossref","unstructured":"Jiang Z, Ding C, Liu M, Tao D (2019) Two-stage cascaded u-net: 1st place solution to brats challenge 2019 segmentation task. In: International MICCAI brainlesion workshop, pp 231\u2013241. Springer","DOI":"10.1007\/978-3-030-46640-4_22"},{"key":"2451_CR12","unstructured":"Kingma DP, Welling M (2013) Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114"},{"key":"2451_CR13","doi-asserted-by":"publisher","first-page":"180011","DOI":"10.1038\/sdata.2018.11","volume":"5","author":"SL Liew","year":"2018","unstructured":"Liew SL, Anglin JM, Banks NW, Sondag M, Ito KL, Kim H, Chan J, Ito J, Jung C, Khoshab N, Lefebvre S, Nakamura W, Saldana D, Schmiesing A, Tran C, Vo D, Ard T, Heydari P, Kim B, Aziz-Zadeh L, Cramer S, Liu J, Soekadar S, Nordvik JE, Westlye L, Wang J, Winstein C, Yu C, Ai L, Koo B, Craddock R, Milham M, Lakich M, Pienta A, Stroud A (2018) A large, open source dataset of stroke anatomical brain images and manual lesion segmentations. Sci Data 5:180011","journal-title":"Sci Data"},{"issue":"2","key":"2451_CR14","doi-asserted-by":"publisher","first-page":"102","DOI":"10.1016\/j.zemedi.2018.11.002","volume":"29","author":"AS Lundervold","year":"2019","unstructured":"Lundervold AS, Lundervold A (2019) An overview of deep learning in medical imaging focusing on MRI. Z Med Phys 29(2):102\u2013127","journal-title":"Z Med Phys"},{"key":"2451_CR15","doi-asserted-by":"crossref","unstructured":"Menze BH, Jakab A, Bauer S, Kalpathy-Cramer J, Farahaniy K, Kirby J, Burren Y, Porz N, Slotboomy J, Wiest R, Lancziy L, Gerstnery E, Webery MA, Arbel T, Avants B, Ayache N, Buendia P, Collins L, Cordier N, Van Leemput K et al (2015) The multimodal brain tumor image segmentation benchmark (BRATS). IEEE Trans Med Imaging 34(10):1993\u20132024","DOI":"10.1109\/TMI.2014.2377694"},{"key":"2451_CR16","doi-asserted-by":"crossref","unstructured":"Pathak D, Krahenbuhl P, Donahue J, Darrell T, Efros AA (2016) Context encoders: feature learning by inpainting. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 2536\u20132544","DOI":"10.1109\/CVPR.2016.278"},{"key":"2451_CR17","doi-asserted-by":"crossref","unstructured":"Sato D, Hanaoka S, Nomura Y, Takenaga T, Miki S, Yoshikawa T, Hayashi N, Abe O (2018) A primitive study on unsupervised anomaly detection with an autoencoder in emergency head CT volumes. In: Medical imaging 2018: computer-aided diagnosis, vol 10575, p 105751P. International Society for Optics and Photonics","DOI":"10.1117\/12.2292276"},{"key":"2451_CR18","doi-asserted-by":"publisher","first-page":"30","DOI":"10.1016\/j.media.2019.01.010","volume":"54","author":"T Schlegl","year":"2019","unstructured":"Schlegl T, Seeb\u00f6ck P, Waldstein SM, Langs G, Schmidt-Erfurth U (2019) f-AnoGAN: fast unsupervised anomaly detection with generative adversarial networks. Med Image Anal 54:30\u201344","journal-title":"Med Image Anal"},{"issue":"1","key":"2451_CR19","first-page":"1929","volume":"15","author":"N Srivastava","year":"2014","unstructured":"Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R (2014) Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res 15(1):1929\u20131958","journal-title":"J Mach Learn Res"},{"key":"2451_CR20","unstructured":"Tan M, Le Q (2019) Efficientnet: rethinking model scaling for convolutional neural networks. In: International conference on machine learning, pp 6105\u20136114. PMLR"},{"issue":"18","key":"2451_CR21","doi-asserted-by":"publisher","first-page":"1821","DOI":"10.1056\/NEJMoa070972","volume":"357","author":"MW Vernooij","year":"2007","unstructured":"Vernooij MW, Ikram MA, Tanghe HL, Vincent AJ, Hofman A, Krestin GP, Niessen WJ, Breteler MM, van der Lugt A (2007) Incidental findings on brain MRI in the general population. N Engl J Med 357(18):1821\u20131828","journal-title":"N Engl J Med"},{"key":"2451_CR22","unstructured":"Zimmerer D, Kohl SA, Petersen J, Isensee F, Maier-Hein KH (2019) Context-encoding variational autoencoder for unsupervised anomaly detection. In: International conference on medical imaging with deep learning"}],"container-title":["International Journal of Computer Assisted Radiology and Surgery"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s11548-021-02451-9.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s11548-021-02451-9\/fulltext.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s11548-021-02451-9.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2021,8,10]],"date-time":"2021-08-10T12:22:08Z","timestamp":1628598128000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s11548-021-02451-9"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,7,12]]},"references-count":22,"journal-issue":{"issue":"9","published-print":{"date-parts":[[2021,9]]}},"alternative-id":["2451"],"URL":"https:\/\/doi.org\/10.1007\/s11548-021-02451-9","relation":{},"ISSN":["1861-6410","1861-6429"],"issn-type":[{"value":"1861-6410","type":"print"},{"value":"1861-6429","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,7,12]]},"assertion":[{"value":"13 January 2021","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"30 June 2021","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"12 July 2021","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"The authors declare that they have no conflict of interest.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Conflict of interest"}},{"value":"This work was conducted retrospectively on data from clinical routine which was completely anonymized. Ethical approval was therefore not required. Also this work relies on the BraTS 2019 and ATLAS data set. For use of these data sets, no ethics statements are necessary.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethical approval"}}]}}