Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204458084> ?p ?o ?g. }
- W3204458084 endingPage "273" @default.
- W3204458084 startingPage "263" @default.
- W3204458084 abstract "The performance of deep neural networks typically increases with the number of training images. However, not all images have the same importance towards improved performance and robustness. In fetal brain MRI, abnormalities exacerbate the variability of the developing brain anatomy compared to non-pathological cases. A small number of abnormal cases, as is typically available in clinical datasets used for training, are unlikely to fairly represent the rich variability of abnormal developing brains. This leads machine learning systems trained by maximizing the average performance to be biased toward non-pathological cases. This problem was recently referred to as hidden stratification. To be suited for clinical use, automatic segmentation methods need to reliably achieve high-quality segmentation outcomes also for pathological cases. In this paper, we show that the state-of-the-art deep learning pipeline nnU-Net has difficulties to generalize to unseen abnormal cases. To mitigate this problem, we propose to train a deep neural network to minimize a percentile of the distribution of per-volume loss over the dataset. We show that this can be achieved by using Distributionally Robust Optimization (DRO). DRO automatically reweights the training samples with lower performance, encouraging nnU-Net to perform more consistently on all cases. We validated our approach using a dataset of 368 fetal brain T2w MRIs, including 124 MRIs of open spina bifida cases and 51 MRIs of cases with other severe abnormalities of brain development." @default.
- W3204458084 created "2021-10-11" @default.
- W3204458084 creator A5006331855 @default.
- W3204458084 creator A5028279799 @default.
- W3204458084 creator A5031248347 @default.
- W3204458084 creator A5031879901 @default.
- W3204458084 creator A5035099926 @default.
- W3204458084 creator A5035530787 @default.
- W3204458084 creator A5035865500 @default.
- W3204458084 creator A5045321937 @default.
- W3204458084 creator A5055496507 @default.
- W3204458084 creator A5060751261 @default.
- W3204458084 creator A5060814361 @default.
- W3204458084 creator A5079918230 @default.
- W3204458084 creator A5082106258 @default.
- W3204458084 creator A5082898801 @default.
- W3204458084 creator A5087020941 @default.
- W3204458084 creator A5091180731 @default.
- W3204458084 date "2021-01-01" @default.
- W3204458084 modified "2023-09-25" @default.
- W3204458084 title "Distributionally Robust Segmentation of Abnormal Fetal Brain 3D MRI" @default.
- W3204458084 cites W1987869189 @default.
- W3204458084 cites W2042587503 @default.
- W3204458084 cites W2589120351 @default.
- W3204458084 cites W2600904149 @default.
- W3204458084 cites W2907193617 @default.
- W3204458084 cites W2909968020 @default.
- W3204458084 cites W2949461491 @default.
- W3204458084 cites W2966060723 @default.
- W3204458084 cites W2979754683 @default.
- W3204458084 cites W2989249732 @default.
- W3204458084 cites W2995393401 @default.
- W3204458084 cites W3011721937 @default.
- W3204458084 cites W3098414035 @default.
- W3204458084 cites W3112701542 @default.
- W3204458084 cites W3182781195 @default.
- W3204458084 cites W3204378754 @default.
- W3204458084 doi "https://doi.org/10.1007/978-3-030-87735-4_25" @default.
- W3204458084 hasPublicationYear "2021" @default.
- W3204458084 type Work @default.
- W3204458084 sameAs 3204458084 @default.
- W3204458084 citedByCount "10" @default.
- W3204458084 countsByYear W32044580842021 @default.
- W3204458084 countsByYear W32044580842022 @default.
- W3204458084 countsByYear W32044580842023 @default.
- W3204458084 crossrefType "book-chapter" @default.
- W3204458084 hasAuthorship W3204458084A5006331855 @default.
- W3204458084 hasAuthorship W3204458084A5028279799 @default.
- W3204458084 hasAuthorship W3204458084A5031248347 @default.
- W3204458084 hasAuthorship W3204458084A5031879901 @default.
- W3204458084 hasAuthorship W3204458084A5035099926 @default.
- W3204458084 hasAuthorship W3204458084A5035530787 @default.
- W3204458084 hasAuthorship W3204458084A5035865500 @default.
- W3204458084 hasAuthorship W3204458084A5045321937 @default.
- W3204458084 hasAuthorship W3204458084A5055496507 @default.
- W3204458084 hasAuthorship W3204458084A5060751261 @default.
- W3204458084 hasAuthorship W3204458084A5060814361 @default.
- W3204458084 hasAuthorship W3204458084A5079918230 @default.
- W3204458084 hasAuthorship W3204458084A5082106258 @default.
- W3204458084 hasAuthorship W3204458084A5082898801 @default.
- W3204458084 hasAuthorship W3204458084A5087020941 @default.
- W3204458084 hasAuthorship W3204458084A5091180731 @default.
- W3204458084 hasBestOaLocation W32044580842 @default.
- W3204458084 hasConcept C104317684 @default.
- W3204458084 hasConcept C105795698 @default.
- W3204458084 hasConcept C108583219 @default.
- W3204458084 hasConcept C119857082 @default.
- W3204458084 hasConcept C122048520 @default.
- W3204458084 hasConcept C142724271 @default.
- W3204458084 hasConcept C153180895 @default.
- W3204458084 hasConcept C154945302 @default.
- W3204458084 hasConcept C185592680 @default.
- W3204458084 hasConcept C207886595 @default.
- W3204458084 hasConcept C33923547 @default.
- W3204458084 hasConcept C41008148 @default.
- W3204458084 hasConcept C55493867 @default.
- W3204458084 hasConcept C63479239 @default.
- W3204458084 hasConcept C71924100 @default.
- W3204458084 hasConcept C89600930 @default.
- W3204458084 hasConceptScore W3204458084C104317684 @default.
- W3204458084 hasConceptScore W3204458084C105795698 @default.
- W3204458084 hasConceptScore W3204458084C108583219 @default.
- W3204458084 hasConceptScore W3204458084C119857082 @default.
- W3204458084 hasConceptScore W3204458084C122048520 @default.
- W3204458084 hasConceptScore W3204458084C142724271 @default.
- W3204458084 hasConceptScore W3204458084C153180895 @default.
- W3204458084 hasConceptScore W3204458084C154945302 @default.
- W3204458084 hasConceptScore W3204458084C185592680 @default.
- W3204458084 hasConceptScore W3204458084C207886595 @default.
- W3204458084 hasConceptScore W3204458084C33923547 @default.
- W3204458084 hasConceptScore W3204458084C41008148 @default.
- W3204458084 hasConceptScore W3204458084C55493867 @default.
- W3204458084 hasConceptScore W3204458084C63479239 @default.
- W3204458084 hasConceptScore W3204458084C71924100 @default.
- W3204458084 hasConceptScore W3204458084C89600930 @default.
- W3204458084 hasLocation W32044580841 @default.
- W3204458084 hasLocation W32044580842 @default.
- W3204458084 hasOpenAccess W3204458084 @default.