FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4297384217> ?p ?o ?g. }

• W4297384217 abstract "Abstract Magnetic resonance imaging (MRI) is the dominant diagnostic technique to non-invasively image the brain, and deep learning has proven a powerful tool for analyzing these images. However, machine learning models trained on such MRI data have empirically shown an ability to detect complex and invisible artifacts, such as which type of machine a scan was taken from to a high degree of accuracy. Such artifacts are potentially invisible to the human eye, but can be identified by machine learning systems, leading them to focus on irrelevant features rather than scientifically and/or medically useful ones. For example, machine learning systems can often “shortcut” past the actual features researchers would like to detect and utilize separate spurious correlations to make predictions. Several such undesired features have been reported to interfere with cross-institutional medical imaging deep learning research, and more are likely to be identified as time goes on. Here, we develop a method capable of removing these spurious correlations in an unsupervised manner, leveraging generative techniques to produce images which maintain image quality while learning how to remove technical artifacts. Generative Adversarial Networks are a class of deep learning architectures which have shown impressive efficacy in image generation and editing tasks, and our work builds upon this success. Here, we propose Generative Editing via Convolutional Obscuring (GECO), a Generative Adverserial Network for MRI deartifacting. GECO is based on a CycleGAN, a GAN architecture designed for image-to-image translation that is transforming an input image into a new image with one or more desirable properties. By formulating the CycleGAN loss as a two-player game with a regularization term and incentivizing the generator to erase spurious correlations the original image quality can be better preserved. Beginning with classifiers trained on original images to identify images based on artifacts of interest, GECO reduced the classifiers’ ability to detect these spurious correlations from 97% down to a difference which is nearly equal to a classifier making purely random guesses. We also observe over 98% structural similarity between the original and deartifacted images, indicating the preservation of the vast majority of non-spurious information contained in the original images. In addition to solving the known problem of avoiding artifacts from scanner type, this method opens the door to potentially removing many other types of spurious correlations from medical images and other data modalities across many fields. Software availability https://github.com/gevaertlab/GECO 1 ." @default.
• W4297384217 created "2022-09-28" @default.
• W4297384217 creator A5001226757 @default.
• W4297384217 creator A5002024060 @default.
• W4297384217 creator A5028102705 @default.
• W4297384217 creator A5029405899 @default.
• W4297384217 creator A5032901303 @default.
• W4297384217 creator A5038354247 @default.
• W4297384217 creator A5050826873 @default.
• W4297384217 creator A5065857785 @default.
• W4297384217 creator A5078274543 @default.
• W4297384217 creator A5084875483 @default.
• W4297384217 date "2022-09-27" @default.
• W4297384217 modified "2023-09-26" @default.
• W4297384217 title "Generative Editing via Convolutional Obscuring (GECO): A Generative Adversarial Network for MRI de-artifacting" @default.
• W4297384217 cites W1548374273 @default.
• W4297384217 cites W1901129140 @default.
• W4297384217 cites W1968426398 @default.
• W4297384217 cites W2011301426 @default.
• W4297384217 cites W2015159529 @default.
• W4297384217 cites W2083927153 @default.
• W4297384217 cites W2508310662 @default.
• W4297384217 cites W2557738935 @default.
• W4297384217 cites W2592929672 @default.
• W4297384217 cites W2811374795 @default.
• W4297384217 cites W2884436604 @default.
• W4297384217 cites W2912989244 @default.
• W4297384217 cites W2962914239 @default.
• W4297384217 cites W2963420272 @default.
• W4297384217 cites W2972006273 @default.
• W4297384217 cites W3099878876 @default.
• W4297384217 doi "https://doi.org/10.1101/2022.09.21.22280206" @default.
• W4297384217 hasPublicationYear "2022" @default.
• W4297384217 type Work @default.
• W4297384217 citedByCount "0" @default.
• W4297384217 crossrefType "posted-content" @default.
• W4297384217 hasAuthorship W4297384217A5001226757 @default.
• W4297384217 hasAuthorship W4297384217A5002024060 @default.
• W4297384217 hasAuthorship W4297384217A5028102705 @default.
• W4297384217 hasAuthorship W4297384217A5029405899 @default.
• W4297384217 hasAuthorship W4297384217A5032901303 @default.
• W4297384217 hasAuthorship W4297384217A5038354247 @default.
• W4297384217 hasAuthorship W4297384217A5050826873 @default.
• W4297384217 hasAuthorship W4297384217A5065857785 @default.
• W4297384217 hasAuthorship W4297384217A5078274543 @default.
• W4297384217 hasAuthorship W4297384217A5084875483 @default.
• W4297384217 hasBestOaLocation W42973842171 @default.
• W4297384217 hasConcept C108583219 @default.
• W4297384217 hasConcept C115961682 @default.
• W4297384217 hasConcept C119857082 @default.
• W4297384217 hasConcept C120665830 @default.
• W4297384217 hasConcept C121332964 @default.
• W4297384217 hasConcept C153180895 @default.
• W4297384217 hasConcept C154945302 @default.
• W4297384217 hasConcept C167966045 @default.
• W4297384217 hasConcept C192209626 @default.
• W4297384217 hasConcept C2779757391 @default.
• W4297384217 hasConcept C2988773926 @default.
• W4297384217 hasConcept C37736160 @default.
• W4297384217 hasConcept C39890363 @default.
• W4297384217 hasConcept C41008148 @default.
• W4297384217 hasConcept C81363708 @default.
• W4297384217 hasConcept C97256817 @default.
• W4297384217 hasConceptScore W4297384217C108583219 @default.
• W4297384217 hasConceptScore W4297384217C115961682 @default.
• W4297384217 hasConceptScore W4297384217C119857082 @default.
• W4297384217 hasConceptScore W4297384217C120665830 @default.
• W4297384217 hasConceptScore W4297384217C121332964 @default.
• W4297384217 hasConceptScore W4297384217C153180895 @default.
• W4297384217 hasConceptScore W4297384217C154945302 @default.
• W4297384217 hasConceptScore W4297384217C167966045 @default.
• W4297384217 hasConceptScore W4297384217C192209626 @default.
• W4297384217 hasConceptScore W4297384217C2779757391 @default.
• W4297384217 hasConceptScore W4297384217C2988773926 @default.
• W4297384217 hasConceptScore W4297384217C37736160 @default.
• W4297384217 hasConceptScore W4297384217C39890363 @default.
• W4297384217 hasConceptScore W4297384217C41008148 @default.
• W4297384217 hasConceptScore W4297384217C81363708 @default.
• W4297384217 hasConceptScore W4297384217C97256817 @default.
• W4297384217 hasLocation W42973842171 @default.
• W4297384217 hasOpenAccess W4297384217 @default.
• W4297384217 hasPrimaryLocation W42973842171 @default.
• W4297384217 hasRelatedWork W3009295188 @default.
• W4297384217 hasRelatedWork W3014413679 @default.
• W4297384217 hasRelatedWork W3015409886 @default.
• W4297384217 hasRelatedWork W3094011899 @default.
• W4297384217 hasRelatedWork W3178813832 @default.
• W4297384217 hasRelatedWork W4200633480 @default.
• W4297384217 hasRelatedWork W4297384217 @default.
• W4297384217 hasRelatedWork W4301431435 @default.
• W4297384217 hasRelatedWork W4312525628 @default.
• W4297384217 hasRelatedWork W4226271949 @default.
• W4297384217 isParatext "false" @default.
• W4297384217 isRetracted "false" @default.
• W4297384217 workType "article" @default.