Adversarial synthesis learning enables segmentation without target modality ground truth
Year: 2018
Type: conference-paper
Abstract: A lack of generalizability is one key limitation of deep learning based segmentation. Typically, one manually labels new training images when segmenting organs in different imaging modalities or segmenting abnormal organs from distinct disease cohorts. The manual efforts can be alleviated if one is able to reuse manual labels from one modality (e.g., MRI) to train a segmentation network for a new modality (e.g., CT). Previously, two stage methods have been proposed to use cycle generative adversarial networks (CycleGAN) to synthesize training images for a target modality. Then, these efforts trained a segmentation network independently using synthetic images. However, these two independent stages did not use the complementary information between synthesis and segmentation. Herein, we proposed a novel end-to-end synthesis and segmentation network (EssNet) to achieve the unpaired MRI to CT image synthesis and CT splenomegaly segmentation simultaneously without using manual labels on CT. The end-to-end EssNet achieved significantly higher median Dice similarity coefficient (0.9188) than the two stages strategy (0.8801), and even higher than canonical multi-atlas segmentation (0.9125) and ResNet method (0.9107), which used the CT manual labels.
Institutions: Vanderbilt University, Incyte (United States)
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funders: National Science Foundation, Vanderbilt University, Vanderbilt Institute for Clinical and Translational Research, Incyte, National Institutes of Health, National Eye Institute, National Institute of Neurological Disorders and Stroke, National Institute of Biomedical Imaging and Bioengineering, National Center for Research Resources, National Center for Advancing Translational Sciences
Awards: Resource Development for the Java Image Science Toolkit, G1182793083, CAREER: Modeling Personalized Brain Development with Big Data, G2116968943, GPU-Accelerated Parallel Computer for Drug Discovery Applications, Computer-Assited Functional Neurosurgery, G2265622010, G2537422520, G2591101265, G3066754252, An Interactive Web-Based Game for Collaborative Labeling of Medical Images, G4011179826, G4107354744, G4418762578, G4425699080, Computer-Assisted functional neurosurgery, G5090652903, G5679716143, G5868276417, G64116031, G648101241, The Vanderbilt Institute for Clinical and Translational Research (VICTR) UL1, G6671297155, G6704072150, G7307553525, G7358180766, G7653134492, The Vanderbilt Institute for Clinical and Translational Research (VICTR), G8287001021, Quantitative Image Analysis Techniques for Optic Nerve Disease, G8668361270, G9000033427
OpenAlex ID: W2963105382
