Standardizing CT Image using Generative Adversarial Network
Jin Chen, Ph.D.,
Associate Professor, Institute of Biomedical Informatics,
Department of Internal Medicine, Department of Computer Science,
University of Kentucky
CT acquisition parameter customization forms a barrier for large-scale CT image analysis, in that capturing CT images with non-standardized imaging protocols is common and it often results in significantly different radiomics features, even for the same patient. To overcome the barrier, we present a new generative adversarial network (GAN) model to learn from a large amount of training CT images and to generate synthetic CT images such that the differences in radiomic features due to using non-standardized imaging protocols are minimized.
Registration is required. Email email@example.com by Monday, June 11, 2018. If you require special physical arrangements to attend, please call 859-323-8545.