Exploiting Shared Knowledge From Non-COVID Lesions for Annotation-Efficient COVID-19 CT Lung Infection Segmentation | IEEE Journals & Magazine | IEEE Xplore