Abstract
The research of uncertainty has shown great potential in the field of medical image processing. However, most of the research in the field of medical image is aimed at quantifying uncertainty. In this paper, we introduce an uncertainty supervised learning method. Specifically, we integrate the dropout variable reference and heterostatic noise model to estimate uncertainty and then guide super-resolution processing. Finally, we evaluate the enhancement effect of uncertainty supervised learning on super-resolution processing under the demonstration of DIQT model.
Publisher
Darcy & Roy Press Co. Ltd.