Abstract: Aiming at the problems of blurred edges and missed detection of small targets caused by the insufficient feature extraction ability of shallow encoders in the UNet model for medical image ...