Random Convolutions for Domain Generalization of Deep Learning-based Medical Image Segmentation Models.

Purpose To evaluate random convolutions as an augmentation strategy for improving domain generalization of deep learning-based segmentation models in medical imaging. Materials and Methods In this retrospective study, a random convolution-based augme…

Preview only. Read the full abstract at the source