One-shot neuroanatomy segmentation through online data augmentation and confidence aware pseudo label

Liutong Zhang, Guochen Ning, Hanying Liang, Boxuan Han, Hongen Liao

Abstract: Highlights•Propose an end-to-end network for one-shot neuroanatomy segmentation.•Develop online data augmentation to generate diverse labeled image pairs.•Introduce confidence aware pseudo label to leverage the unlabeled images.•Experimental results show our network consistently outperforms other methods.