Bidirectional Meta-kronecker Factored Optimizer And Hausdorff Distance Loss For Few-shot Medical Image Segmentation

This article presents a bidirectional meta-Kronecker factored optimizer (BM-KFO) framework to quickly optimize semantic segmentation tasks using just a few magnetic resonance imaging (MRI) images as input. The model-agnostic approach can be implemented without altering network components and is capable of learning the learning process and meta-initial points while training on previously unseen data. Experiments demonstrate that BM-KFO with AHD-loss is suitable for general medical image segmentation applications and achieves superior performance compared to the baseline method in few-shot learning tasks.