Lifelong nnU-Net: a framework for standardized medical continual learning

An effective framework for continual image segmentation has the following requirements:

1. 1.

   It has all components for achieving high-quality static segmentation results and supports both two- and three-dimensional architectures (like the nnU-Net),

2. 2.

   Simplifies the evaluation of incremental domain scenarios by relying on widely accepted dataset formats and the alignment of label characteristics across…