This repository provides a wrapper around nnU-Net that enables interoperability with dataset CSVs generated by QTIM's preprocessing library. Its primary goal is to decouple nnU-Net model training and deplotment from its default dataset organizational assumptions by introducing custom data loading utilities that operate on CSV-based datasets rather than requiring a fixed directory structure.
This repository is intended primarily as an internal compatibility layer for QTIM. It is not designed to function as a fully standalone framework or replacement for nnU-Net. Most users should interact with this repository through a parent project rather than using it independently.