Paper: View it like a radiologist: Shifted windows for deep learning augmentation of CT images (Link)

This repository contains an implementation of "Random windowing" – a novel preprocessing and augmentation scheme for CT images. Random windowing augments CT images during preprocessing, avoiding artifacts and leverage more of the data distribution compared to traditional augmentation schemes.

Random windowing is described in the 2025 paper "Random Window Augmentations for Deep Learning Robustness in CT and Liver Tumor Segmentation".

A preliminary version of Random windowing, called "Window shifting" was described in the MLSP 2023 paper "View it like a radiologist: Shifted windows for deep learning augmentation of CT images".