PyTorch dataset loader for TDSC-ABUS 2023

[mralinp]

Hi everyone,

I've been working with the TDSC-ABUS 2023 dataset for my research on 3D breast ultrasound, and ended up building a small PyTorch package to make working with the dataset easier.

It handles downloading the dataset automatically and provides the 3D volumes, masks, labels, bounding boxes, and tumor crops through a standard PyTorch Dataset interface.

pip install tdsc-abus2023-pytorch

from tdsc_abus2023_pytorch import TDSC, DataSplits

dataset = TDSC("./data", split=DataSplits.TRAIN, download=True)
volume, mask, label, bbox = dataset[0]

GitHub: https://github.com/mralinp/tdsc-abus2023-pytorch
PyPI: https://pypi.org/project/tdsc-abus2023-pytorch/

I'm using it in my own ABUS research, but thought it might be useful to others working with this dataset too. Feedback and contributions are very welcome!