MrBlankness / LightM-UNet

Pytorch implementation of "LightM-UNet: Mamba Assists in Lightweight UNet for Medical Image Segmentation"
https://arxiv.org/abs/2403.05246
Apache License 2.0
295 stars 26 forks source link

Clarification on Dataset Format: .nii vs .nii.gz #5

Open xaswq opened 7 months ago

xaswq commented 7 months ago

First of all, I would like to extend my deepest appreciation for your incredible work on LightM-UNet. It's truly remarkable and has greatly contributed to the community's progress.

As I was preparing to utilize the Dataset801_LiverCT for my work, I encountered a slight confusion regarding the dataset format. After downloading the dataset from https://www.kaggle.com/datasets/gauravduttakiit/3d-liver-and-liver-tumor-segmentation/data, I noticed that the images are in .nii format. However, in the project's documentation and tutorials, there seems to be a reference to .nii.gz files.

Could you please clarify the expected format for the dataset? If the dataset should indeed be in .nii.gz format, could you advise on whether I should compress the .nii files myself or if there's an alternative download link for the .nii.gz versions?

In addition, I noticed that the image files (CHNCXR.png) and label files (MCUCXR.png) in the Montgomery&Shenzhen dataset have different prefixes. Is this the correct correspondence between images and labels, or should they have matching filenames?

Could you please clarify how to ensure the correct image-label pairing when preparing this dataset for training?

Thank you very much for your time and assistance. image image

eclipse0922 commented 7 months ago

You can get detailed information from here As you can see in the document, you can specify the file extensions(nrrd nii nii.gz and etc) author just made an example with .nii.gz, you can use other format for your dataset and for my case, I use .nrrd

xaswq commented 7 months ago

thx