anwai98 / Brain-Tissue-Seg

Brain Tissue Segmentation on IBSR18 Dataset

Brain-Tissue-Seg

Brain Tissue Segmentation on MRI Images

Dataset used - IBSR18

How to play with the codes?

1. Use the Colab Notebooks for 2D Pipelines of UNet, ResUNet, DenseUNet, nnUNet likewise according to your utility/requirement. This will segment the three tissues of the Brain MRI Images - CSF, White Matter and Gray Matter.
2. Play with the straight forward code. You must remember to read the documentations of nnUNet (https://github.com/MIC-DKFZ/nnUNet) and versions of UNet (https://github.com/ashishpatel26/satellite-Image-Semantic-Segmentation-Unet-Tensorflow-keras) from these repositories.

Note : Comments are provided for explanation (almost). For more information, contact me!

Qualitative Results

Quantitative Results

Mean DSC = 0.931, Mean HD (Hausdorff Distance) = 15.67

Challenge Results (MAIA - 3rd Semester)

Our method had the best DSC Score on Test Dataset (Ranked 1st) in the Subject MISA - Medical Image Segmentation and Applications.