rochadiego/MultiResUNet-lung-segmentation

MultiResUNet implementation for lung segmentation using Tensorflow 2

Prediction

Documentation

Files

• Train.ipynb - Train the model
• Test.ipynb - Test the model, predict all images in a folder and save them
• Prediction.ipynb - Predict an single image, improve the quality and plot

Folders

Data

• data/cxr - Chest X-ray images to train
• data/masks - Chest X-ray masks to train
• data/test - Chest X-ray images for testing from same train dataset
• data/production - Chest X-ray images for testing from different dataset

Root

• models - Model file, weights and jacard index
• predictions - Predictions from Test.ipynb
• results - Self-evaluation during train

MultiResUNet

Rethinking the U-Net architecture for multimodal biomedical image segmentation

This repository contains the original implementation of "MultiResUNet : Rethinking the U-Net architecture for multimodal biomedical image segmentation" in Tensorflow 2.

Thanks https://github.com/nibtehaz/MultiResUNet

Paper

MultiResUNet has been published in Neural Networks

Ibtehaz, Nabil, and M. Sohel Rahman. "MultiResUNet: Rethinking the U-Net architecture for multimodal biomedical image segmentation." Neural Networks 121 (2020): 74-87.

• Read the Paper
• View the Preprint

License

MIT license