This project implements lung segmentation using a U-Net architecture with residual connections. Lung segmentation is a crucial step in medical image analysis, and this model helps identify the regions of interest within chest X-rays for various diagnostic purposes.
The lung segmentation dataset is available in the /kaggle/input/chest-xray-masks-and-labels/data/ directory. It includes lung images and their corresponding masks. The dataset was sourced from [source link or citation].
To get started with this project, follow these steps:
-
Clone this repository to your local machine:
git clone <https://github.com/Venkat-Kancherla-1/lung_segmentation_using_resunet.git>
segmentation-with-res-u-net.ipynb contains the code for data preprocessing, model creation, training, and testing. The test_on_image function allows you to visualize model predictions on test images. The get_metrics function can be used to visualize training metrics.
The lung segmentation model achieved an impressive accuracy of 98% on the test dataset, making it a robust tool for identifying lung regions within chest X-rays. You can see the resultant images in Results directory
Python (>=3.6) TensorFlow (>=2.0) Numpy Pandas OpenCV Matplotlib
Contributions are welcome! If you'd like to contribute to this project, please follow these guidelines:
Fork the repository. Create a new branch for your feature or bug fix. Make your changes and commit them with clear, concise commit messages. Push your branch to your fork. Open a pull request to the master branch of this repository.
Here, I attached the kaggle link of my work https://www.kaggle.com/code/kancherlavenkat/segmentation-with-res-u-net