This project is about organ segmentation focused on liver. Segmenting liver will be help in extracting the liver component for further task.
Public Dataset: LITS 2017 data Competition Page - https://competitions.codalab.org/competitions/17094
Architecture used is 3d-Dense-U-net for liver segmentation
- The model was trtained on Tesla T-4 (16 GB)
- The Data was split into Training - 70%, Validation - 20% and testing - 10%
- Loss used for training the algorithm is dice loss.
- Metrics used for training and validation is dice score.
- The testing is performed using a pre-trained model
- The model is trained on LITS 2017 dataset
- Testing can be performed with the model using a 'python test.py'
- Finally, we can evalute the results using dice metrics for the testing data if the ground truth labels are available