Classifies 6 HCC, gender, age, and RAF score based on frontal chest radiographs
Data directory structure set up as follows:
- raf_data
- train.csv
- test.csv
- data
Usage:
- Training: python train.py --data_dir <path_to_data> --checkpoint_dir <checkpoint_save_directory> --size <size_of_xrays> -age_norm <normalization_for_age> --raf_norm <normalization_for_raf> --lr <learning_rate> --epochs <number_of_epochs> --train_batch_size <batch_size_for_training> --test_batch_size <batch_size_for_testing> --num_workers <number_of_workers> --decay_start_epoch <epoch_to_start_lr_decay>
- Testing: python test.py --data_dir <path_to_data> --checkpoint_path <path_to_checkpoint> --out_path <test_csv_location> --size <size_of_xrays> --age_norm <normalization_for_age> --raf_norm <normalization_for_raf> --only_pred <if_no_csv_file_is_used> --calc_stats <get_specificity_and_sensitivity>
Model weights available upon request to [email protected] for non-commercial use, research purposes only. MIT license.