An attention-enhanced cross-task network to analyse lung nodule attributes in CT images

Fu, X., Bi, L., Kumar, A., Fulham, M., & Kim, J. (2022). An attention-enhanced cross-task network to analyse lung nodule attributes in CT images. Pattern Recognition, 126, 108576. https://doi.org/10.1016/j.patcog.2022.108576

Implementation of the proposed deep learning model in Python and PyTorch.

Requirements

Python 3.7.8 to 3.8.8

PyTorch 1.5.0 to 1.10.2

Run pip install -r requirements.txt in your shell to install additional dependencies.

Data

CT Images:

• Nodules resized to 64x64 (see paper for details).
• .nii files containing all the slices for each nodule.
• For example: LIDC-IDRI-0001-1/ct_axial.nii

Labels:

• .csv file containing the ground truth attribute ratings for all nodules.
• File contains 10 columns. First column is nodule IDs. Subsequent columns are for the 9 attributes (subtlety, internal structure, calcification, sphericity, margin, lobulation, spiculation, texture, and malignancy).
• Nodule IDs are in the format LIDC-IDRI-0001-1, LIDC-IDRI-0002-1, etc.
• Ratings are normalised to [0, 1]

Cross-validation fold splits:

• .txt files each containing a list of nodule IDs for different cross-validation folds.
• For example for fold 1: 1_train.txt, and 1_test.txt.

Operation

Modify hyperparameters and locations of data files in the .json config file inside /configs.

Train the model using train.py.

Additional arguments for training:

• --config_file — path to config file
• --fold_id — cross-validation fold number
• --resume_epoch — None for train from scratch, or int number (e.g., 5) to resume training from saved model

Test the model using test.py.

Additional arguments for testing:

• --config_file — path to config file
• --fold_id — cross-validation fold number
• --test_epoch — which epoch to test (int number, -1 for latest saved model, or -2 for all saved models)