f-AnoGAN for prediction of anomalies in high-risk breast cancer women

Code for reproducing f-AnoGAN training and anomaly scoring presented in "Deep learning for predicting future lesion emergence in high-risk breast MRI screening: a feasibility study". This work extends f-AnoGAN: "f-AnoGAN: Fast Unsupervised Anomaly Detection with Generative Adversarial Networks" (accepted manuscript) for the application to dce-MRI in high-risk breast cancer women.

Referencing and citing f-AnoGAN

If you use (parts of) this code in your work please refer to this citations:

• Burger, B., Bernathova, M., Seeböck, P. et al. Deep learning for predicting future lesion emergence in high-risk breast MRI screening: a feasibility study. Eur Radiol Exp 7, 32 (2023). https://doi.org/10.1186/s41747-023-00343-y
• Schlegl, T., Seeböck, P., Waldstein, S.M., Langs, G., Schmidt-Erfurth, U., 2019. f-AnoGAN: Fast Unsupervised Anomaly Detection with Generative Adversarial Networks. Medical Image Analysis 54, 30-44. DOI: https://doi.org/10.1016/j.media.2019.01.010

Prerequisites

• Python (2.7), TensorFlow (1.2), NumPy, SciPy, Matplotlib
• A recent NVIDIA GPU

f-AnoGAN building blocks

• wgangp_64x64.py: Training a 64x64 WGAN architecture yields a trained generator (G) and discriminator (D). Modifies and extends (including tflib/) Ishaan Gulrajani's Tensorflow implementation of the WGAN-GP model proposed in "Improved Training of Wasserstein GANs" (GitHub).
• z_encoding_izif.py: Training the izi_f encoder (E) based on the trained WGAN model (G and D). Please refer to the full paper for more detailed information.
• new_anomaly_detection.py: Code for anomaly scoring utilizing the trained G, D, and E.

Setting (image) paths

Image paths are set in tflib/img_loader.py and tflib/new_img_loader.py. Images should be provided as "*.png" files structured in the following way:

• trainset_path = "path-to-folder-holding-normal-training-images"
• trainset_val_path = "path-to-folder-holding-normal-validation-images"
• test_normal_path = "path-to-folder-holding-normal-test-images"
• test_anom_path = "path-to-folder-holding-anom-test-images"

Additionally, the minimum and maximum gray value of each image need to be stored in mat.files corresponding to the images as variables "Min" and "Max":

• trainset_range_path= "path-to-folder-holding-ranges-of-normal-training-images"
• trainset_val_range_path = "path-to-folder-holding-ranges-of-normal-validation-images"
• test_normal_range_path = "path-to-folder-holding-ranges-of-normal-test-images"
• test_anom_range_path = "path-to-folder-holding-ranges-of-anom-test-images"

Please edit the files to specify the paths to your datasets.

Misc

Results of related research work are provided at CIR.