MammoGANesis: Controlled Generation of High-Resolution Mammograms for Radiology Education (2020-10-11T00:00:00.000000Z)

TL;DR

A generative adversarial network is trained to synthesize 512 x 512 high-resolution mammograms to distinguish between generated and real images, ascribing to the high visual quality of the synthesized and edited mammograms, and to their potential use in advancing and facilitating medical education.

Abstract

During their formative years, radiology trainees are required to interpret hundreds of mammograms per month, with the objective of becoming apt at discerning the subtle patterns differentiating benign from malignant lesions. Unfortunately, medico-legal and technical hurdles make it difficult to access and query medical images for training. In this paper we train a generative adversarial network (GAN) to synthesize 512 x 512 high-resolution mammograms. The resulting model leads to the unsupervised separation of high-level features (e.g. the standard mammography views and the nature of the breast lesions), with stochastic variation in the generated images (e.g. breast adipose tissue, calcification), enabling user-controlled global and local attribute-editing of the synthesized images. We demonstrate the model's ability to generate anatomically and medically relevant mammograms by achieving an average AUC of 0.54 in a double-blind study on four expert mammography radiologists to distinguish between generated and real images, ascribing to the high visual quality of the synthesized and edited mammograms, and to their potential use in advancing and facilitating medical education.

Authors

Cyril Zakka

1 papers

Ghida Saheb

1 papers

Elie Najem

1 papers

TL;DR

Abstract

Authors

References27 items

Artificial intelligence in medical imaging

Editing in Style: Uncovering the Local Semantics of GANs

GANSpace: Discovering Interpretable GAN Controls

Analyzing and Improving the Image Quality of StyleGAN

Data augmentation using generative adversarial networks (CycleGAN) to improve generalizability in CT segmentation tasks

Interpreting the Latent Space of GANs for Semantic Face Editing

MammoGAN: High-Resolution Synthesis of Realistic Mammograms

A Style-Based Generator Architecture for Generative Adversarial Networks

Towards generative adversarial networks as a new paradigm for radiology education

Generation of Synthetic Electronic Medical Record Text

Medical Image Synthesis for Data Augmentation and Anonymization using Generative Adversarial Networks

Deep Feature Factorization For Concept Discovery

Opportunities and challenges in developing deep learning models using electronic health records data: a systematic review

Opportunities in Machine Learning for Healthcare

Do GANs learn the distribution? Some Theory and Empirics

Generative Adversarial Networks: An Overview

Deep MR to CT Synthesis Using Unpaired Data

Opportunities and obstacles for deep learning in biology and medicine

Medical Image Synthesis with Context-Aware Generative Adversarial Networks

Invertible Conditional GANs for image editing

Generative Visual Manipulation on the Natural Image Manifold

De-identification of Medical Images with Retention of Scientific Research Value.

Conditional Generative Adversarial Nets

If You Don’t Find It Often, You Often Don’t Find It: Why Some Cancers Are Missed in Breast Cancer Screening

Machine learning and radiology

Mammo{GAN}: High-Resolution Synthesis of Realistic Mammograms in International Conference on Medical Imaging with Deep Learning – Extended Abstract Track (London, United Kingdom, July 2019)

When Radiologists Perform Best: The Learning Curve in Screening Mammogram Interpretation

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names