Synthesis of COVID-19 Chest X-rays using Unpaired Image-to-Image Translation

TL;DR

This paper presents a novel method for generating high-quality synthetic COVID-19 chest X-ray images using unpaired image-to-image translation, improving detection performance and enabling data anonymization.

Contribution

The authors introduce an unsupervised domain adaptation approach for synthetic image generation and demonstrate its benefits for COVID-19 detection and data privacy.

Findings

Synthetic images improve COVID-19 detection accuracy.

Synthetic data achieves comparable detection performance for anonymization.

The dataset contains 21,295 high-fidelity synthetic COVID-19 X-ray images.

Abstract

Motivated by the lack of publicly available datasets of chest radiographs of positive patients with Coronavirus disease 2019 (COVID-19), we build the first-of-its-kind open dataset of synthetic COVID-19 chest X-ray images of high fidelity using an unsupervised domain adaptation approach by leveraging class conditioning and adversarial training. Our contributions are twofold. First, we show considerable performance improvements on COVID-19 detection using various deep learning architectures when employing synthetic images as additional training set. Second, we show how our image synthesis method can serve as a data anonymization tool by achieving comparable detection performance when trained only on synthetic data. In addition, the proposed data generation framework offers a viable solution to the COVID-19 detection in particular, and to medical image classification tasks in general. Our publicly available benchmark dataset consists of 21,295 synthetic COVID-19 chest X-ray images. The insights gleaned from this dataset can be used for preventive actions in the fight against the COVID-19 pandemic.