Data augmentation approaches using cycle-consistent adversarial networks for improving COVID-19 screening in portable chest X-ray images

Abstract

The current COVID-19 pandemic, that has caused more than 100 million cases as well as more than two million deaths worldwide, demands the development of fast and accurate diagnostic methods despite the lack of available samples. This disease mainly affects the respiratory system of the patients and can lead to pneumonia and to severe cases of acute respiratory syndrome that result in the formation of several pathological structures in the lungs. These pathological structures can be explored taking advantage of chest X-ray imaging. As a recommendation for the health services, portable chest X-ray devices should be used instead of conventional fixed machinery, in order to prevent the spread of the pathogen. However, portable devices present several problems (specially those related with capture quality). Moreover, the subjectivity and the fatigue of the clinicians lead to a very difficult diagnostic process. To overcome that, computer-aided methodologies can be very useful even taking into account the lack of available samples that the COVID-19 affectation shows. In this work, we propose an improvement in the performance of COVID-19 screening, taking advantage of several cycle generative adversarial networks to generate useful and relevant synthetic images to solve the lack of COVID-19 samples, in the context of poor quality and low detail datasets obtained from portable devices. For validating this proposal for improved COVID-19 screening, several experiments were conducted. The results demonstrate that this data augmentation strategy improves the performance of a previous COVID-19 screening proposal, achieving an accuracy of 98.61% when distinguishing among NON-COVID-19 ( i.e. normal control samples and samples with pathologies others than COVID-19) and genuine COVID-19 samples. It is remarkable that this methodology can be extrapolated to other pulmonary pathologies and even other medical imaging domains to overcome the data scarcity.

Does imbalance in chest X-ray datasets produce biased deep learning approaches for COVID-19 screening?

“…We consider this configuration to be suitable enough for our sex and age study, as it has provided satisfactory results in similar works [15,14,12].…”

“…The performance of the presented computational approaches was evaluated by comparing the predictions provided by the models with the ground truth labels annotated in the X-ray image datasets. Then, the values of True Positives (TP), True Negatives (TN), False Positives (FP) and False Negatives (FN) were considered to calculate different metrics that are commonly used in the literature [15,14,12] to assess the stability of computational methods for medical imaging problems. Following the reference of these similar works, we also decided to use these metrics for our analysis of the sex and age factors.…”

“…However, we modified its original structure to support the binary output defined in our computational approaches, as depicted in Table 2. This architecture provided satisfactory results in similar works aimed at classifying chest X-rays of patients with COVID-19 [15,12,14], which led us to choose it for this work.…”

“…Waheed et al [13] addressed the lack of COVID-19 chest X-ray and they tried to solve this by developing CovidGAN, a model based on Auxiliary Classifier Generative Adversarial Network that generates synthetic COVID-19 images. In the work of Morís et al [14], the authors proposed a strategy to improve the performance of COVID-19 screening [15] by using 3 CycleGAN architectures to generate synthetic images from portable chest X-ray devices.…”

Does imbalance in chest X-ray datasets produce biased deep learning approaches for COVID-19 screening?

“…Waheed et al [ 18 ] addressed the lack of COVID-19 chest X-ray and they tried to solve this by developing CovidGAN, a model based on Auxiliary Classifier Generative Adversarial Network that generates synthetic COVID-19 images. In the work of Morís et al [ 19 ], the authors proposed a strategy to improve the performance of COVID-19 screening [ 20 ] by using 3 CycleGAN architectures to generate synthetic images from portable chest X-ray devices.…”