Measuring the Impact of Data Augmentation Techniques in Lung Radiograph Classification Using a Fractional Factorial Design: A Covid-19 Case Study

Convolutional neural networks (CNNs) have become dominant in various computer vision tasks, obtaining state-of-the-art results in medical image analysis. Nevertheless, CNNs require large datasets to achieve high performance, which might not always be available in medical settings. Hence, different data augmentation strategies have been proposed to synthetically increase the size and diversity of a dataset. According to the state of the art, the relationship between data augmentation operations and the classification accuracy of a neural network has not been fully explored. In this work, the effect that basic augmentation techniques have in the detection of COVID-19 on chest X-Ray images is analyzed using a 2(7-1) fractional factorial experimental design. The experimental results show that zoom in and height shift operations have a significant positive effect on the accuracy, while horizontal flip operation hinders the performance. Moreover, by applying a cube plot analysis, the data augmentation operations and values that maximize the accuracy of the CNN are found. A 97% accuracy, 93% precision, and 97.7% recall scores are attained on a publicly available COVID-19 dataset using these data augmentation operations. © 2022 IEEE.

Analyzing the Effect of Basic Data Augmentation for COVID-19 Detection through a Fractional Factorial Experimental Design

The COVID-19 pandemic has created a worldwide healthcare crisis. Convolutional Neural Networks (CNNs) have recently been used with encouraging results to help detect COVID-19 from chest X-ray images. However, to generalize well to unseen data, CNNs require large labeled datasets. Due to the lack of publicly available COVID-19 datasets, most CNNs apply various data augmentation techniques during training. However, there has not been a thorough statistical analysis of how data augmentation operations affect classification performance for COVID-19 detection. In this study, a fractional factorial experimental design is used to examine the impact of basic augmentation methods on COVID-19 detection. The latter enables identifying which particular data augmentation techniques and interactions have a statistically significant impact on the classification performance, whether positively or negatively. Using the CoroNet architecture and two publicly available COVID-19 datasets, the most common basic augmentation methods in the literature are evaluated. The results of the experiments demonstrate that the methods of zoom, range, and height shift positively impact the model's accuracy in dataset 1. The performance of dataset 2 is unaffected by any of the data augmentation operations. Additionally, a new state-of-the-art performance is achieved on both datasets by training CoroNet with the ideal data augmentation values found using the experimental design. Specifically, in dataset 1, 97% accuracy, 93% precision, and 97.7% recall were attained, while in dataset 2, 97% accuracy, 97% precision, and 97.6% recall were achieved. These results indicate that analyzing the effects of data augmentations on a particular task and dataset is essential for the best performance. © 2023 by the authors. Licensee ESJ, Italy.