Genetic Convolutional Neural Networks Approach for Disease Detection From Chest X-Ray Images

Advanced technologies for medical imaging, such as chest radiography (CXR), have shown the potential to accurately predict diseases through deep learning models. These models play a significant role in detecting life-threatening lung diseases. However, there are challenges associated with the similarities between the patterns and symptoms of these diseases. These similarities can lead to misinterpretations and fatal errors. In addition, DL models generate a large number of parameters. In this study, we propose an innovative approach to identifying lung diseases using CXR images employing a powerful Genetic Deep Convolutional Neural Network. This model combines a deep CNN model with a Genetic Algorithm (GA), allowing us to extract only the most suitable feature representations from the large-scale of DL model parameters while optimizing the model hyper-parameters simultaneously. The targeted dataset is the COVID-19 radiography dataset. The evaluation shows that our GA approach achieves an accuracy of 77.31%, 78.23%, and 78.87% for three different dataset samples, outperforming traditional methods by 0.51%, 1.56%, and 1.7% respectively. Our findings highlight the potential of combining CNN and GA to enhance lung disease detection. This offers promising implications for improved medical image analysis and diagnosis, showcasing promising prospects in this field.