Genomic data science systems of Prediction and prevention of pneumonia from chest X-ray images using a two-channel dual-stream convolutional neural network

Pneumonia is one of the prime causes of death among children and old people around the world. It is an infection caused by a virus, bacteria, or other germs, which results in inflammation of the lungs and it can be life-threatening if not diagnosed on time, although it is a vaccine-preventable disease. Pneumonia can be identified by analyzing chest X-rays through very tedious, rigorous, and time-consuming procedures and also requires accurate precision. A better and advanced artificial intelligence system for pneumonia analysis and recognition is a vital step in the process of diagnosis, especially during this critical COVID-19 pandemic period. A robust, efficient, and highly accurate classification method is required to perform the classification of chest X-ray diseases due to an increase in the high rate of mortality cases. However, the favorable results of deep learning algorithms in inspecting and investigating medical images using convolutional neural networks (CNN) have gained much attention for disease classification because of their speed and levels of accuracy on image classification. This research, therefore, proposes a novel methodology of a CNN-based classification model that is capable of detecting whether an X-ray image contains pneumonia or not, without applying transfer learning or data augmentation. This method utilizes feeding of more than one input source of X-ray images into two separate CNNs simultaneously and optimizing the result using a conjugate gradient descent optimizer. Remarkable classification performance is achieved using the techniques of TCDS-CNN which outperformed other state-of-the-art models.