Artificial Intelligence in the Diagnosis of Diseases of the Respiratory System

Artificial intelligence (AI) arises from the desire of human beings to reproduce their intelligent behavior by means of computers. Recently, AI has been demonstrating remarkable success in medical image analysis owing to the rapid progress of deep learning algorithms, which have shown increasing power to solve complex real-world problems in computer vison and image analysis. The aim of this work is to introduce the main AI techniques for an interdisciplinary reader profile, specifically deep learning approaches applied nowadays in Radiology, for the intelligent diagnosis of lung cancer and respiratory diseases such as viral and bacterial pneumonia, tuberculosis and Covid-19. An overview of current techniques is presented, framed mainly in the area of deep learning and in particular convolutional neural networks. Reference to current reviews on techniques, applications, needs, software and databases is made. Then, the Deep Learning paradigm is introduced from its origins to then focus on CNNs, their main characteristics, aspects to be considered in training, representative architectures, implementation issues, including transfer learning. Finally, the software and hardware platforms widely used for the problem are described and then the concluding remarks are presented.