Use Case: Application of Deep Learning Methods in Diagnosis of Lung Diseases from Medical Imaging Data

Lung diseases are a broad group of illnesses and conditions that affect the lungs and impair their ability to function properly. It is estimated that more than 500 million people have chronic respiratory diseases and that they account for 7% of all deaths. The use of imaging tests is crucial for the accurate diagnosis, management, and treatment of various chest diseases as it gives clear visualization to any changes that occurred due to a lung disease presence. In order to help doctors in the interpretation of medical images, numerous computer-aided diagnosis systems have been developed, most of the recent ones being based on deep learning. The goal of this chapter is to show various deep learning approaches in diagnosis and detection of several lung diseases and lung conditions: lung cancer, COVID-19, tuberculosis, chronic obstructive pulmonary disease (COPD), and pneumothorax. Convolutional neural networks showed to be the most commonly used deep learning approach, especially pretrained models. Even though numerous research studies have been conducted on the use of deep learning for diagnosis of lung disease from medical imaging, direct comparisons between diagnoses from deep learning models and those from medical experts need to become more common practice in order to verify the generalizability and robustness of developed deep learning systems.