ARTIFICIAL INTELLIGENCE BASED PNEUMONIA DETECTION USING IMAGING MODALITIES: A REVIEW

Abstract

Pneumonia remains to be one of the leading causes of morbidity and mortality across the world. Timely and correct diagnosis is important for clinical treatment and to minimize the mortality. Medical imaging modalities play an important role in the detection of pneumonia, its severity and monitoring of treatment. This is a general review of the imaging technologies applied in detecting pneumonia covering the evolution of the early conventional methods and the modern advanced technologies into emerging technologies. The conventional modalities are explained in relation to the diagnostic performance, clinical utility, and limitations, including radiation exposure and access issues, i.e. chest X-ray (CXR), computed tomography (CT), the methods of ultrasound and their increasing place as bedside, radiation-free methods. The recent advances in the field of artificial intelligence (AI) and machine learning is to detect and classify pneumonia in different imaging modalities are reviewed in a systematic manner. This review will mainly concentrate on the use of artificial intelligence (AI) in the pneumonia detection using imaging. The application of machine learning (ML) and deep learning (DL) methods to chest X-ray and computer tomography images to obtain automatic classification of diseases and the evaluation of their performance is thoroughly reviewed. Moreover, the novel findings of AI-based pneumonia detection in thermal imaging are discussed, which is one of the radiation-free, non-contact, and fast screening modality. This review compares model, data sets, and reported performance metrics of imaging modalities and determines the challenges concerning data quality, generalizability, and clinical validation. Lastly, research directions in the future are described, which include the necessity to further examine the thermal imaging modality and multimodal AI systems to create safe, accessible, and clinically portable pneumonia diagnostic systems.