A Review of Diagnostic Strategies for Pulmonary Embolism Prediction in Computed Tomography Pulmonary Angiograms

Abstract

Pulmonary Embolism (PE) occurs when blood clots travel to the lungs from different parts of the body. It is amongst the most lethal cardio-respiratory diseases after stroke and heart attack. It occurs due to injury or inactivity due to Deep Vein Thrombosis (DVT). Over the last decade, the PE mortality rate has increased by 23%. Moreover, Vein Thromboembolism has been one of the leading causes of mortality among hospitalized COVID-19 patients. As a result, the necessity for early pulmonary embolism identification has immense significance in saving human lives. Computed Tomography Pulmonary Angiograms (CTPA) are the optimal medical imaging technique for diagnosing pulmonary embolism because of their superior sensitivity and specificity. The ability to distinguish between malignant and benign lungs using CTPA is critical for the early identification of the disease. The field of medical imaging analysis has significantly advanced due to the use of Machine Learning (ML) and Deep Learning (DL) techniques, notably Convolutional Neural Networks (CNN), for automated disease detection. These Computer-Aided Diagnosis (CAD) systems assist healthcare workers in rapid and knowledgeable decision-making, improving patient outcomes globally. This review paper offers a systematic study of current improvements in identifying PE by medical image processing, motivated by an extensive overview. This study attempts to bridge the gap between research and practice by providing a broad framework that covers both baseline and state-of-the-art approaches. It is a valuable resource for researchers and medical practitioners by providing insights into the effective utilization of advanced techniques at each stage of medical image analysis.