PE-DeepNet: A deep neural network model for pulmonary embolism detection

Abstract

Machine learning in medical image processing has shown to be a useful way for discovering patterns in both poorly labelled and unlabeled datasets. Venous thromboembolism, which includes deep vein thrombosis and pulmonary embolism, is a major cause of death in patients and requires quick detection by specialists. Using an artificial neural network, the suggested study was carried out to aid doctors in identifying and forecasting the risk level of pulmonary embolism in patients. This research presents a hybrid deep learning convolutional neural network model called PE-DeepNet (Pulmonary Embolism detection using Deep neural Network) to perform quick and accurate pulmonary embolism detection. The experiment uses a case study from the standard RSNA STR Pulmonary Embolism Chest CT scan image dataset. The proposed work results in an accuracy of 94.2%, an improvement over existing CNN models with minor trainable parameters.

Keywords

• Convolutional neural network
• Deep learning
• Medical imaging
• Machine learning
• Neural networks
• Thrombosis detection