Pulmonary embolism, often referred to as PE, is a condition in which a blood clot becomes trapped in a pulmonary artery and prevents flow of blood to the lungs. If left ignored, this might be life-threatening and in most circumstances, fatal. Since the identification of whether a scan contains an embolus or not is a cumbersome process, we propose an approach using VGG16 and XGBoost to classify whether an image contains an embolus or not. The dataset used has been downloaded from Kaggle and segregated into two classes, namely “PE” (the images that contain embolus) and “no PE” (the images without any embolus in the lungs). Each directory contains over 1000 images. The methodology employed in this paper using VGG16 to extract features and XGBoost to further classify images rendered an accuracy of 97.59% and a sensitivity of 97.00% with 5 misclassifications.
CITATION STYLE
Dua, R., Ronald Wallace, G., Chotso, T., & Francis Densil Raj, V. (2023). Classifying Pulmonary Embolism Cases in Chest CT Scans Using VGG16 and XGBoost. In Lecture Notes on Data Engineering and Communications Technologies (Vol. 131, pp. 273–292). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-19-1844-5_22
Mendeley helps you to discover research relevant for your work.