Evaluation of computed tomographies of pulmonary emboli patients admitted to the emergency department with artificial intelligence

Objectives; Pulmonary embolism is an important cause of mortality and morbidity. The timely and accurate detection of patients with this pathology in emergency departments and other health units in our country is directly related to the mortality rates of patients. The main aim of this study was to improve the time to diagnosis of pulmonary embolism patients and the workforce of physicians working in the emergency department, and to minimize computed tomography pulmonary angiography (CTPA) evaluation errors caused by radiologists or emergency physicians. In our study, computed tomography pulmonary angiography images of healthy patients and patients with pulmonary embolism were evaluated by artificial intelligence programs. Introducing BTPA images to artificial intelligence and interpreting them by artificial intelligence will save time for emergency physicians, significantly reduce the workload of radiologists and minimize the patient's stay in the emergency department. Method; In our study, BTPA images taken at Inonu University Turgut Özal Medical Center between 2014 and 2024 and stored in the hospital information system SECTRA were retrospectively evaluated without exclusion criteria such as age, gender, and race. Tomography images were accessed through our hospital's imaging system (SECTRA). The accuracy criterion was based on the patients' tomography reports written by radiology physicians available in the hospital registration system or the verbal opinions of radiology physicians for images without available tomography reports. Findings; In this study, 232 CT scans were used, which were selected by excluding BTPA images with motion artifacts, inadequate timing of intravenous contrast administration, and suboptimal contrast filling of the pulmonary artery and its branches. Simultaneously, the available CT scans were divided into healthy (normal) images and images with pulmonary embolism. 150 CT scans with pulmonary embolism and 82 healthy CT scans were used in the study. Conclusion; With the data obtained, the ability of artificial intelligence programs to detect pulmonary embolism in computed tomography pulmonary angiography was found to be positive and we hope that the use of artificial intelligence will become widespread in today's practice. The current programs will make it easier for emergency physicians to reach the correct diagnosis, reduce the margin of error to the most reasonable level and contribute positively to the workload of radiology physicians. The data obtained in our study present results that support this situation. In the near future, with similar studies, the role of artificial intelligence will gradually increase in healthcare services and will become more frequent and more effective in today's practice. Key Words; Computed tomography pulmonary angiography, artificial intelligence, pulmonary embolism