AI in the Identification of Lung Contusions Through Chest Radiological Examination in Blunt Thoracic Trauma

Overview

The observational study focuses on comparing the interpretation of chest radiological examinations performed using a computer-based system with the standard interpretation conducted by a radiologist. The "LUNIT" system serves as a tool designed to assist radiologists in detecting the 10 most common abnormalities visible on chest radiographs, with proven efficacy in large case series. The investigation addresses the need to evaluate lung injuries resulting from thoracic trauma, which are linked to a higher risk of complications requiring close monitoring to detect potential respiratory failure. The primary aim of the study is to assess the accuracy of the LUNIT system in interpreting chest radiographs for the identification of lung contusions compared to the standard radiologist-based interpretation.

The study is observational, cross-sectional, retrospective, and involves a post-marketing device (LUNIT INSIGHT CXR, CE-marked), with a non-profit framework. Patients presenting to the general Emergency Department (ED) of the IRCCS AOU of Bologna with blunt thoracic trauma, who underwent a chest HRCT within 48 hours of a standard chest X-ray, between June 1, 2014, and June 1, 2024, will be retrospectively included. Enrollment will be consecutive and based on discharge reports from the ED. Chest X-rays will be analyzed using LUNIT by two expert radiologists. The identification of contusions by LUNIT will be compared with the standard chest X-ray reading (based on reports issued at the time of ED presentation), using chest HRCT as the gold standard. No changes will be made to the routine diagnostic-therapeutic pathway of patients. The Radiology Unit involved will be the Radiology Department of IRCCS AOUBO, Policlinico di Sant'Orsola; activities performed by the radiologists will include image reading, interpretation, and reporting, as per standard clinical practice, both before and after the application of LUNIT. The identification of pulmonary contusions is crucial in the assessment of patients with blunt thoracic trauma, as it significantly impacts clinical management, often necessitating clinical observation due to the risk of respiratory failure and bacterial superinfection. Accurate identification of contusions on chest X-rays can reduce the need for HRCT in high-risk patients. The use of software like LUNIT could represent a valid alternative to HRCT for patients presenting to the ED with blunt thoracic trauma, potentially reducing ionizing radiation exposure and shortening ED management times. Patients presenting to the general ED of IRCCS AOU of Bologna with blunt thoracic trauma and undergoing HRCT within 48 hours of a chest X-ray between June 1, 2014, and June 1, 2024, will be retrospectively included.