AUTHOR=Huang Shu-Tien , Liu Liong-Rung , Chiu Hung-Wen , Huang Ming-Yuan , Tsai Ming-Feng 

TITLE=Deep convolutional neural network for rib fracture recognition on chest radiographs

JOURNAL=Frontiers in Medicine

VOLUME=Volume 10 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2023.1178798

DOI=10.3389/fmed.2023.1178798

ISSN=2296-858X

ABSTRACT=Rib fractures represent a prevalent injury among trauma patients and pose significant risks if not promptly and accurately diagnosed. Unfortunately, missed rib fractures are not uncommon, with approximately 50% escaping detection on radiographs, consequently leading to heightened morbidity and mortality rates. While more sensitive imaging modalities exist, their high cost and associated radiation exposure make them impractical for routine use. Point of care ultrasound offers an alternative but suffers from time-consuming procedures and reliance on operator expertise. Thus, the objective of this study was to leverage the potential of deep convolutional neural networks (DCNNs) in identifying rib fractures on chest radiographs. A comprehensive retrospective dataset of chest radiographs, accompanied by formal image reports documenting rib fractures, was collected from a single medical center over the last five years. The DCNN models were trained using 2000 region-of-interest (ROI) slices for each category, including fractured ribs, non-fractured ribs, and background regions. The images were segmented into pixel dimensions of 128 × 128 to facilitate training of the deep learning models (DLMs). The resulting trained models exhibited remarkable validation accuracies, with AlexNet achieving 92.6%, GoogLeNet achieving 92.2%, EfficientNetb3 achieving 92.3%, DenseNet201 achieving 92.4%, and MobileNetV2 achieving 91.2%. By integrating DCNN models capable of rib fracture recognition into clinical decision support systems, the incidence of missed rib fracture diagnoses can be significantly reduced, thereby leading to tangible decreases in morbidity and mortality rates among trauma patients. This innovative approach holds the potential to revolutionize the diagnosis and treatment of chest trauma, ultimately improving clinical outcomes for individuals affected by these injuries.