Classification of the quality of canine and feline ventrodorsal and dorsoventral thoracic radiographs through machine learning. Veterinary Radiology & Ultrasound

Tahghighi, P., Appleby, R. B., Norena, N., Ukwatta, E., & Komeili, A. (2024). Classification of the quality of canine and feline ventrodorsal and dorsoventral thoracic radiographs through machine learning. Veterinary Radiology & Ultrasound. https://doi.org/10.1111/vru.13373

Thoracic radiographs are vital in diagnosing respiratory, cardiovascular, and neoplastic conditions in companion animals. This study introduces a machine learning-based model to classify the quality of canine and feline thoracic radiographs in ventrodorsal and dorsoventral views. The model addresses three aspects: collimation, positioning, and exposure, using a dataset of 899 radiographs, and achieves F1 and AUC scores of 91.33% and 91.10%, respectively, via fivefold cross-validation. The results suggest the potential for clinic implementation, which could enhance radiographic quality and reduce nondiagnostic imaging.