Automatic classification and grading of canine tracheal collapse on thoracic radiographs by using deep learning

• DOI: 10.1111/vru.13413
• Publication Date: 2024-07-16T12:30:24Z
• Authors: Hathaiphat Suksan...; Nan Choisunirachon; Teerawat Tongloy; Santhad Chuwongin; Siridech Boonsang; Veerayuth Kittichai; Chutimon Thanaboo...
• Keywords: Grading (engineering)

Tracheal collapse is a chronic and progressively worsening disease; the severity of clinical symptoms experienced by affected individuals depends on degree airway collapse. Cutting-edge automated tools are necessary to modernize disease screening using radiographs across various veterinary settings, such as animal clinics hospitals. This primarily due inherent challenges associated with interpreting uncertainties among veterinarians. In this study, an artificial intelligence model was developed screen canine tracheal archived lateral cervicothoracic radiographs. can differentiate between normal collapsed trachea, ranging from early severe degrees. The you-only-look-once (YOLO) models, including YOLO v3, v4, v4 tiny, were used train test data sets under in-house XXX platform. results showed that tiny-416 had satisfactory performance in grade 1-2 collapse, 3-4 98.30% sensitivity, 99.20% specificity, 98.90% accuracy. area curve precision-recall >0.8, which demonstrated high diagnostic intraobserver agreement deep learning radiologists κ = 0.975 (P < .001), all observers having excellent (κ 1.00, P .001). intraclass correlation coefficient >0.90, represented consistency. Therefore, be useful reliable method for effective classification based routine