racheal collapse is a progressive disease in which the tracheal diameter decreases due to the degeneration of the cartilaginous rings that make up the trachea, causing changes in tracheal shape and compliance, impairing breathing. The diagnosis is based on anamnesis, physical examination and complementary tests, with radiographic examination being the most accessible and used in clinical routine. It is recommended to use two radiographic projections to measure the tracheal diameter, a simple radiograph of the cervical region in the laterolateral projection and the same positioning with compression of the ventral region of the cervical trachea. Thus, the present study aims to evaluate the feasibility of not using the radiography of the cervical region with the ventral compression of the trachea, aiming to reduce the emission of ionizing radiation, the reduction of the animal's stress to obtain the compressive radiography and, consequently, the reduction of costs. and the time of diagnosis. For this, artificial intelligence will be used through convolutional neural networks applied to data from simple radiographic images of the cervical region in the laterolateral projection of dogs with a positive previous diagnosis for tracheal collapse and dogs without tracheal collapse (control group), in addition to machine learning techniques on basic patient data. Data will be collected from the archive of radiographic images of the Diagnostic Imaging sector of the Veterinary Hospital Luiz Quintiliano de Oliveira, Faculty of Veterinary Medicine, Universidade Estadual Paulista - UNESP, campus of Araçatuba-SP.
News published in Agência FAPESP Newsletter about the scholarship: