Classification of long bone fractures in dogs from X-ray images according to the time of occuration

Deep learning algorithms, which play an important role in the biomedical field, are one of the most popular topics in recent years. These algorithms provide successful results for many applications such as disease and fracture detection, biological data prediction, tissue and organ segmentation, and missing data completion using imaging methods. Fracture detection in bones is particularly one of the most researched topics in this field. However, while the majority of these applications are used in human medicine, veterinary medicine applications have been less studied. The lack of research in this field has been the main motivation for the thesis topic. Within the scope of this thesis study, the aim is to detect the presence of a fracture in the long bones in dogs using a dataset containing X-ray images of dogs, and to classify the fracture according to its time of occurrence if it exists. Like many studies in the field of the biomedical image processing, different deep learning architectures are compared, and the results are tried to be optimized.