RADYOGRAFİ GÖRÜNTÜLERİ VE SINIFLANDIRMA ALGORİTMALARI KULLANILARAK OMUZ PROTEZLERİNİN ÜRETİCİLERİNİN BELİRLENMESİ

Abstract

Shoulder prostheses may need to be maintained or replaced over time for different reasons. These maintenance procedures are also performed by surgeries. There are different types of shoulder prostheses produced by different manufacturers, and different equipment is required to remove and care for each. In cases where sufficient information about the prosthesis type cannot be provided, some problems may be encountered. Visual examination and comparison of radiographic images by experts is both tiring and prolonged. In order to select the correct equipment and procedures before surgery, a fast and highly accurate solution is needed to assist the surgeon who will perform the operation in identifying unknown prostheses. In this study, 12 different classification algorithms were used to identify shoulder prostheses from 3 different manufacturers from radiographic images and the performances of these algorithms were compared. It has been observed that K-Nearest Neighbor algorithm performs better than other algorithms. It is thought that this algorithm will be the right choice for prosthesis recognition from radiography images and can be used to identify other prosthesis types.