Detecting the correct image in diagnosis of hip displacement using machine learning

Developmental hip dysplasia is an orthopedic disorder that occurs during infancy and early development. It has easy and painless treatment methods if diagnosed early. However, if the diagnosis is delayed, difficult, painful and costly treatments such as surgery are required. In addition, delay in diagnosis causes permanent orthopedic disorders in patients. This study aims to extract accurate image detection, which is the first step in the diagnosis of developmental hip dysplasia, from ultrasound recordings with machine learning algorithms. The aim of the study is to increase the rate of early diagnosis, thus to provide treatment with easier and lower-cost methods, and to eliminate permanent orthopedic disorders. In the study, the features of the images captured from ultrasound recordings were obtained with the LBP method and pre-trained VGG16, VGG19, InceptionV3, MobileNet, DenseNet12