X-Işını Görüntülerinin Giriş Seviyesinde Birleştirilmesi Kullanılarak Tibial Plato Kırığı Tespiti için Çift Görünümlü Derin Öğrenme
Year: 2026
Type: article
Abstract: Tibial plateau fractures are serious injuries of the knee joint. If they are not diagnosed and treated on time, they can lead to important functional problems. Therefore, accurate detection from X-ray images is very important. In this study, a deep learning-based method is proposed for detecting tibial plateau fractures using dual-view X-ray images. Anterior–posterior and lateral images are combined at the input level to create a 6-channel representation. This simple design allows the model to learn from both views at the same time. In experiments, five pre-trained convolutional neural network models (ResNet18, ResNet50, DenseNet201, EfficientNet-B0, and MobileNetV2) are used and evaluated under the same conditions. The results show that all models perform well, and EfficientNet-B0 achieved the best balance between sensitivity (87.23%) and specificity (94.23%) on the test set. In addition, Grad-CAM and t-SNE analyses are used to understand model decisions better. Overall, the proposed method provides a simple and effective approach for dual-view fracture detection and can support clinical diagnosis.
Source: DergiPark (Istanbul University)
Institution: Izmir Kâtip Çelebi University
Language: English
Topic: Medical Imaging and Analysis
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7166967088