Classification of different implant types from X-ray images using deep learning methods

This study focuses on using Deep Learning models to classify different types of implants in X-ray images. It was conducted using a dataset consisting of 597 images of 4 classes of shoulder implants and 2376 images of dental implants which are sourced from open data repositories. The classification task was performed using deep learning models such as InceptionResNetV2, ResNet152V2, Xception, and DenseNet201. Model training was involved with the usage of Adam and RMSprop optimizers with varying learning rates. Additionally, data augmentation and the application of the CLAHE filter were employed to enhance model performance. Results indicate that the Xception model outperformed others in both datasets. In the shoulder implant dataset, the model achieved a test accuracy of 0.85, F1-score of 0.85, and an AUC score of 0.95 while in the dental implant dataset, the test accuracy was 0.90, F1-score was 0.91, and AUC score was 0.97. This research underscores the utility and effectiveness of deep learning models in classifying different implant types in x-ray images. The results show that this study can be suggested as potential guide in clinical applications. Furthermore, the application of data augmentation and the CLAHE filter improves model performance.