Gömülü kaninlerin lokalizasyon zorluğunu değerlendirmeye yönelik derin öğrenme modeli

The aim of this study is to examine if an artificial intelligence algorithm can be used for identifying the bucco-palatinal position of the maxillary impacted canine from the panoramic X-rays.

A total of 810 panoramic x-rays were obtained from the archive of University, Faculty of Dentistry, Department of Orthodontics. X-rays included cases with unilateral/bilaterally impacted canines. We used a Convolutional Neural Network (CNN) to forecast where the impacted canines crown would be situated. CNNs excel at classifying images as they can autonomously and flexibly grasp hierarchies of features, from input images. The implementation of the proposed deep learning model has been done using the Python programming language and libraries (TensorFlow and Numpy). The dataset that was used to train the proposed model categorized into buccal, middle, and palatinal positioned samples. These samples are mainly 2D panoramic X-rays combined with clinical information about the location of the canine. The expectation from the model is to determine the location of the crown of the maxillary-impacted canine. The model's prediction of the impacted canine's location was assessed in bucco-palatinal directions.

The proposed deep learning model predicts the position of the impacted canine in the buccal, middle, or palatinal position with 68% accuracy.

This multidisciplinary research study developed a deep learning model to automate the detection and positioning of impacted canines on panoramic dental X-rays. Lastly, further research is required to refine the model for clinical implementation and to explore its integration into routine orthodontic practice