Automatic Detection of Radiological Features to Fast Cephalometric Diagnosis: Preliminary Results of a Computer Vision Model

Mayhuasca R.; Huamanchumo L.; Rosales J. · Lecture Notes in Networks and Systems (2026) · DOI: 10.1007/978-3-032-12885-0_22

Resumen

Artificial intelligence techniques improve accuracy in imaging diagnosis. Human errors in radiological diagnosis can be avoided using deep learning and computer vision techniques. These automation tools add speed and reproducibility to the results delivered to users. This study aims to design a model with convolutional neural networks that automatically detects five anatomical areas in a cephalometric radiograph and evaluate the accuracy of the detection process. Among the structures of interest are the areas of the pituitary gland, nasal bones, symphysis, gonion, and the anterior maxillomandibular sector, important structures in diagnosis in dentistry and medicine. The database will consist of the five regions extracted from each of the 400 cephalometric radiographs grouped into training (75%) and test (25%) data. Each radiograph will be standardized by size and pixel resolution before they are entered into the YOLOv5 algorithm. In the training stage, the model will be optimized using gradient descent considering around 100 repetitions (epochs). To evaluate the performance of the model, the precision and intersection over union (IOU) measures will be analyzed. Preliminary detection results showed overall accuracy greater than 0.89. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Instituciones: Universidad Nacional de Ingeniería