Integration of Preprocessing Techniques and Artificial Intelligence for Accurate Inferior Alveolar Nerve Segmentation in Cone Beam Computed Tomography

Accurate segmentation of the Inferior Alveolar Nerve (IAN) in Cone Beam Computed Tomography (CBCT) is critical for improving patient outcomes and reducing surgical risks in dental and maxillofacial procedures. This study explored UNet, Residual UNet, and Attention UNet models combined with preprocessing techniques like image sharpening and Contrast Limited Adaptive Histogram Equalization (CLAHE) for IAN segmentation. Evaluation metrics included precision, recall, F1 score, accuracy, and Intersection over Union (IoU). Preprocessing led to a significant improvement in performance. Residual UNet with CLAHE and image sharpening achieved the highest results, with an accuracy of 0.940, precision of 0.916, recall of 0.833, F1 score of 0.872, and IoU of 0.675. These findings highlight the effectiveness of Residual UNet combined with CLAHE and image sharpening for accurate IAN segmentation. This method enhances the segmentation accuracy and may lead to better clinical decision-making in oral surgery.