PanMB2-Net: a deep learning framework for preoperative MB2 canal risk assessment on panoramic radiographs

This study aimed to develop and validate a clinically motivated artificial intelligence framework for preoperative risk assessment of the second mesiobuccal (MB2) canal in maxillary first molars using panoramic radiographs.

A total of 388 panoramic radiographs were retrospectively collected. Stage 1 used YOLOv5 to localize maxillary first molars and crop tooth-level regions of interest. Stage 2 applied Panoramic-based MB2 Prediction Network (PanMB2-Net) (ResNet50 with gray-guided attention, receptive field block, contrastive regularization, and Sobel edge loss) for MB2 classification. Data were split 80/20 at the image level for training/validation. Detection was assessed by localization accuracy; classification via accuracy, precision, recall, and F1-score. Ablation studies and class activation heatmaps were used for validation and visualization.

PanMB2-Net achieved 72.9% validation accuracy, exceeding the 50% statistical baseline for the balanced dataset. For MB2 prediction, precision, recall, and F1-score were 0.687, 0.673, and 0.679, respectively. Comparative experiments showed consistent improvements over common convolutional and transformer-based baselines, and ablation studies confirmed the complementary contributions of gray-guided attention, multi-scale enhancement, and auxiliary regularization losses.

PanMB2-Net is a feasible clinically aligned tool for MB2 risk assessment on panoramic radiographs. Although not a replacement for CBCT, it may support preoperative decision-making by highlighting cases that warrant increased attention without additional radiation exposure or examination cost.

PanMB2‑Net provides preoperative MB2 canal risk assessment from routine panoramic radiographs before endodontic therapy, identifying teeth that merit closer exploration or CBCT, supporting decisions without added radiation or cost. It augments clinician judgment and CBCT rather than replacing them.