ABSTRACT The prevalence of pulpal and periapical diseases exceeds 50% in the general population. Root canal treatment is currently recognized as the gold standard treatment, in which precise measurement of working length (WL) is critical for treatment success. In this study, a total of 100 eligible extracted teeth were collected and scanned using cone‐beam computed tomography (CBCT) to obtain high‐resolution three‐dimensional images. For WL calculation, we employed a V‐net‐based segmentation network for simulated paths of the root canal file, incorporating an encoder–decoder structure and a multi‐task learning strategy, and achieved a sensitivity of 94.7%. Ablation studies revealed that integrating the decoder's mask branch, key point branch, and boundary branch significantly improved the segmentation accuracy. The WL calculation comprised three stages: skeleton extraction and noise suppression, branch extraction and generation of the simulated paths of the root canal file, and length calculation based on three‐dimensional spline curves. The model achieved an average prediction error of 0.28 mm and an accuracy of 86.67% in WL prediction. These findings indicate that this V‐net‐based multi‐branch framework for precise WL estimation from CBCT holds substantial clinical application potential. Future work will focus on enhancing generalization and addressing challenges posed by calcified or anatomically complex root canals.

Current out-of-distribution (OOD) detection methods typically assume balanced in-distribution (ID) data, while most real-world data follow a long-tailed distribution. Previous approaches to long-tailed OOD detection often involve balancing the ID data by reducing the semantics of head classes. However, this reduction can severely affect the classification accuracy of ID data. The main challenge of this task lies in the severe lack of features for tail classes, leading to confusion with OOD data. To tackle this issue, we introduce a novel Prioritizing Attention to Tail (PATT) method using augmentation instead of reduction. Our main intuition involves using a mixture of von Mises-Fisher (vMF) distributions to model the ID data and a temperature scaling module to boost the confidence of ID data. This enables us to generate infinite contrastive pairs, implicitly enhancing the semantics of ID classes while promoting differentiation between ID and OOD data. To further strengthen the detection of OOD data without compromising the classification performance of ID data, we propose feature calibration during the inference phase. By extracting an attention weight from the training set that prioritizes the tail classes and reduces the confidence in OOD data, we improve the OOD detection capability. Extensive experiments verified that our method outperforms the current state-of-the-art methods on various benchmarks.