Abstract
Accurate understanding of anatomical structures is essential for reliably staging certain dental diseases. A way of introducing this within semantic segmentation models is by utilising hierarchy-aware methodologies. However, existing hierarchy-aware segmentation methods largely encode anatomical structure through the loss functions, providing weak and indirect supervision. We introduce a general framework that embeds an explicit anatomical hierarchy into semantic segmentation by coupling a recurrent, level-wise prediction scheme with restrictive output heads and top-down feature conditioning. At each depth of the class tree, the backbone is re-run on the original image concatenated with logits from the previous level. Child class features are conditioned using Feature-wise Linear Modulation of their parent class probabilities, to modulate child feature spaces for fine grained detection. A probabilistic composition rule enforces consistency between parent and descendant classes. Hierarchical loss combines per-level class weighted Dice and cross entropy loss and a consistency term loss, ensuring parent predictions are the sum of their children. We validate our approach on our proposed dataset, TL-pano, containing 194 panoramic radiographs with dense instance and semantic segmentation annotations, of tooth layers and alveolar bone. Utilising UNet and HRNet as donor models across a 5-fold cross validation scheme, the hierarchical variants consistently increase IoU, Dice, and recall, particularly for fine-grained anatomies, and produce more anatomically coherent masks. However, hierarchical variants also demonstrated increased recall over precision, implying increased false positives. The results demonstrate that explicit hierarchical structuring improves both performance and clinical plausibility, especially in low data dental imaging regimes.
Abstract (translated)
对解剖结构的准确理解对于可靠地分期某些牙科疾病至关重要。在语义分割模型中引入这一要素的一种方法是利用层次感知的方法论。然而，现有的层次感知分割方法大多通过损失函数来编码解剖结构，提供的是弱且间接的监督。我们介绍了一种通用框架，该框架将明确的解剖层次嵌入到语义分割中，通过结合递归、级联预测方案和限制性输出头以及自顶向下的特征调节实现这一目标。 在类别树的每个深度上，骨干网络都会在原始图像与前一层的logits（未标准化分数）串联后重新运行。子类别的特征通过对其父类概率进行逐通道线性调制来调节其子类特征空间，以促进细粒度检测。一种概率组合规则确保了父级和后代类别之间的连贯性。层次损失结合了每个级别的加权Dice损失、交叉熵损失以及一致性术语损失，确保父预测是其子项的总和。 我们在提出的TL-pano数据集上验证了我们的方法，该数据集包含194张全景X光片，带有密集实例和语义分割注释，包括牙齿层和牙槽骨。我们使用UNet和HRNet作为供体模型，并通过五折交叉验证方案进行实验。层次变种在所有IoU、Dice以及召回率上持续提高，特别是在细粒度解剖结构方面，并产生更符合解剖学的掩码。然而，层次变种也显示出了更高的召回率而不仅仅是精度，暗示了更多的假阳性。 结果表明，在数据量较少的牙科影像模式下，显式的层次结构能够显著改善性能和临床合理性。
URL
https://arxiv.org/abs/2512.07984
