MAGEFormer：学习各向异性CT分割的度量一致表示
MAGEFormer: Learning Metric-Consistent Representations for Anisotropic CT Segmentation
查看机构详情
- Durham University（杜伦大学）
机构由 AI 辅助整理，请以论文原文为准。
浏览论文内容
中文总结 AI 辅助
针对各向异性CT分割中度量失配与精度下降问题，提出几何校准框架MAGEFormer，通过度量自适应空间嵌入、几何约束注意力和几何视图投票嵌入物理度量约束，在BTCV和FLARE 22上取得最优边界精度。
中文摘要 AI 辅助
视觉Transformer（ViT）在体积分割中表现出强大的性能，但其在临床CT上的有效性受到各向同性欧几里得晶格假设的限制。这一假设与各向异性CT采集相冲突，导致两个关键问题：（1）体素索引与物理解剖结构之间的度量不匹配，（2）各向同性重采样导致的精度下降。为解决这些问题，我们提出了MAGEFormer，一个几何校准框架，将物理度量约束直接嵌入表示学习中。我们的方法引入了度量自适应空间嵌入（MASE），利用体素间距校准位置频率；几何约束注意力（GCA），抑制物理上不合理的特征相关性；以及几何视图投票（GVV），在推理期间减少离散化偏差。我们在两个多器官腹部CT基准数据集BTCV和FLARE 22上，在统一协议下，与强CNN和基于Transformer的基线方法进行了评估。MAGEFormer在比较的方法中实现了最强的边界精度，在BTCV上HD95为10.58毫米，在FLARE 22上HD95为3.40毫米，并在相同协议下在Dice指标上表现出一致的提升。这些结果表明，对于各向异性CT分割，几何感知的内部校准比仅依赖传统的各向同性预处理更为有效。
英文摘要
Vision Transformers (ViTs) have shown strong performance in volumetric segmentation, but their effectiveness on clinical CT is limited by an isotropic Euclidean lattice assumption. This conflicts with anisotropic CT acquisition, leading to two key issues: (1) a metric mismatch between voxel indices and physical anatomy, and (2) accuracy degradation from isotropic resampling. To address this, we propose MAGEFormer, a geometry-calibrated framework that embeds physical metric constraints directly into representation learning. Our method introduces Metric-Adaptive Spatial Embedding (MASE) to calibrate positional frequencies using voxel spacing, Geometry-Constrained Attention (GCA) to suppress physically implausible feature correlations, and Geometric View Voting (GVV) to reduce discretization bias during inference. We evaluate MAGEFormer on two multi-organ abdominal CT benchmarks, BTCV and FLARE 22, under a unified protocol against strong CNN and Transformer-based baselines. MAGEFormer achieves the strongest boundary accuracy among the compared methods, with 10.58 mm HD95 on BTCV and 3.40 mm HD95 on FLARE 22, and shows consistent gains in Dice under the same protocol. These results show that geometry-aware internal calibration is more effective than relying on conventional isotropic preprocessing alone for anisotropic CT segmentation.
