面向体数据 CT Transformer 的自监督与跨模态预训练规模化
Scaling Self-Supervised and Cross-Modal Pretraining for Volumetric CT Transformers
对做3D医学影像或大体积数据的研究者，可复用其「局部高分辨率Transformer + 全局扫描上下文Transformer」的双层注意力拆分方案，把极端词元规模下的3D注意力变得可计算；同时DINO自蒸馏与SigLIP视觉语言对齐联合预训练的配方，可迁移到其他缺乏标注的体数据域。
原文依据 jointly optimizes a local transformer for high-resolution volumetric feature extraction and a global transformer for whole-scan context modeling
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论本文提出 SPECTRE，一个面向体数据计算机断层扫描（CT）的纯 Transformer 基础模型。我们的「用于 CT 表征提取的自监督与跨模态预训练」（SPECTRE）方法采用可扩展的 3D Vision Transformer 架构以及现代自监督与视觉语言预训练策略，以学习通用 CT 表征。体数据 CT 带来独特挑战，例如极端的词元规模、几何各向异性以及微弱或含噪的临床监督，这些挑战使标准 Transformer 与对比学习配方开箱即用失效。该框架联合优化一个用于高分辨率体数据特征提取的局部 Transformer 和一个用于全扫描上下文建模的全局 Transformer，使大规模 3D 注意力在计算上可行。值得注意的是，SPECTRE 仅使用公开可得的 CT 数据集训练，表明无需依赖私有数据也能获得高性能、可泛化的表征。预训练结合了 DINO 风格的自蒸馏与基于 SigLIP 的视觉语言对齐（使用配对的放射学报告），从而得到既几何一致又具临床意义的特征。在多个 CT 基准上，SPECTRE 在零样本与微调设置下均持续优于先前的 CT 基础模型，确立了 SPECTRE 作为面向 3D 医学影像的可扩展、开放且纯 Transformer 的基础模型。
英文原文摘要 1458 字符
We introduce SPECTRE, a fully transformer-based foundation model for volumetric computed tomography (CT). Our Self-Supervised & Cross-Modal Pretraining for CT Representation Extraction (SPECTRE) approach utilizes scalable 3D Vision Transformer architectures and modern self-supervised and vision-language pretraining strategies to learn general-purpose CT representations. Volumetric CT poses unique challenges, such as extreme token scaling, geometric anisotropy, and weak or noisy clinical supervision, that make standard transformer and contrastive learning recipes ineffective out of the box. The framework jointly optimizes a local transformer for high-resolution volumetric feature extraction and a global transformer for whole-scan context modeling, making large-scale 3D attention computationally tractable. Notably, SPECTRE is trained exclusively on openly available CT datasets, demonstrating that high-performing, generalizable representations can be achieved without relying on private data. Pretraining combines DINO-style self-distillation with SigLIP-based vision-language alignment using paired radiology reports, yielding features that are both geometrically consistent and clinically meaningful. Across multiple CT benchmarks, SPECTRE consistently outperforms prior CT foundation models in both zero-shot and fine-tuned settings, establishing SPECTRE as a scalable, open, and fully transformer-based foundation model for 3D medical imaging.