发表机构
Center for Data Science, New York University; NYU Grossman School of Medicine; Courant Institute of Mathematical Sciences, NYU（纽约大学数据科学中心; 纽约大学格罗斯曼医学院; 纽约大学柯朗数学科学研究所）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究比较了基于解剖特征、CNN和ViT基础模型的三种特征提取范式，发现线性解剖模型性能可与复杂AI模型匹敌，并提出解剖分割预训练（ASP）方法，在生物学年龄估计上优于现有模型。
AI 中文摘要
在这项工作中，我们全面评估了基于AI的神经影像建模中三种流行的特征提取范式：（1）解剖表面和体积的计算，（2）使用卷积神经网络（CNN）的监督学习，以及（3）视觉变换器（ViT）基础模型的无监督预训练，随后进行监督微调。我们的研究基于18个公开可用的数据集，包含来自约80,000名参与者的3D结构T1加权MRI扫描，涵盖七个不同的临床任务。我们观察到，基于解剖特征的线性模型能够匹配由复杂AI框架（包括在数千次扫描上训练的基础模型）学习到的复杂非线性特征的诊断性能。相反，CNN和预训练的ViT学习到的特征隐式地捕获了相关的解剖信息，绕过了显式特征提取的需要。基于这些见解，我们提出了解剖分割预训练（ASP），一种在基础模型预训练期间纳入解剖信息的新方法，该方法在生物学年龄估计方面优于现有模型。
英文摘要
In this work, we comprehensively evaluate three popular feature-extraction paradigms in AI-based neuroimaging modeling: (1) computation of anatomical surfaces and volumes, (2) supervised learning with convolutional neural networks (CNNs), and (3) unsupervised pretraining of vision transformer (ViT) foundation models, followed by supervised finetuning. Our study is based on 18 publicly available datasets containing 3D structural T1-weighted MRI scans from approximately 80,000 participants across seven distinct clinical tasks. We observe that a linear model based on anatomical features matches the diagnostic performance of complex nonlinear features learned by sophisticated AI frameworks, including foundation models trained on thousands of scans. Conversely, CNNs and pretrained ViTs learn features that implicitly capture relevant anatomical information, bypassing the need for explicit feature extraction. Building upon these insights, we propose Anatomy Segmentation Pretraining (ASP), a novel method to incorporate anatomical information during foundation-model pretraining, which outperforms existing models in biological age estimation.
