用对比学习提升脑部MRI的自监督表示，同时兼顾局部结构与全局区分性。
Contrastive Joint-Embedding Prediction for Representation Learning in Structural MRI
- 结合联合嵌入预测与对比损失，学习3D脑部MRI的多尺度特征。
- 在双胞胎检索中排名1的召回率达84%，年龄预测误差仅2.55年。
- 适合需要高质量无标签医学影像表征的研究者使用。
自监督学习为医疗影像提供了有效方案，尤其适用于标注数据稀缺、采集成本高的场景。本文提出COJEPA，一种用于三维脑部结构磁共振成像的自监督框架，融合联合嵌入预测（JEPA）与对比损失（CO），旨在同时实现局部可预测性与全局可区分性。模型在两个队列（HCP-YA和AABC，N=2286，年龄22至90岁）的T1加权图像上无标签训练，扩展I-JEPA至3D，采用前景感知块掩码、分层卷积补丁嵌入及世界空间正弦位置编码。在零样本双胞胎检索、脑肿瘤分割（BraTS 2024）和年龄回归（OpenBHB）任务中评估，COJEPA在单卵双胞胎检索中排名1的召回率达到0.84，在OpenBHB 3.0T上的微调年龄平均绝对误差为2.55年，并在BraTS全肿瘤分割中与对比损失方法持平，表明该联合目标能生成兼具判别力与局部结构特性的表示。
原文摘要 · Abstract (English)
Self-supervised learning offers a compelling approach for medical imaging, where labeled data are scarce and acquisition costs are high. We present COJEPA, a self-supervised framework for volumetric brain MRI that combines a joint-embedding predictive architecture (JEPA) with a contrastive loss (CO), targeting two complementary properties: local predictivity and global discriminability. The model is trained without labels on T1-weighted structural MRI from two cohorts (HCP-YA and AABC, $N{=}2286$, ages 22 to 90), extending I-JEPA to 3D with foreground-aware block masking, a hierarchical convolutional patch embedding, and world-space sinusoidal positional encodings. We evaluate all three objectives across zero-shot twin retrieval, brain tumor segmentation (BraTS 2024), and age regression (OpenBHB). COJEPA achieves the best monozygotic twin recall at rank@1 (0.84), the best finetuning age MAE (2.55 years on OpenBHB 3.0T), and matches CO on BraTS whole-tumor Dice, demonstrating that the combined objective yields representations that are simultaneously discriminative and locally structured.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。