BrainSegFounder：迈向神经影像分割的3D基础模型
BrainSegFounder: Towards 3D foundation models for neuroimage segmentation
DOI 原文链接
用sci-hub下载
如无法下载，请从 Sci-Hub 选择可用站点尝试。
影响因子:11.8
分区:医学1区 Top / 计算机：人工智能1区 计算机：跨学科应用1区 工程：生物医学1区 核医学1区
发表日期:2024 Oct
作者:
Joseph Cox, Peng Liu, Skylar E Stolte, Yunchao Yang, Kang Liu, Kyle B See, Huiwen Ju, Ruogu Fang
DOI:
10.1016/j.media.2024.103301
摘要
脑健康研究的快速发展日益依赖人工智能（AI）分析和解读神经影像数据。医学基础模型显示出优越的性能和更好的样本效率。本文提出了一种通过自监督训练创建多模态神经影像分割的3维（3D）医学基础模型的新方法。该方法采用一种创新的两阶段预训练策略，基于视觉变换器（vision transformers）。第一阶段利用大型无标签的多模态脑磁共振成像（MRI）数据集（41,400名参与者）编码正常大脑的解剖结构，重点识别不同脑结构的形状和大小等关键特征。第二阶段则识别疾病特异性特征，如肿瘤和病变的几何形状及其在大脑中的空间位置。该双阶段方法显著减少了传统AI模型在神经影像分割中所需的大量数据，同时具有适应多种成像模态的灵活性。在脑肿瘤分割（BraTS）挑战和脑卒中后病变解剖追踪（ATLAS v2.0）数据集上对模型BrainSegFounder进行了严格评估。结果显示，该模型性能显著优于以往完全监督学习的获胜方案。研究表明，扩大模型复杂度及利用源自正常大脑的海量无标签训练数据，均能提升模型在神经影像分割任务中的准确性和预测能力。我们的预训练模型和代码可在https://github.com/lab-smile/BrainSegFounder获取。
Abstract
The burgeoning field of brain health research increasingly leverages artificial intelligence (AI) to analyze and interpret neuroimaging data. Medical foundation models have shown promise of superior performance with better sample efficiency. This work introduces a novel approach towards creating 3-dimensional (3D) medical foundation models for multimodal neuroimage segmentation through self-supervised training. Our approach involves a novel two-stage pretraining approach using vision transformers. The first stage encodes anatomical structures in generally healthy brains from the large-scale unlabeled neuroimage dataset of multimodal brain magnetic resonance imaging (MRI) images from 41,400 participants. This stage of pertaining focuses on identifying key features such as shapes and sizes of different brain structures. The second pretraining stage identifies disease-specific attributes, such as geometric shapes of tumors and lesions and spatial placements within the brain. This dual-phase methodology significantly reduces the extensive data requirements usually necessary for AI model training in neuroimage segmentation with the flexibility to adapt to various imaging modalities. We rigorously evaluate our model, BrainSegFounder, using the Brain Tumor Segmentation (BraTS) challenge and Anatomical Tracings of Lesions After Stroke v2.0 (ATLAS v2.0) datasets. BrainSegFounder demonstrates a significant performance gain, surpassing the achievements of the previous winning solutions using fully supervised learning. Our findings underscore the impact of scaling up both the model complexity and the volume of unlabeled training data derived from generally healthy brains. Both of these factors enhance the accuracy and predictive capabilities of the model in neuroimage segmentation tasks. Our pretrained models and code are at https://github.com/lab-smile/BrainSegFounder.