Wulan Naren, An Lijun, Zhang Chen, Kong Ru, Chen Pansheng, Bzdok Danilo, Eickhoff Simon B, Holmes Avram J, Yeo B T Thomas
Centre for Sleep & Cognition & Centre for Translational Magnetic Resonance Research, Yong Loo Lin School of Medicine, National University of Singapore.
Department of Electrical and Computer Engineering, National University of Singapore, Singapore.
bioRxiv. 2024 Jan 2:2023.12.31.573801. doi: 10.1101/2023.12.31.573801.
Individualized phenotypic prediction based on structural MRI is an important goal in neuroscience. Prediction performance increases with larger samples, but small-scale datasets with fewer than 200 participants are often unavoidable. We have previously proposed a "meta-matching" framework to translate models trained from large datasets to improve the prediction of new unseen phenotypes in small collection efforts. Meta-matching exploits correlations between phenotypes, yielding large improvement over classical machine learning when applied to prediction models using resting-state functional connectivity as input features. Here, we adapt the two best performing meta-matching variants ("meta-matching finetune" and "meta-matching stacking") from our previous study to work with T1-weighted MRI data by changing the base neural network architecture to a 3D convolution neural network. We compare the two meta-matching variants with elastic net and classical transfer learning using the UK Biobank (N = 36,461), Human Connectome Project Young Adults (HCP-YA) dataset (N = 1,017) and HCP-Aging dataset (N = 656). We find that meta-matching outperforms elastic net and classical transfer learning by a large margin, both when translating models within the same dataset, as well as translating models across datasets with different MRI scanners, acquisition protocols and demographics. For example, when translating a UK Biobank model to 100 HCP-YA participants, meta-matching finetune yielded a 136% improvement in variance explained over transfer learning, with an average absolute gain of 2.6% (minimum = -0.9%, maximum = 17.6%) across 35 phenotypes. Overall, our results highlight the versatility of the meta-matching framework.
基于结构磁共振成像（MRI）的个性化表型预测是神经科学的一个重要目标。预测性能会随着样本量的增大而提高，但参与者少于200人的小规模数据集往往不可避免。我们之前提出了一个“元匹配”框架，用于转换从大型数据集中训练的模型，以改进在小型收集工作中对新的未见表型的预测。元匹配利用表型之间的相关性，当应用于以静息态功能连接作为输入特征的预测模型时，相较于经典机器学习有大幅提升。在此，我们通过将基础神经网络架构改为3D卷积神经网络，对我们之前研究中表现最佳的两个元匹配变体（“元匹配微调”和“元匹配堆叠”）进行调整，使其适用于T1加权MRI数据。我们使用英国生物银行（N = 36,461）、人类连接组计划青年成人（HCP - YA）数据集（N = 1,017）和HCP - 老年数据集（N = 656），将这两个元匹配变体与弹性网络和经典迁移学习进行比较。我们发现，无论是在同一数据集中转换模型，还是在使用不同MRI扫描仪、采集协议和人口统计学特征的数据集之间转换模型，元匹配都大幅优于弹性网络和经典迁移学习。例如，当将一个英国生物银行模型转换到100名HCP - YA参与者时，元匹配微调在可解释方差方面比迁移学习提高了136%，在35种表型上平均绝对增益为2.6%（最小值 = -0.9%，最大值 = 17.6%）。总体而言，我们的结果突出了元匹配框架的通用性。
