Deep Learning for Prognosis Using Task-fMRI: A Novel Architecture and Training Scheme
Deep Learning for Prognosis Using Task-fMRI: A Novel Architecture and Training Scheme
复制标题
DOI：
10.1145/3534678.3539362
复制
发表时间：
2022-08
期刊：
影响因子：
--
通讯作者：
Ge Shi;J. Smucny;I. Davidson
中科院分区：
文献类型：
--
作者：
Ge Shi;J. Smucny;I. Davidson
大多数现有的脑成像工作集中在静息状态fMRI（rs-fMRI）数据，其中受试者在扫描仪中休息，通常用于疾病诊断问题。在这里，我们分析任务功能磁共振成像（t-fMRI）的数据，受试者执行多事件任务在多个试验。t-fMRI数据允许探索更具挑战性的应用，例如治疗预后，但代价是分析更加复杂。不仅存在多种类型的试验，而且每种类型的试验对每个受试者重复不同的次数。这导致了多视图（多种类型的试验）和多实例（每个受试者的每种类型的多个试验）设置。我们提出了一个深度的多模型架构，从t-fMRI数据编码多视图的大脑活动和多层感知器集成模型联合收割机这些视图模型，并作出主题明智的预测。我们探索了模型之间的域自适应迁移学习，以解决不平衡的观点和一种新的方法来预测多实例嵌入。我们评估我们的模型在主题交叉验证上的性能，以准确地确定性能。实验结果表明，所提出的方法优于发表的方法的AX-CPT功能磁共振成像数据的预测治疗改善近期发作的儿童精神分裂症的预后问题。据我们所知，这是第一个数据驱动的研究上述任务的体素明智的t-fMRI数据的整个大脑。
Most existing brain imaging work focuses on resting-state fMRI (rs-fMRI) data where the subject is at rest in the scanner typically for disease diagnosis problems. Here we analyze task fMRI (t-fMRI) data where the subject performs a multi-event task over multiple trials. t-fMRI data allows exploring more challenging applications such as prognosis of treatment but at the cost of being more complex to analyze. Not only do multiple types of trials exist but the trials of each type are repeated a varying number of times for each subject. This leads to a multi-view (multiple types of trials) and multi-instance (multiple trials of each type of each subject) setting. We propose a deep multi-model architecture to encode multi-view brain activities from t-fMRI data and a multi-layer perceptron ensemble model to combine these view models and make subject-wise predictions. We explore domain adaptation transfer learning between models to address unbalanced views and a novel way to make predictions out of multi-instance embeddings. We evaluate our model's performance on subject-wise cross-validations to accurately determine performance. The experimental results show the proposed method outperforms published methods on the AX-CPT fMRI data for the prognosis problem of predicting treatment improvement in recent-onset childhood schizophrenia. To our knowledge, this is the first data-driven study of the aforementioned task on voxel-wise t-fMRI data of the whole brain.