会议ISBI发表论文：通过强化学习方法探索大脑连接中枢的新型多模态框架

论文“A NOVEL MULTI-MODALITY FRAMEWORK FOR EXPLORING BRAIN CONNECTIVITY HUBS VIA REINFORCEMENT LEARNING APPROACH”被会议 “IEEE International Symposium on Biomedical Imaging”接收

时间：2023-01-23

关键词：脑影像人工智能，文章接收

  近日，硕士研究生张海洋在医学图像处理领域会议“IEEE International Symposium on Biomedical Imaging”发表题为“Differentiating brain states via multi-clip random fragment strategy-based interactive bidirectional recurrent neural network”的文章，通讯作者为张枢教授。

  Exploring the brain connectivity and identifying the connectivity hubs is an important issue for better understanding the working mechanism of the brain as well as assisting to investigate the brain of disease and disorders. In recent years, on one hand, plenty research have been proposed to study brain connectivity hubs either on functional or structural perspective, but very few studies are focusing on integration them together; on the other hand, efficient learning approach to deal with the complex brain network is urgently needed. To address above mentioned issues, in this paper, we propose a novel Multi-Modality Reinforcement Learning (MM-RL) approach,50 brain connectivity hubs are identified and discussed. This work sheds the new insights that reinforcement learning approach can be adopted to study the brain connectivity, identify the potential hubs and interpret the relationship between function and structure.

图1：The pipeline of proposed MM-RL approach.