Cheng Shuguan, Li Mengmeng, Yu Haifei, Zhao Kun, Liu Shuo, Wan Hong
Annu Int Conf IEEE Eng Med Biol Soc. 2020 Jul;2020:38-41. doi: 10.1109/EMBC44109.2020.9175413.
It is a hot research direction to reveal the working mechanism of brain by measuring the connection characteristics of brain function network. In this paper, to decode pigeon behavior outcomes in goal-directed decision task, an experiment based on plus maze was designed and the nidopallium caudolaterale (NCL) of the pigeon was selected as the target brain region. The local field potential (LFP) signals in the waiting area (WA) and turning area (TA) were recorded when the pigeons performed the goal-directed tasks. Then, the brain functional connection networks of the LFPs were constructed and the extracted features were applied to decode pigeon behavior outcomes. Firstly, continuous wavelet transform (CWT) was used to carried out time-frequency analysis and the task-related frequency band (40-60 Hz) was extracted. Then, weighted sparse representation (WSR) method was used to construct the functional connectivity network and the related network features were selected. Finally, k-nearest neighbor (kNN) algorithm was used to decode behavior outcomes. The results show that the energy difference between TA and WA in 40-60 Hz band is significantly higher than those in other bands. The selected features have good discriminability for the representation of the differences between WA and TA. The decoding results also suggest the classification performance of the different behavior outcomes. These results show the effectiveness of the WSR to construct the function network to decode behavior outcomes.
通过测量脑功能网络的连接特征来揭示大脑的工作机制是一个热门的研究方向。本文为了解码鸽子在目标导向决策任务中的行为结果，设计了一项基于十字迷宫的实验，并选择鸽子的尾外侧巢皮质（NCL）作为目标脑区。当鸽子执行目标导向任务时，记录等待区（WA）和转弯区（TA）的局部场电位（LFP）信号。然后，构建LFP的脑功能连接网络，并将提取的特征应用于解码鸽子的行为结果。首先，使用连续小波变换（CWT）进行时频分析，提取与任务相关的频段（40 - 60 Hz）。然后，使用加权稀疏表示（WSR）方法构建功能连接网络，并选择相关的网络特征。最后，使用k近邻（kNN）算法解码行为结果。结果表明，40 - 60 Hz频段中TA和WA之间的能量差异明显高于其他频段。所选特征对WA和TA之间差异的表示具有良好的可辨别性。解码结果也表明了不同行为结果的分类性能。这些结果表明WSR构建功能网络以解码行为结果的有效性。
