• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Conference on Medical Imaging Physics and Engineering >Decoding Facial Emotion from Activity in The Human Visual Cortex using Functional Magnetic Resonance Imaging
【24h】

Decoding Facial Emotion from Activity in The Human Visual Cortex using Functional Magnetic Resonance Imaging

机译：使用功能磁共振成像解码人类视觉皮质活动中的面部情感

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

At present, the neuroimaging research on visual cortex mainly focuses on decoding semantics of visual stimuli. However, it is still not clear how visual cortex can represent the emotional information in images. At the same time, the huge number of visual voxels and the subtle emotion factors bring great difficulties to accurate emotion decoding. In this paper, by using functional magnetic resonance imaging, we explore the methods of extract facial emotion information from the activity of human visual cortex through multi-voxel analysis based machine learning. Firstly, primary visual areas and facial recognition-related brain areas such as IOG, pSTS, mFus and pFus are extracted from the expanded cerebral cortex as regions of interest (ROI), and the responses of all cortical vertices to five kinds of facial emotional image stimuli are estimated by general linear model (GLM). Secondly, to solve the problem that the number of voxels in ROI is too large to be decoded further, a Principal Component-Max Relevance (PC-MR) method is proposed to extract voxel features and optimize voxel selection, and the voxel feature dimension is reduced to facilitate decoder recognition and decoding. Lastly, a parameter learning optimal SVM classifier is proposed for decoding. The experimental results show that the selected ROI contains the factor information of facial emotion, and can decode it through the voxel signal characteristics of these regions.
机译：目前，Visual Cortex的神经影像学研究主要侧重于可视刺激的解码语义。但是，目前尚不清楚Visual Cortex如何代表图像中的情绪信息。与此同时，大量的视觉体素和微妙的情绪因素带来了巨大的情感解码困难。本文通过使用功能磁共振成像，我们通过基于多体素分析的机器学习探索从人类视觉皮层的活动中提取面部情感信息的方法。首先，从膨胀的脑皮层作为感兴趣区域（ROI）区域，以及所有皮质顶点的响应到五种面部情感图像的脑皮层中提取诸如IOG，PST，MFU和PFU等与面部识别相关的大脑区域。通过一般线性模型（GLM）估计刺激。其次，为了解决ROI中的体素数量太大而无法进一步解码的问题，提出了一个主组件 - 最大相关性（PC-MR）方法以提取体素特征并优化体素选择，并且体素特征尺寸是优化的减少以促进解码器识别和解码。最后，提出了一个参数学习最佳SVM分类器进行解码。实验结果表明，所选ROI包含面部情感的因子信息，并且可以通过这些区域的体素信号特性进行解码。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号