Decoding passive tactile shape from functional MRI signals

Decoding passive tactile shapes is a challenge in neuroscience. In this study, we extracted BOLD signals from three regions of interest, the intraparietal sulcus (IPS), the lateral occipital cortex (LOC), and the dorsolateral prefrontal cortex (dlPFC). All are known to be the shared areas between visions and haptics. Then we used a VDVAE (Very Deep Variational Auto-Encoder) to visualize the haptic information from these areas. This approach opens a new way for explaining the human tactile perception and possibly contributes to haptic-related diagnosis in future work.解码被动触觉形状是神经科学领域的一项研究挑战。在本研究中，我们从三个感兴趣脑区——顶内沟（intraparietal sulcus, IPS）、外侧枕叶皮层（lateral occipital cortex, LOC）以及背外侧前额叶皮层（dorsolateral prefrontal cortex, dlPFC）——提取了血氧水平依赖（Blood Oxygen Level Dependent, BOLD）信号。上述脑区均被证实为视觉与触觉共享的功能脑区。随后我们采用超深度变分自编码器（Very Deep Variational Auto-Encoder, VDVAE）对来自这些脑区的触觉信息进行可视化。该方法为阐释人类触觉感知提供了全新路径，有望在未来研究中为触觉相关诊断提供助力。