**My research objective** is to decode, encode and understand the neural activities (e.g., [[Human Motor Process]]) in a robust way. # Neural Decoding ## Reconstruction of Continuous Hand Grasp Movement from EEG Using Deep Learning >Y. Tang, N. Robinson, X. Fu, K. P. Thomas, A. A. P. Wai and C. Guan, "Reconstruction of Continuous Hand Grasp Movement from EEG Using Deep Learning," _2024 46th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC)_, Orlando, FL, USA, 2024, pp. 1-4, doi: 10.1109/EMBC53108.2024.10781850. <b class='rainbow-text'>Significance of study: </b> + #NeuralDecoding This study address the research gap that **hand movement receives less attention compared to other limbs**, by collecting a novel dataset with a novel experiment paradigm. + #BCI This study achieves the **continuous** hand joints reconstruction from EEG such that $r=0.627$ with channel-wise attention CNN. + #BCI The continuous trajectory can be converted into discrete BCI control command (i.e., a grasp is 1, while no grasp is 0). **The accuracy of this command would be $83.5\%$.** ![[finger_visualization_1.png]] ![[finger_visualization_2.png]] > [!Experiment Setup] > **Left** - Every trial began with a 10-second rest period, followed by a 10-second motor execution task phase. During the execution, a metronome cue from a speaker instructed the subject to perform a single cycle of the grasp action, the hand starting from the original open position. The subject was instructed to perform grasp actions 5-7 times following the metronome cues within each task phase. Each grasp last for 2 ± 0.5 seconds. > > **Right** - Each participant was seated at a table with a screen monitor in the front displaying text instructions as well as a pad for resting their wrist. > ![[finger_visualization_3.png]] > [!Channel Importance Analysis] > Channel important analysis is conducted through an XAI algorithm, DeepLift. > > **Topographic Channel Importance Maps:** The resultant topographic maps are presented below, where the FC1, C3 and FC3 channels on the left head region have the most significant impact to the right hand reconstruction according to the high accuracy group. > ![[finger_visualization_4.png]] ![[canvas_2.png]] # Neural Rehabilitation Participating in clinical trials about BCI-driven stroke therapy is an important part during my research. ![[Psychopy Logotext.png|200]] # Neural Signal Analysis Below are my tutorials on neural signal processing: + [Removing artefact removal with EEGLab](https://blog.csdn.net/weixin_44517742/article/details/132660904) # Other Research I have participated in several research projects about Natural Language Processing and Knowledge Graph during the undergraduate. To view more details, please view [[Other Research]].