Intracortical brain-machine interface
- To provide motor functions via high-performance neural prosthetics to patients with tetraplegia.
- To develop efficient decoding algorithm that can predicts kinematics information from ongoing neural activities and to evaluate its effectiveness through primates.
- To understand sensory copy between motor cortex and posterior parietal cortex, when it provides stimuli associated with proprioceptive information during recording neural activities.
- Neural firing pattern model
- The best kinematics parameters in the three dimensional spaces for the BMIs
- Neural dynamic state model related with 3D arm movements
- Decoding algorithm for neuronal ensemble activities
- Encoding based on the proprioceptive feedback for the bidirectional BMI
- Neural simulator for arm movements in three dimensional spaces
- Medical Device Development Center, Daegu Gyeongbuk-Medical Innovation Foundation (Dr. Jeong-Woo Sohn)
- Seoul National University Hospital (Prof. Chun Kee Chung, Prof. June Sic Kim)
- Seoul National University (Prof. Yoon-Kyu Song)
- KAIST (Prof. Jaeseung Jeong)
- Hanyang University (Prof. Dong-Pyo Jang)
- The Brain Research Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Science, ICT & Future Planning
Bidirectional brain-machine interfaces
Neural prosthetics based on brain-machine interfaces (BMIs) have become one of the key neurotechnologies, holding promises to restore lost functions in patients with tetraplegia. Particularly for neural motor prosthetics, intra-cortical BMIs provide a means to convert discharge activities of single neurons into motor kinematic information.
However, proprioceptive feedback is required to make deliberate BMIs like as our body controls arm movement in bidirectional way. Our goal is 1) to decode arm movement information from intra-cortical neuron population signal or Electrocorticography (ECoG) signal and 2) to develop combined encoding and decoding technology that can extract movement information from neuronal activity dynamically changed by proprioceptive information encoding.
Neuronal firing pattern model
Information transferred by neurons can be decoded using firing rate that is the number of spikes in unit time. However, the temporal pattern of firing also have useful information. Our goal is to make use of firing rate and temporal information of neuron’s activity to improve performance of BMIs
Decoding algorithm for neuronal ensemble activities
To decode arm movement information in 2D or 3D space, Kalman filter of linear filter have been used. The decoding algorithm predict end point velocity of hand based on population neuronal signal. Our goal is to develop decoding algorithm that decodes speed information and direction information separately. Speed information will be extracted using Multi-layer perceptron and direction information will be extracted using kalman filter. And finally, the two algorithm will be combined to predict endpoint velocity.
Encoding process for proprioceptive feedback
We are developing a technology to transfer the proprioceptive feedback to brain artificially based on the assumption that the feedback will improve the BMIs performance. To realize the technology, first, neuronal activity of posterior parietal cortex related with proprioceptive sense will be studied. Second, the encoding process will be developed that maps the directional information of arm movement with stimulation pattern in real time.
Neural simulator for arm movements in three dimensional spaces
Our research team is developing neural simulator using VR technology and leap motion device. To make use of neural simulator, user conducts an experiment such as reaching hand to targets visualized in VR space and the movement information is transferred to spike generator that generates spike trains in real time. Finally, decoder extract movement information and visualize those information in VR space to give visual feedback to user. And further, proprioceptive feedback is given using powered exoskeleton. Through these process, how various decoding algorithm and feedback method affect to decoder performance can be assessed.
- International Papers
- International Conferences
A simulation study on the generative neural ensemble decoding algorithms
S.-P. Kim, M.-K. Kim and G.-T. Park
International Conference on Pattern Recognition, Istanbul, Aug. 2010.