چكيده به لاتين
A lot of people lose their ability to walk every year for different reasons such as spinal cord injury. Using Brain-Computer Interfaces (BCI) is a way to help these people with by using it in either their rehabilitation processes or in controlling neuroprostheses. In order to reach this goal, in the first step, the movement information must be extracted from brain electrical signals, and in the next step, this information must be translated into motion commands. This study is focused on estimating the rat locomotion kinematic parameters using the primary motor cortex (M1) signals. In this study, Local Field Potentials were used because they are more stable over time than neuron spikes and are more suitable for long-term use. First rats were trained to walk smoothly on a treadmill at different speeds and with different slopes. Then micro-wire arrays were implanted in the hindlimb area of their primary motor cortex and after recovery electrical signals of M1 was recorded during walking on the treadmill. In recording sessions, the position of hip, knee, ankle and metatarsal-phalangeal joint was recorded simultaneously with brain signals. Finally, different types of features were extracted from the signals and by using Partial Least Squares Regression kinematic parameters such as position, velocity, and angle of joints, while running, were estimated. The best results were obtained using the power and phase of frequency bands as features. For example, the correlation coefficient of the estimated and measured value in horizontal and vertical position of ankle and its angle are 0.48 ± 0.15, 0.4 ± 0.15 and 0.43 ± 0.14 respectively.
