چكيده به لاتين
The use of closed loop control in the electrical stimulation prostheses used for motor rehabilitation requires the need for a feedback signal from the position of the limbs. Our goal is to investigate the possibility of using the sensory regions of the spinal cord in order to extract the motion information. For this purpose, experiments were performed on rats to get acquainted with the correct extracellular recording and processing. Then, basic experiments were carried out on five cats. In past works, motion information, ie, the angles of an animal's leg, has been extracted through the recording of proprioceptive neurons in the dorsal horn gangili. In this study, as far as we know, for the first time, we extracted the motion information from the spinal cord, the dorsal horn of the spinal cord gray mater. For this purpose, we moved the animal's hind limb in an anesthetized state with a pattern similar to the normal walking pattern and simultaneously recorded the position of the markers attached to the leg joints as well as the electrical signals of the spinal cord. Then, this information was used to create a mathematical model for decoding neural signals in order to estimate the angles of the animal's feet.
In order to improve decoding performance, we simultaneously utilized two features, namely fire rate and continuous inter-spike intervals. In order to integrate the resulting information, we constructed a neural network. Also, estimates have been made by using only the fire rate or inter-spike intervals. The results indicate that, in any case, the decoding is carried out with a high accuracy of up to 6.5 percent for the normalized root mean square error, and also the combination of the fire rate and the inter-spike interval in the stack network improved decoding. We also examined the effect of spike sorting, which results indicate a negligible effectiveness of this stage in processing neural data (p = 0.12)
