EVOKED POTENTIAL PRIMITIVES OF RAT MOTOR CORTEX SIGNAL ANALYSIS BASED ON ITERATED FUNCTION SYSTEMS

Abstract

The function of the brain has been the focus of neuroscience studies for nearly half a century. The studies found that the evoked potential signals of the motor cortex are the main source to command action. An action signal is composed of several signal primitives that are mainly generated by the motor cortex. It was found that signal primitives of the motor cortex can be produced by several fixed rules and a group of codes called iterated function systems (IFS) code. The goal of our research is to find the relationships between the signal primitives of the motor cortex and actions. We recorded the action signals of the rat motor cortex using 8-channel micro-electrodes and used independent component analysis (ICA) to find the independent source signals called signal primitives. Then, the IFS algorithm was used to find the signal primitive codes, which is the IFS code. The experimental results showed that the source signals of actions produce the IFS rules and a set of codes by the IFS algorithm and conversely, using the IFS rules and the set of codes can reconstruct the source signals. Every 20-character length of action signals will generate unique 6-character IFS codes, meaning that the action signals can be replaced with IFS codes to achieve the compression. We found that the IFS rules and codes can be used to represent different cortex commands which have distinct IFS codes that can be used to classify the movements of rat. The classification result reached 78.75% for rough movement and nearly 50% for subtle movement, where the rough movement is that the rat performs two motions and the subtle movement is three motions. This result shows that the motor cortex command can consist of distinct signal primitives and the huge file size of the motor cortex command is reduced three times by the IFS algorithm.