Cell categories and K-nearest neighbor algorithm based decoding of primary motor cortical activity during...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Robotics

Background:

• Neural decoding translates brain activity into movement commands for controlling external devices.
• Intelligent prostheses require accurate and timely decoding of intended movements.

Purpose of the Study:

• To investigate the impact of cell classification on neural decoding accuracy for a 3-D reach-to-grasp task.
• To determine if primary motor cortex (M1) activity during the planning stage can predict movement parameters.

Main Methods:

• Recorded neural signals from monkey M1 during a 3-D reach-to-grasp task.
• Classified task-related cells based on correlation with movement direction and orientation.
• Utilized the k-nearest neighbor (KNN) algorithm for decoding movement parameters.

Main Results:

• Cell classification enhanced decoding accuracy, requiring fewer cells, even during the movement planning (cortical ready time - CRT) stage.
• High pre-movement decoding accuracy was achieved.
• Populations of task-related cells in M1 demonstrated preferences for specific movement directions and orientations.

Conclusions:

• Classifying neural cells significantly improves decoding performance for prosthetic control.
• M1 plays an active role in movement planning, not just executing commands.
• Understanding cell population preferences in M1 is key for developing sophisticated neural interfaces.