Decoding Bilateral Hindlimb Kinematics From Cat Spinal Signals Using Three-Dimensional Convolutional Neural Network.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Motor Control

Background:

• Decoding limb kinematics traditionally relies on neural signals from peripheral nerves, spinal cord gray matter, or the sensorimotor cortex.
• Investigating alternative neural signal sources is crucial for advancing brain-computer interfaces and understanding motor control.

Purpose of the Study:

• To investigate the potential of neural signals from the spinal cord's dorsal and lateral columns for decoding hindlimb kinematics during locomotion.
• To compare decoding accuracy between ipsilateral and contralateral hindlimbs and between dorsal and lateral column signals.

Main Methods:

• Experiments were conducted on intact cats trained to walk on a treadmill in a hindlimb-only condition.
• Local field potential signals were recorded using microelectrode arrays implanted in the dorsal and lateral columns of the cat spinal cord.
• Hindlimb joint angles were decoded, and time-frequency and mutual information analyses were performed.

Main Results:

• Hindlimb kinematics were accurately decoded from both dorsal and lateral spinal cord columns, with no significant difference in performance.
• Contralateral hindlimb kinematics were decoded with accuracy comparable to ipsilateral kinematics.
• The theta frequency band demonstrated significantly higher limb kinematic information content compared to other bands, with power correlating with locomotion speed.

Conclusions:

• Neural signals from the spinal cord's dorsal and lateral columns are viable sources for decoding hindlimb kinematics during locomotion.
• The theta frequency band in spinal cord signals carries substantial information about limb movement dynamics.
• These findings offer new insights into neural signal processing for motor control and potential applications in neuroprosthetics.