Neural decoding based on probabilistic neural network.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Brain-machine interfaces (BMIs) offer potential solutions for severe paralysis by enabling direct control of prosthetic devices.
• Effective neural decoding algorithms are crucial for translating brain activity into intended movements.

Purpose of the Study:

• To introduce and evaluate two novel neural decoding methods, a probabilistic neural network (PNN) decoder and a modified PNN (MPNN) decoder.
• To compare the performance of MPNN against traditional decoders like Wiener filter (WF) and Kalman filter (KF).

Main Methods:

• Rats were trained to operate a lever, with motor cortex neural activity and lever pressure recorded synchronously using microelectrode arrays and pressure sensors.
• Neural signals were decoded using PNN and MPNN algorithms to estimate pressure values.
• Decoder performance was quantified using correlation coefficient (CC) and mean square error (MSE).

Main Results:

• The MPNN decoder achieved a high correlation coefficient (0.8657) and low mean square error (0.2563), outperforming WF and KF decoders.
• MPNN performance was independent of discretization level, suggesting broad applicability.
• The MPNN decoder demonstrated capability in both movement state detection and continuous kinematic parameter estimation.

Conclusions:

• The modified PNN (MPNN) decoder represents a significant advancement in neural decoding for brain-machine interface applications.
• MPNN's robustness and high performance indicate its potential for controlling prosthetic devices and aiding individuals with paralysis.