Unsupervised, piecewise linear decoding enables an accurate prediction of muscle activity in a multi-task brain...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Intracortical brain-computer interfaces (iBCIs) aim to improve user experience through seamless task transitions.
• Nonlinearity in neural activity poses a significant challenge for developing effective global iBCI decoders.

Purpose of the Study:

• To develop a novel decoding method that addresses the nonlinearity of neural activity for iBCIs.
• To create a decoder that differs from traditional globally optimized approaches.

Main Methods:

• An unsupervised approach was used to implement a piecewise linear decoder based on neural manifold structure.
• Neural signals from motor cortex and electromyographs (EMGs) were recorded from monkeys performing diverse tasks.
• Linear and nonlinear dimensionality reduction techniques identified neural manifolds, and unsupervised algorithms clustered these spaces.
• A linear EMG decoder was fitted for each identified cluster, activating a specific decoder based on neural data point clusters.

Main Results:

• Clusters in neural manifolds corresponded to different tasks or task sub-phases.
• Piecewise decoding performance improved with an increasing number of clusters, eventually plateauing.
• Even with two clusters, piecewise decoding outperformed a global linear decoder and a global recurrent neural network (RNN) decoder.

Conclusions:

• A computationally lightweight solution for iBCI decoders effective across various tasks was introduced.
• Piecewise linear decoding approximates the nonlinearity between neural activity and motor outputs.
• This approach enhances understanding of neural manifold structure in the motor cortex for improved iBCI functionality.