Prior knowledge improves decoding of finger flexion from electrocorticographic signals

Abstract

Brain-computer interfaces (BCIs) use brain signals to convey a user's intent. Some BCI approaches begin by decoding kinematic parameters of movements from brain signals, and then proceed to using these signals, in absence of movements, to allow a user to control an output. Recent results have shown that electrocorticographic (ECoG) recordings from the surface of the brain in humans can give information about kinematic parameters (e.g., hand velocity or finger flexion). The decoding approaches in these studies usually employed classical classification/regression algorithms that derive a linear mapping between brain signals and outputs. However, they typically only incorporate little prior information about the target movement parameter. In this paper, we incorporate prior knowledge using a Bayesian decoding method, and use it to decode finger flexion from ECoG signals. Specifically, we exploit the constraints that govern finger flexion and incorporate these constraints in the construction, structure, and the probabilistic functions of the prior model of a switched non-parametric dynamic system (SNDS). Given a measurement model resulting from a traditional linear regression method, we decoded finger flexion using posterior estimation that combined the prior and measurement models. Our results show that the application of the Bayesian decoding model, which incorporates prior knowledge, improves decoding performance compared to the application of a linear regression model, which does not incorporate prior knowledge. Thus, the results presented in this paper may ultimately lead to neurally controlled hand prostheses with full fine-grained finger articulation.

Keywords: brain–computer interface; decoding algorithm; electrocorticographic; finger flexion; machine learning; prior knowledge.

Figure 1

Experimental setup for this study.

Figure 2

The schematic diagram of Bayesian decoding.

Figure 3

Examples of two flexion traces.

Figure 4

A diagram of possible state transitions for finger movements.

Figure 5

SNDS model in which S_t, Y_t, Z_t represent the moving states, real finger position, and the estimated finger position at time t, respectively.

Figure 6

(A) Probabilistic density function (PDF) of Y_t−1 given S_t−1 = extension and S_t = flexion; (B) Probabilistic density function of Y_t−1 given S_t−1 = flexion and S_t = extension.

Figure 7

(A) Kernel locations for $\widehat{p}\left(Y_{t-1},Y_t\right)$ under extension state; (B) kernel locations for $\widehat{p}\left(Y_{t-1},Y_t\right)$ under flexion state; (C) kernel locations for $\widehat{p}\left(Y_{t-1},Y_t\right)$ under rest state. Numbers on the axis are the normalized amplitude of the fingers’ flexion.

Figure 8

(A) Actual finger flexion (dotted trace) and decoded finger flexion (solid trace) using pace regression (mean square error 0.68); (B) Actual finger flexion (dotted trace) and decoded finger flexion (solid trace) using SNDS (mean square error 0.40); (C) Actual finger flexion (dotted trace) and state prediction (solid trace).

Figure 9

Comparison of computational models that incorporate different types of prior knowledge. See text for details.

Figure 10

Performance (given as mean squared error for pace regression, intermediate models (a–d), and the final model.)

Figure 11

Exemplary decoding results of the models in Figure 10.