A Single-Trial P300 Detector Based on Symbolized EEG and Autoencoded-(1D)CNN to Improve ITR Performance in BCIs

Abstract

In this paper, we propose a breakthrough single-trial P300 detector that maximizes the information translate rate (ITR) of the brain-computer interface (BCI), keeping high recognition accuracy performance. The architecture, designed to improve the portability of the algorithm, demonstrated full implementability on a dedicated embedded platform. The proposed P300 detector is based on the combination of a novel pre-processing stage based on the EEG signals symbolization and an autoencoded convolutional neural network (CNN). The proposed system acquires data from only six EEG channels; thus, it treats them with a low-complexity preprocessing stage including baseline correction, windsorizing and symbolization. The symbolized EEG signals are then sent to an autoencoder model to emphasize those temporal features that can be meaningful for the following CNN stage. This latter consists of a seven-layer CNN, including a 1D convolutional layer and three dense ones. Two datasets have been analyzed to assess the algorithm performance: one from a P300 speller application in BCI competition III data and one from self-collected data during a fluid prototype car driving experiment. Experimental results on the P300 speller dataset showed that the proposed method achieves an average ITR (on two subjects) of 16.83 bits/min, outperforming by +5.75 bits/min the state-of-the-art for this parameter. Jointly with the speed increase, the recognition performance returned disruptive results in terms of the harmonic mean of precision and recall (F1-Score), which achieve 51.78 ± 6.24%. The same method used in the prototype car driving led to an ITR of ~33 bit/min with an F1-Score of 70.00% in a single-trial P300 detection context, allowing fluid usage of the BCI for driving purposes. The realized network has been validated on an STM32L4 microcontroller target, for complexity and implementation assessment. The implementation showed an overall resource occupation of 5.57% of the total available ROM, ~3% of the available RAM, requiring less than 3.5 ms to provide the classification outcome.

Keywords: CNN; P300; autoencoder; brain–computer interface (BCI); single-trial detection.

Figure 1

BCI stimulation protocol. (a) Snapshot of the 4 targets composing the prototype car driving odd-ball paradigm; (b) experimental setup for the prototype car test; (c) 6 × 6 character matrix for P300 speller diagram from BCI competition III; (d) row and columns indexes distribution.

Figure 2

Overall architecture overview.

Figure 3

Implementation of 1D-LBP code. (a) 40-sample EEG data subset with LBP code extraction (b) sEEG from LBP routine application on a 168 samples trial.

Figure 4

5-layer Autoencoder Implementation.

Figure 5

8-layer sequential NN implementation.

Figure 6

Snapshot of an in vivo proof of concept validation on the acrylic prototype car system designed in [3].

Figure 7

Accuracy rate on subjects A and B for Dataset 2.

Figure 8

ITR graph for four algorithms on Dataset 2.

Figure 9

Histogram plot of the number of NN model parameters and input data size (bytes) for the analyzed methods.

Figure 10

Validation on target flow by X-CUBE-AI.

Figure 11

RAM and I/O memory usage of the autoencoder stage.

Figure 12

RAM and I/O memory usage of the sequential NN following the autoencoder.

Figure 13

Precision versus recall versus F1-score bubble scatter plot for model comparison.

Figure 14

Precision versus recall scatter plot for LBP impact on neural network architecture comparison.