Fabrication and characterization of polyimide-based 'smooth' titanium nitride microelectrode arrays for neural stimulation and recording.

OBJECTIVE: As electrodes are required to interact with sub-millimeter neural structures, innovative microfabrication processes are required to enable fabrication of microdevices involved in such stimulation and/or recording. This requires the development of highly integrated and miniaturized systems, comprising die-integration-compatible technology and flexible microelectrodes. To elicit selective stimulation and recordings of sub-neural structures, such microfabrication process flow can beneficiate from the integration of titanium nitride (TiN) microelectrodes onto a polyimide substrate. Finally, assembling onto cuffs is required, as well as electrode characterization. APPROACH: Flexible TiN microelectrode array integration and miniaturization was achieved through microfabrication technology based on microelectromechanical systems (MEMS) and complementary metal-oxide semiconductor processing techniques and materials. They are highly reproducible processes, granting extreme control over the feature size and shape, as well as enabling the integration of on-chip electronics. This design is intended to enhance the integration of future electronic modules, with high gains on device miniaturization. MAIN RESULTS: (a) Fabrication of two electrode designs, (1) 2 mm long array with 14 TiN square-shaped microelectrodes (80 × 80 µm2), and (2) an electrode array with 2 mm × 80 µm contacts. The average impedances at 1 kHz were 59 and 5.5 kΩ, respectively, for the smaller and larger contacts. Both designs were patterned on a flexible substrate and directly interconnected with a silicon chip. (b) Integration of flexible microelectrode array onto a cuff electrode designed for acute stimulation of the sub-millimeter nerves. (c) The TiN electrodes exhibited capacitive charge transfer, a water window of -0.6 V to 0.8 V, and a maximum charge injection capacity of 154 ± 16 µC cm-2. SIGNIFICANCE: We present the concept, fabrication and characterization of composite and flexible cuff electrodes, compatible with post-processing and MEMS packaging technologies, which allow for compact integration with control, readout and RF electronics. The fabricated TiN microelectrodes were electrochemically characterized and exhibited a comparable performance to other state-of-the-art electrodes for neural stimulation and recording. Therefore, the presented TiN-on-polyimide microelectrodes, released from silicon wafers, are a promising solution for neural interfaces targeted at sub-millimeter nerves, which may benefit from future upgrades with die-electronic modules.

Feature selection on movement imagery discrimination and attention detection.

Noninvasive brain-computer interfaces (BCI) translate subject's electroencephalogram (EEG) features into device commands. Large feature sets should be down-selected for efficient feature translation. This work proposes two different feature down-selection algorithms for BCI: (a) a sequential forward selection; and (b) an across-group variance. Power rar ratios (PRs) were extracted from the EEG data for movement imagery discrimination. Event-related potentials (ERPs) were employed in the discrimination of cue-evoked responses. While center-out arrows, commonly used in calibration sessions, cued the subjects in the first experiment (for both PR and ERP analyses), less stimulating arrows that were centered in the visual field were employed in the second experiment (for ERP analysis). The proposed algorithms outperformed other three popular feature selection algorithms in movement imagery discrimination. In the first experiment, both algorithms achieved classification errors as low as 12.5% reducing the feature set dimensionality by more than 90%. The classification accuracy of ERPs dropped in the second experiment since centered cues reduced the amplitude of cue-evoked ERPs. The two proposed algorithms effectively reduced feature dimensionality while increasing movement imagery discrimination and detected cue-evoked ERPs that reflect subject attention.

Visual gate for brain-computer interfaces.

Brain-Computer Interfaces (BCI) based on event related potentials (ERP) have been successfully developed for applications like virtual spellers and navigation systems. This study tests the use of visual stimuli unbalanced in the subject's field of view to simultaneously cue mental imagery tasks (left vs. right hand movement) and detect subject attention. The responses to unbalanced cues were compared with the responses to balanced cues in terms of classification accuracy. Subject specific ERP spatial filters were calculated for optimal group separation. The unbalanced cues appear to enhance early ERPs related to cue visuospatial processing that improved the classification accuracy (as low as 6%) of ERPs in response to left vs. right cues soon (150-200 ms) after the cue presentation. This work suggests that such visual interface may be of interest in BCI applications as a gate mechanism for attention estimation and validation of control decisions.

Comparison of EEG pattern classification methods for brain-computer interfaces.

The aim of this study is to compare 2 EEG pattern classification methods towards the development of BCI. The methods are: (1) discriminant stepwise, and (2) Principal Component Analysis (PCA)-Linear Discriminant Analysis (LDA) joint method. Both methods use Fisher's LDA approach, but differ in the data dimensionality reduction procedure. Data were recorded from 3 male subjects 20-30 years old. Three runs per subject took place. The classification methods were tested in 240 trials per subject after merging all runs for the same subject. The mental tasks performed were feet, tongue, left hand and right hand movement imagery. In order to avoid previous assumptions on preferable channel locations and frequency ranges, 105 (21 electrodesx5 frequency ranges) electroencephalogram (EEG) features were extracted from the data. The best performance for each classification method was taken into account. The discriminant stepwise method showed better performance than the PCA based method. The classification error by the stepwise method varied between 31.73% and 38.5% for all subjects whereas the error range using the PCA based method was 39.42% to 54%.