Post-stroke aphasia rehabilitation using an adapted visual P300 brain-computer interface training: improvement over time, but specificity remains undetermined

Abstract

Introduction: This study aimed to evaluate the efficacy of visual P300 brain-computer interface use to support rehabilitation of chronic language production deficits commonly experienced by individuals with a left-sided stroke resulting in post-stroke aphasia.

Methods: The study involved twelve participants, but five dropped out. Additionally, data points were missing for three participants in the remaining sample of seven participants. The participants underwent four assessments-a baseline, pre-assessment, post-assessment, and follow-up assessment. Between the pre-and post-assessment, the participants underwent at least 14 sessions of visual spelling using a brain-computer interface. The study aimed to investigate the impact of this intervention on attention, language production, and language comprehension and to determine whether there were any potential effects on quality of life and well-being.

Results: None of the participants showed a consistent improvement in attention. All participants showed an improvement in spontaneous speech production, and three participants experienced a reduction in aphasia severity. We found an improvement in subjective quality of life and daily functioning. However, we cannot rule out the possibility of unspecific effects causing or at least contributing to these results.

Conclusion: Due to challenges in assessing the patient population, resulting in a small sample size and missing data points, the results of using visual P300 brain-computer interfaces for chronic post-stroke aphasia rehabilitation are preliminary. Thus, we cannot decisively judge the potential of this approach.

Keywords: P300 – event related potential; aphasia; brain-computer interface (BCI); quality of life; rehabilitation; stroke.

Figure 1

Assessment overview. Each participant underwent a total of 32 sessions: t0, t1, t2, and t3, with three sessions each (S1, S2, S3) and 20 sessions of BCI training. Based on individual needs, up to 7 additional therapy sessions were offered. TAP, Test Battery for Attention Performance; SASS, Social Activity Self-rating Scale; ASF, Aachen Self-efficiency Questionnaire; BDI-II, Beck Depression Inventory-II; SF-36, Short-Form-36 Health Survey; AFIB, Aachen Functioning Item Bank; AAT, Aachen Aphasia Test; ANT, Attention Network Task; SART, Sustained Attention to Reaction Task; VO, Visual Oddball Task.

Figure 2

Spelling matrix in the facilitated version and the regular version. Stages of the visual P300 speller grid 6×6 containing letters A to Z and numbers 0 to 9 cropped in half for convenience. The cardboard condition is shown in the upper half (quadrants A,B), while quadrants C,D show a standard P300 display. The instruction is to spell the word BRAIN letter per letter, in this case, starting with the letter B. Participants always see the target inside the matrix (quadrant A) in the facilitated condition. During visual stimulation, the target is intensified stronger (quadrant B). Targets are not highlighted in the standard P300 setup (quadrant C) and do not stand out from other stimuli when highlighted during intensification (quadrant D).

Figure 3

P300 amplitudes at electrode position Pz. Black = targets, dark grey = non-targets. Target end refers to session 20 for participants A, B, C, and F; to session 14 for participant D, to session 12 for participant E and to session 15 for participant G.

Figure 4

(A–E) Reaction time of all participants for each comparison of each behavioral test across sessions. Box-plots indicate quartiles, and whiskers extend until 1.50 interquartile range. (A–C) For the comparisons “Alerting”, “Orienting” and “Executive function” of the ANT test. (D) The “Inhibition” comparison of the SART test. (E) The “Oddball” comparison, reporting the reaction time press the button during deviant stimuli. (F) shows the P300 EEG amplitude between 250 to 600 ms post stimulus expressed in micro-Volts between deviant and repetitive cues. These plots illustrate the contrasts that were statistically evaluated between sessions. No significant interaction was found.

Figure 5

T value changes in the AAT repetition subtest for individual participants.

Figure 6

T value changes in the AAT Written Language test for individual participants.

Figure 7

Raw value changes in the AAT Spontaneous Speech test for individual participants. Raw values range from 0 (worst) to 30 (best) points.