Phase Segment Analysis Alleviates Contamination from Similar Frequency Targets in a 35 Target Phase Discriminating SSVEP BCI

Recently brain-computer interfaces (BCI) utilising the steady state visual evoked potential paradigm have begun to employ previously unutilised phase information [1, 2]. Absolute phase response can be used to discriminate between targets that flash at identical frequency rates but are offset in phase of their alternation rate [3]. In traditional frequency tagged SSVEP BCI’s SSVEP response need only to be distinguished from ongoing brain rhythms and background EEG. However in phase tagged BCI’s, frequency similar targets are subject to contamination from not only ongoing brain rhythms but crucially the other non-attended targets that share alternation rates within the system. A main motivation for phase tagging is to increase the number of stable targets presentable on typical computer monitors that operate at a 60Hz refresh rate. Screen realty is typically limited in terms of physical size and resolution. Thus increasing the number of targets dictates reducing the size of individual targets or the separation between them, increasing the likelihood of cross contamination. This paper introduces a vector segmentation analysis method that operates on single cycle Fourier components to classify phase angle and compares this to a phase weighting/projection approach. The comparison is facilitated by an offline study with five subjects using a 35 phase tagged target system.