Using ANNs to predict a subject’s response based on EEG traces

Effectiveness of different sensing modalities in predicting targets of reaching movements

“…Individual modalities EEG allows early prediction of the reach target, giving an accuracy of 72.8% for 2-object trials and 54.1% for 3-object trials a second after objects appear. This accuracy is similar to findings in literature: Logar et al [14] found 75% accuracy when predicting true/false responses while Lee [5] found 40% accuracy when predicting among 4 reach targets.…”

Visuo-Motor Tasks in a Brain-Computer Interface Analysis

“…All of these methods are relatively simple; however, to find the optimal methodology constellation that yields the optimal gripping-force or wrist-movement estimations for the forthcoming task trials, the EEG measurements underwent several combinations of signal-processing procedures, parameter fitting, optimization and fuzzy model options. Similar methods of signal processing have proved to be suitable for extracting the EEG information from working-memory tasks (Logar, Belič, Koritnik, Brežan, Zidar, Karba & Matko, 2008), and now we have shown that they can also, with some modifications, be used for extracting the information from VM tasks. In order to use such a methodology for the information decoding of VM tasks the required modifications include a replacement of the model's parameters to comply with the theory of the brain's visuo-motor integration.…”

“…Therefore, for the needs of a different cognitive task, the filtering intervals and carrier-wave frequencies need to be adapted to meet the needs of a motor task instead of a working-memory task. Briefly, this means that all the frequency parameters that were placed in the theta frequency band (Logar, Belič, Koritnik, Brežan, Zidar, Karba & Matko, 2008) had to be shifted to the beta frequency band and precisely re-fitted. Parameter re-fitting proved to suit the needs of the static VM task data processing; however, to handle the data of a more complex dVM task an extension of the data processing had to be performed.…”

“…As is generally known, many proposed BCIs need extensive training for the subjects so that they can gain control over their brain rhythms in order to properly use the BCI (Neuper & Pfurtscheller, 2001;Wolpaw et al, 1991). However, the approach presented in this paper deviates from the subject-training ideas and is instead based on EEG data pre-processing and fuzzy classification, which does not need any preliminary subject training (Logar, Belič, Koritnik, Brežan, Zidar, Karba & Matko, 2008;Logar, Škrjanc, Belič, Brežan, Koritnik & Zidar, 2008). The proposed methodology, which is capable of interpreting the measured EEG information in a certain predefined action, uses different beta brain-rhythm filters, phase demodulation and a principal component analysis for the signal pre-processing.…”

Modelling in systems biology, neurology and pharmacy

“…Although it is not reasonable to expect that in a completely idle state the brain would produce single-frequency sine wave oscillations, using the concept, one should be able to identify the most dominant tasks that are performed in the brain at the time of observation. To test the concept, EEG data from five healthy volunteers, performing different hand-gripping tasks [60,61], and of three healthy volunteers performing working memory tests [62] were used. The EEG signals were phase-demodulated and fed into artificial neural network to predict corresponding activity of the persons.…”