Brain computer interface as a forensic tool

Abstract

Event related potential can provide important information about the human memory content. One can use ERP signals to determine if a piece of information is significant to a person or not. This property of ERP appears to be greatly useful to a wide range of applications. The use of ERP as a forensic tool is being investigated in this paper. The used experiment paradigm, data collection method, and signal processing techniques are described. The classification results for the different feature extraction and classification techniques are reported. The results obtained lead to an optimistic view of the systempsilas reliability. The maximum classification rate obtained was 90% using the minimum (Euclidean) distance classifier. The rate was achieved using the band Fourier estimation alpha band parameter applied to the signals from the Pz site. Moreover, innocents were a 100% correctly classified and criminals were 80% correctly classified. Numerous other features scored more than 80% accuracy using the various classifiers.