Development Of A Lie Detector

by Chin Chuan ChenSemester 2, 2004/2005

Attempt to detect P300 is carried out using visual stimulus made of two capital alphabets. Four subjects (two males and two females) aged 21 to 25 are involved in the experiment, designed in an oddball paradigm comprising of Targets (‘X’) and Irrelevants (‘O’). Targets are prompted at a probability of 0.2.
Fz, Cz, Pz and Oz montages are used to obtain EEG signals. ECG, throat EMG and EOG signals are collected in the experiment to monitor artifacts. EEG data contaminated with eye artifacts are removed through visual inspection. Data are extracted one second before to one second after the response tag. These are filtered digitally with an IIR lowpass filter, at 3.5 Hz cutt-off frequency. Filtered data are separated into a matrix of Targets and a matrix of Irrelevants for separate signal processing.
The P300 detection program uses the filtered and baseline corrected input signal to detect the presence of P300 based on its amplitude and length of the P300 complex. The total classification rate of the program is 66.6%.

Supervisor: Ng Siew Cheok