Abstract. Anxiety patients that are presented with neutral and panic disorder triggering stimuli shon. different event-related brain potentials (ERP) within the electroencephalogram (EEG). In this paper, we investigate this difference by time-frequency (TF) revealing transforms leading to an identification of a small number of significant parameterising coefficients to be able to differentiate between the presented stimulus categories.
I N T R O D U C T I O NIndividuals with panic disorder are characterised by an abnormal fear of certain anxiety connected sensations such as palpitation, breathlessness, or dizziness [l]. The research into this disorder has led to studies investigating its symptoms by means of appropriate stimulation and measurement of the subsequent ERP [2, 31. In this context, visual stimulation has been performed with words causing panic disorder, whereby the EEG can be recorded showing event related potentials. Previous studies have resulted in revealing a low frequent transient waveform appearing approximately 300 ms after stimulus onset as a distinctive characteristic nhich is referred to as P300.Analysis of variances (ANOVA) [4] is one method of detecting the P300 in panic disorder and normal response ERP. Since the P300 has a transient behaviour, the application of time frequency (TF) analysis appears well suited, as it takes both spectral and temporal information into account 151. In this paper we aim to investigate various transforms -such as wavelet, wavelet packet, and Gabor transforms -with respect to their suitability for revealing the TF characteristics of the transient P300. We further optimise these transforms such that the distinction between panic disorder and normal responses is concentrated in only few transform coefficients, to which a statistical test can be applied.The paper is organised as follows. Sec. 2 will introduce the background and experimental conditions under which panic disorder data was obtained. In Sec. 3 suitable TF transforms will be reviewed, which can parameterise the elicited event related potentials, while Sec. 4 discusses a method to isolate indicative parameters, which can be used for distinguishing between panic disorder and normal EEG. Finally, test results and conclusions are presented in Secs. 5 and 6.
PANIC D I S O R D E R E R PThe panic disorder ERP were measured for an anxiety patient who was presented with fear-inducing or neutral words tachistoscopically at the perception threshold of panic disorder. The patient's perception threshold for correctly identifying 50% of the words vas determined with neutral words not used in t,he experi: ment. It can be assumed that the patient will recognise a greater number of anxiety words given at his perception threshold than neutral words [4]. Thus, it can be expected that the EEG exhibit an difference when neutral and anxiety words are presented.The EEG was measured at the vertex electrode (Cz) synchronously to the stimuli, whereby the recordings were started 100 ms before the onset of the visual word st,im...
