Invited Symposium: Nonlinear Dynamical Systems in Psychiatry
Materials and Methods
1. Subjects Three depressed patients (Mrs. G., Mss. H. and Mrs. R.) and a control subject (Mrs. I.) gave their written consent to participate to this research. The depressed patients stayed in La Salpêtrière Hospital (Paris) and received either pharmacological treatment or ECT in regard to depression intensity and resistance. They were selected according to the DSM-IV (ref. 1) criteria for a major depressive episode. Their scores obtained to depression rating scales assessed their depressive state at the arrival (Table 1). In the case of the control subject, depression scales proved the absence of depressive symptoms. Table 1 : Scores to global rating scales of depression (HAMD-21 & MADRS), to psychomotor retardation (ERD), and anxiety (COVI) scales performed upon arrival and discharge of each patient.
ARRIVAL / MADRS HAMD-21 ERD COVI / DISCHARGE / MADRS HAMD-21 ERD COVI ---------------------------------------------------------------------- Mrs G. / 27 25 8 22 / Mrs G. / 12 11 3 18 Mss S. / 20 21 12 4 / Mss S. / 10 10 5 2 Mrs R. / 24 20 25 4 / Mrs R. / 8 9 15 22. Experimental protocol Subjects have been examined every two days during their hospitalization (between two and three weeks). Clinical evaluation Before each EEG-recording, the subjects filled the BfS' scale questionnaire (ref. 14). This scale permits a scalar evaluation (between 0 and 56) of depressive mood intensity from " relaxed " (low scores) to " extremely depressed " (high scores). EEG-recording EEG has been recorded, in a resting eyes closed condition, on 31 derivations referred to the ears and set on the scalp in an equidistributed manner. 3. Methods Non-linear quantification For each recording session, ten 8-second multi-channel EEG-segments free of artifacts were selected. A total of 325 EEG-segments were thus used in this study. Each EEG-segment was analyzed using a numerical method based on multi-channel recordings and non-linear forecasting (ref. 9, 10). It permits one to obtain local and global entropies which quantify the loss of predictability for each of the 31 recording sites (k) and for the global EEG-recordings (K). In turn, an increase of entropy corresponds to less predictable signals and thus to a more complex dynamics. Since signals generated by linear stochastic processes can exhibit similar characteristics as non-linear deterministic systems (ref. 12), the presence of non-linear structure in EEG-segments was ensured by testing the significance of the difference between K obtained for raw data and K obtained for a set of 39 multi-channel surrogate data (ref. 11). Principal component analysis (PCA) PCA was used to better characterize the neuronal changes involved in the remission of depressive episode. It was computed on the data sets composed of 31 entropy measurements obtained for all the EEG-segments and for each subject. We obtained spatial pattern of the dynamics by projecting each principal component score onto the electrode space (Fig. 3).
| Discussion Board | Next Page | Your Symposium |
|Thomasson, N; (1998). Brain Dynamics Bifurcation and Remission of Depressive Episode. Presented at INABIS '98 - 5th Internet World Congress on Biomedical Sciences at McMaster University, Canada, Dec 7-16th. Invited Symposium. Available at URL http://www.mcmaster.ca/inabis98/sulis/thomasson0507/index.html|
|© 1998 Author(s) Hold Copyright|