Análise de registros eletroencefalográficos de pacientes comatosos: estudo quantitativo para definir estatisticamente a maneira mais apropriada de agrupar pacientes em coma

Abstract

The electroencephalogram (EEG) clinical examination is essential in the Intensive Care Unit (ICU) environments, and is commonly used in comatose patients, since it provides essential information regarding brain electrical activity. The main causes of hospitalization of patients in ICU’s are caused by the etiologies of Cranioencephalic Trauma (TBI), Stroke (Stroke) and Metabolic Coma (CM), which together with EEG techniques, are assessed by the degree of consciousness using scales such as the Glasgow Scale (ECG) and the Ritchmond Agitation Sedation Scale (RASS). Based on the EEG signals collected from these patients, with different characteristics, the present study aims to verify quantitatively, through the use of three different quantifiers, the degree of linear correlation between patient data, when they are submitted to two criteria distinct, in order to identify the best among them. 75 EEG signals (0-40Hz) measured in comatose patients at Hospital das Clínicas, Federal University of Uberlândia, and 30 neurologically healthy individuals were analyzed. The EEG stretches were measured by different quantifiers that measure normalized and percentage power, as well as the symmetry of the brain signal. The correlation calculation was used in order to measure the degree of correlation in the groups performed, using two criteria, the etiology, and the degree of consciousness (assessed by the Glasgow and RASS scales), being compared with the control group. The results obtained show that the groups formed by level of consciousness lead to a better level of statistical correlation than when grouped by etiology, with the control group as the maximum correlation pattern. The conclusions suggest that the quantitative assessment of EEG’s of comatose patients of different characteristics is based on the grouping by level of consciousness, where it will be possible to obtain a greater similarity of the data of each group, and thus present more reliable results.