Análisis no lineal de registros magnetoencefalográficos de pacientes con Trastorno obsesivo compulsivo mediante la complejidad de Lempel-Ziv

Zusammenfassung

Obsessive compulsive disorder (OCD) is characterized by the presence of obsessions and compulsions that cause emotional distress and affect different aspects of a patient´s life which are clinically significant. The main feature of OCD is obsession, understanding as such any thought, feeling, idea or image, unusual and repetitive despite the efforts to ignore or confront them. The second and also essential element is compulsion: every repetitive and stereotyped action done to mitigate the anxiety that stems from obsessions. With the compulsion the subject obtains temporary relief, but it is not in itself pleasant and may cause more distress than the obsessions. OCD is classified within anxiety disorders in ICE10 and DSM-IV. This is due to: -the presence of obsessions, that cause anxiety and generate compulsions, with the objective of relieving anxiety. -Both in OCD and other anxiety disorders avoidance behaviour appear, aimed at preventing situations which cause anxiety. However, its classification is frequent cause of discussion between those who consider that its symptomatology, course, epidemiology and neurobiological findings differ considerably from the rest of anxiety disorders. An important group of investigators proposed the creation of an Obsessive Compulsive Spectrum, which would include OCD but also other disorders such as Tourette syndrome body dysmorphic or pathological gambling. The biomedical signal processing is directed to the development of specific algorithm for every specific kind of signal, to reach a more accurate diagnosis (2). The biomedical signal processing is divided in several steps: registration, processing and classification (3). Electrocardiography, phonocardiography, electroencephalography (EEG), magnetoencephalography (MEG) or electromyography are examples of biomedical signals. The information of these biomedical signals may not be obtained immediately, needing complex signal processing tools (4). The main objectives of biomedical signal processing are (4): -reduce the subjectivity of manual measurements. -extract features in order to characterize information contained in a signal. -noise reduction. -mathematical (signal) modelling and simulation for a better understanding of physiological processes. In the present work we are going to analyse magnetoencephalograms, which are a record of magnetic fields produced by electrical activity within the brain. In spite of the fact that neurophysiological techniques have broadly been used in the investigation of mental disorders, and although they made possible the creation of statistical models with high sensitivity and specificity, their use in every day practice has been quite limited. Surely, this is partly due to the incorrect choice of analysis process, which would bring about the loss of significant information contained in the brain signal (5). In this point, the nonlinear analysis methods, such as complexity become important for extracting clinically significant information hidden in the signal. The analysis of complex parameters, obtained from EEG or MEG, estimates the predictability of brain oscillations and/or the number of oscillators that produce the recorded signals (5). Moreover, since complexity parameters are sensitive to the temporal components of brain activity, their study may allow us to reach a better understanding of the dynamical nature of psychiatric disorders. We chose the Lempel-Ziv complexity (LZC) algorithm, which works by scanning through the input string for successively longer substrings until it finds one that has not previously been registered. This algorithm has been used to assess several mental disorders such as Alzheimer, Schizophrenia, Depression or ADHD. In the present work we analysed the MEG recordings of 13 patients diagnosed of OCD (DSM and ICD) and 60 healthy controls, using the LZC algorithm. The results show a positive correlation between age and LZC values within controls and OCD patients. LZC values were smaller in the OCD group although the regression coefficient in the OCD group was bigger than in the control group, showing a completely different behaviour of LZC among OCD and control group. An asymmetry between both lateral regions was found in the OCD sample but not in the controls. Moreover, the frontal region of OCD showed alterations that could be related with previous findings in the frontal lobe.