Test-Retest Reliabilität von fMRT-Paradigmen: Wortgenerierung und Subliminale Präsentation emotionaler Stimuli

Background: Since the early 1990s the implementation of functional magnetic resonance imaging (fMRI) in the field of medical diagnostics has fundamentally changed the scientific landscape of psychiatry and neurology. Ever since, assumptions about the neurological correlates of human cognitive capabilities evolved quickly. Soon, the transfer of that knowledge to therapeutic means approached realistic dimensions. Yet, scientific standards for the measurement and interpretation of those results have not been developed accordingly (Bennet and Miller, 2010). It is against this background that this study aims to examine the reliability of the fMRI measurement in different cognitive fields. The paradigms of the thesis at hand are situated within the broader context of the study “Effects of genetic risk variants for psychiatric disorders on brain structure and function in healthy and affected subjects”. Using fMRI paradigms in the temporal processes of healthy and affected individuals, this study evaluates, amongst others, neuronal activity and functional connectivity with a view to risk variants. Being longitudinal in nature, it is essential that this study be based on reliable and comparable fMRI data, which is collected over a long period of time. Methods: In a test-retest-procedure, applied across an interval of 28,4 ± 6,0 days, 20 healthy individuals were examined, using the fMRI measurement technique. The data of both measurements was then used to qualitatively review the overlap of the activated areas. Subsequently, with the aid of the so called ICC (Intra-class Correlation Coefficient), this study ran a statistic calculation analysis. Thereafter, based on the ICC values, the reliability of the activation coefficients was assessed in view of their reproducability. Overall, the measurements reviewed five different domains of cognitive brain functions by six paradigms: the activation network at rest (Resting State); semantic word generation (Verbal Fluency, WG); the episodic memory (Encoding and Retrieval); subliminal presentation of emotional stimuli (Subprime, SPES); and the working memory (n-back). The thesis focused on examining the reliability of the WG-paradigm as well as the SPES-paradigm. Results: Both the WG-paradigm as well as the SPES-paradigm mainly showed activations in expected brain areas. However the evaluation of the qualitative presentation by the cluster-overlap method and the calculation analyses showed considerable differences. The WG-paradigm generated considerably more activation during the control period of the experiment. In the cluster overlap method mainly the deactivated areas interleaved. Statistically, the calculated reliabilities confirmed the overall deficient reliability of the paradigm. As for the SPES-paradigm, the analysis showed characteristic activations in the processing of faces. The analysis of reliabilities of the activated and deactivated network showed comparatively high median ICC values in the range of moderate to good reliability. To the contrary, the examination of subliminal stimulated processes of emotions with the example of the emotion fear only showed low activities and in characeristic regions a low reliability. In summary, it can be stated that the reliability of the SPES paradigm can be judged moderate and good. Discussion: When seeking to examine the transferability of the generated results in light of the overall contextual study mentioned above as well as other groups of subjects, various aspects need to be considered: At first, it needs to be clarified that the variance of the acquired fMRI data depends essentially on experimental and technical conditions as well as the experimental design (Bennet und Miller, 2013). Despite good technical conditions, experimental deficits regarding the group of subjects emerged in this study. Furthermore, various deficiencies in the experimental design of both paradigms affected the low reliability. At the same time, the aspect of analysing reliability by ICC values has to be questioned. Even though it represents the current state and latest practice in the evaluation of reliability, it still shows considerable deficits in the transfer to different groups of subjects (Fliessbach et al., 2010). Conclusion: With a view to other studies, the results of the investigated paradigms have to be interpretated with caution. In regard of future scientific findings that are based on fMRI data it will be essential to standardize the measurement of reliability. The aim should be to develop and systematize a solid basis for prospective diagnostic and therapeutic implementation.