Reliable MEG metrics of cortical language function

For brain imaging the reliability of the findings is the cornerstone for their precision. However, the reliability of brain imaging findings has been found to be variable. This thesis identified reliable magnetoencephalography (MEG) evoked responses, modulation of oscillations, and functional connectivity related to language production. In addition, the thesis studied the factors affecting reliability, as well as reliability at the level of individual participants. Study 1 identified reliable evoked activity and modulations of oscillations related to picture naming and semantic judgment, while Study 2 identified reliable functional connectivity related to picture naming. Reliability was assessed using Intraclass Correlation Coefficient (ICC). The naming task showed the reliability of the left-hemisphere evoked activity sequence described in previous MEG studies and in a neurocognitive model of picture naming. The naming task also demonstrated reliable beta oscillatory suppression in premotor and sensorimotor regions showed in previous MEG studies. For the naming-related functional connectivity, a reliable beta connectivity increase in the left motor and frontal regions was found. In the semantic judgment task, the spatial extent of reliable evoked activity and oscillatory modulation was more limited. It is important to query the factors that might influence the reliability. Study 2 showed that the reliable functional connections were linked to stronger functional connectivity, as well as to stronger structural connectivity and shorter structural path length than the non-reliable connections. These findings may suggest that the most efficient communication between brain regions happens especially in reliable connections. It was also found that the reliable connections were linked to language performance, providing additional validation of the findings. Study 3 assessed the reliability (ICC) of resting state sensorimotor power spectral characteristics across a range of analysis parameters. It was found that these characteristics were reliable across different parameters. However, the reliability was affected by variability across individuals. Current methods, such as ICC, mostly assess reliability at the group level. New methods to identify reliable activation at the level of individual participants are needed. Study 4 developed a method to extract reliable individual-level patterns of evoked responses during picture naming. The Study revealed notable variability among participants, highlighting the need for individual measures of reliability. In summary, this thesis demonstrated different MEG metrics giving complementary reliable information on language production in the brain. When studying reliability, this thesis suggests that it might be beneficial to utilize the brain's functional and structural information, and also behavior-based validation. Furthermore, this thesis highlights the need for individual-level measures of reliability.