Evaluating parameter selection and analysis approaches on quality and reproducibility of functional MRS

Functional magnetic resonance spectroscopy (fMRS) extends conventional MRS by acquiring data while participants receive stimuli or are engaged in a task, with analysis focused on segmenting data to align with stimulus- or task-related metabolite changes. In this study, we propose and systematically evaluate three fMRS analysis pipelines: block, event-related, and sliding window approaches, to optimise parameter selection and assess reproducibility with respect to data quality. Using empirical and simulated fMRS datasets, we examine the impact of data quality parameters and investigate the influence of the number of transients per block/window on the trade-off between reproducibility and temporal resolution with respect to data quality. Our results show that while the number of transients required for quantification can be reduced for Glx quantification in both block and event-related analyses, a minimum of 32 transients is required for reliable GABA+ measurement using spectral editing sequences. Optimal data quality, characterized by low noise and a spectral linewidth of 6-8 Hz, is preferred, especially for analyses with fewer transients. These findings highlight the importance of balancing data quality and acquisition parameters to ensure robust fMRS outcomes. Additionally, they provide a framework and consideration for implementing high temporal resolution analysis of GABA and glutamate, positioning fMRS as a powerful tool for advancing our understanding of neural mechanisms underlying brain function in both health and disease.