Spatial and Temporal Quality of Brain Networks for Different Multi-Echo fMRI Combination Methods

Abstract

The application of multi-echo functional magnetic resonance imaging (fMRI) studies has considerably increased in the last decade due to superior BOLD sensitivity compared to single-echo fMRI. Various methods have been developed that combine fMRI data derived at different echo times to improve data quality. Here, we evaluated five multi-echo combination schemes: 'optimal combination' (OC, T2 *weighted), T2 * -FIT (T2 * -weighted, calculated per volume), average-weighted (Avg), temporal Signal-to-Noise Ratio (tSNR) weighted, and temporal Contrast-to-Noise Ratio weighted combination. The effect of these combinations, with and without additional postprocessing, on the quality of functional resting-state networks was assessed. Sixteen healthy volunteers were scanned during a 5-minutes resting-state fMRI session. After network extraction, several quality metrics in the temporal and spatial domain were calculated for their respective time-series and spatial maps. Our results showed that OC and T2 * -FIT outperformed the other methods in both domains. Whereas the OC and T2 * -FIT time-series were found to be the least associated with artifacts, OC resulted in the highest quality spatial maps. Furthermore, spatial smoothing, bandpass filtering and ICA-AROMA merely improved networks derived from the least performing combinations (Avg and tSNR). Because similar network quality was obtained following OC and T2 * -FIT without postprocessing, we recommend future studies to implement these combinations without these postprocessing steps. This minimizes the amount of image modifications and processing, potentially leading to enhanced BOLD contrast. The results highlight the benefits of T2 * -weighted multi-echo combinations on resting-state network quality and raise its potential value in dynamic fMRI analyses or for diagnosis and prognosis purposes of neuropsychiatric disorders.INDEX TERMS Multi-echo imaging, resting-state, fMRI, brain networks