Image quality in task-based BOLD-fMRI with multi-band and multi-echo imaging protocols

Blood-oxygen-level-dependent functional magnetic resonance imaging (BOLD-fMRI) serves as a critical interface between the cognitive sciences and biology, providing a non-invasive, physically observable derivative of neural activity which changes predictably with cognitive processing. While the development of novel pulse sequences seeks to push the boundaries of spatiotemporal resolution, the transition of these protocols into standard research practice is often slowed by concerns regarding image fidelity and signal accuracy. This study investigates two prominent acquisition strategies: multi-echo (ME) and multi-band (MB) sequences. We evaluate their hypothesized improvements to BOLD signal quality through a dual framework, analyzing their performance from both a traditional biophysical perspective and their practical efficacy in task based fMRI. To evaluate the efficacy of these sequences across diverse analytical frameworks, this study utilized 2 image quality metrics (IQM) for assessing the image quality in different settings. The temporal signal-to-noise ratio (tSNR) was employed to characterize the biophysical stability and noise profile of the signal. In parallel, a Z-score derived image quality metric (ZIQM)—extracted via constrained principal components analysis for fMRI (fMRI-CPCA)—was implemented to quantify performance within the context of cognitive tasks. By integrating these distinct IQMs, this work provides a comprehensive characterization of how acquisition parameters influence data integrity across both fundamental physical and applied psychological domains. Results indicate that multi-echo acquisition effectively denoises the functional data by mitigating signal artifacts through noise averaging according to tSNR measures. However, a significant trade-off was observed: while time-weighted averaging improved the tSNR, it resulted in a measurable reduction in ZIQM. This discrepancy is attributed to the inherent dependence of MRI signal intensity on echo time (T∗₂ decay), where the averaging process may prioritize biophysical stability over the statistical variance required for cognitive modeling. Conversely, the multi-band investigation revealed no significant degradation in image quality across either metric. Contrary to prevailing concerns that multi-band acceleration introduces detrimental signal leakage, these findings demonstrate that high-acceleration protocols can be employed to enhance spatiotemporal resolution without compromising tSNR or ZIQM. Consequently, this study suggests that multi-band sequences offer a robust, high-efficiency alternative for rapid data acquisition without the fidelity trade-offs associated with multi-echo averaging. This provides an insight for further development of multi-band sequences, as it reduces the of cost of imaging time (and so run time cost), while ensuring same level of image quality. Making it a superior protocol for BOLD-fMRI studies that requires large field of view and high spatiotemporal resolution.