Variation of noise in multi-run functional MRI using generalized autocalibrating partially parallel acquisition...

Area of Science:

• Radiology
• Neuroimaging
• Medical Physics

Background:

• Multi-run functional MRI (fMRI) is crucial for neuroscience research.
• Generalized Autocalibrating Partially Parallel Acquisition (GRAPPA) accelerates fMRI acquisition.
• Understanding noise variation is key for reliable fMRI data.

Purpose of the Study:

• Investigate noise variation in multi-run fMRI scans acquired with GRAPPA.
• Identify the primary cause of observed noise fluctuations between scanning sessions.
• Assess the impact of GRAPPA acceleration factors on noise stability.

Main Methods:

• Simulated multi-run fMRI exams using a phantom with continuous scanning.
• Evaluated noise variation across different GRAPPA acceleration factors.
• Conducted real fMRI experiments with human subjects at acceleration factor two.
• Analyzed noise variation causes via offline reconstruction and numerical simulations.

Main Results:

• Image noise distribution is stable within a single run but varies randomly between runs.
• Noise variation was confirmed in both phantom and human subject fMRI experiments.
• Reconstructing images using consistent reference scan data significantly reduced noise variation.

Conclusions:

• Noise patterns can indeed change across different fMRI runs, affecting both phantom and human data.
• The primary cause of this run-to-run noise variation is attributed to random noise within the separate reference scans used by GRAPPA for each run.