Image Quality Issues

Abstract

In this chapter, we first describe the main MRI data acquisition methods and some more recent advances. The requirement of a fMRI pulse sequence is BOLD sensitivity, which means predominantly T2*-weighted sequences such as gradient echo-echo-planar imaging (GE-EPI), although spin echo sequences such as spin echo-EPI (SE-EPI) can also be used (Bandettini et al., NMR Biomed 7:12–20, 1994; Norris et al., Neuroimage 15:719–726, 2002; Schmidt et al., Neuroimage 26:852–859, 2005). We therefore describe the main limitations of GE-fMRI in terms of image artefacts (distortion, dropout, blurring and ghosting) highlighting potential enhancement of these problems that EEG equipment can introduce. Approaches to optimise image quality are then described in relation to these artefacts. Recent developments that have become increasingly popular are sequences that utilise ‘parallel imaging’ in the slice direction, where RF receiver coil spatial sensitivity is used to encode spatial position in addition to imaging gradients. This can either reduce the slice thickness while maintaining the TR to increase resolution while reducing through-slice dropout or be used to increase scan repetition rate (decrease TR). This approach has particular relevance at higher field strengths where signal dropout becomes more problematic. The dramatic reduction of the TR to increase fMRI signal sampling rates by an order of magnitude remains an approach where the benefits are unclear owing to signal loss and image quality degradation. However, where full sampling of physiological noise signals is required (particularly important for resting-state fMRI), it may confer net benefit. Nevertheless, the limitations of rapid signal loss where TRs fall below tissue T1s and noise enhancement from the poorly conditioned problem of separating signal aliased from many different slices make it likely that a modest reduction in TR using an SMS factor of 2–3 is optimal for many applications. Although not yet mainstream, 3D echo-planar imaging is being used increasingly, and as the technical challenges associated with segmented acquisition are solved by improved acquisition and reconstruction methods, this is likely to change over the coming years. fMRI is limited by its sensitivity to noise sources particularly motion and physiological noise, and these issues and methods for their mitigation are summarised. Lastly, there has been a notable increase in the application of multi-echo EPI which may allow for a much better separation between BOLD activity and physiological noise. In the last part of the chapter, we introduce and describe fMRI quality assurance methods that can be used to optimise imaging parameters and monitor performance during studies.