Data acquisition speed is one of the major limitations of MRI, as the data points are acquired sequentially in MRI. At the same time, many clinical applications require fast scanning either due to physiological processes they are trying to depict (e.g., cardiac motion), need to capture first pass of contrast agent, or requirement of patient breath-hold to reduce motion. We will begin by demonstrating effects of faster sampling through reduction of the number of acquired k-space lines such as loss of resolution or aliasing artifacts. We will then introduce parallel imaging reconstruction as a way to resolve the aliasing using spatial differences in sensitivities of individual elements in multi-channel receiver arrays. We will also discuss two representative parallel imaging approaches that compensate for the missing data while operating in either image space and or k-space. We will pay special attention to noise properties of parallel MRI. The reduction in the acquired data and the reconstruction lead to noise increase, which is two-fold. First, there is a fundamental loss of signal-to-noise ratio (SNR) that is proportional to the square root of reduction factor and is common for all techniques. Next, there is SNR loss during the image reconstruction process traditionally described by a spatially varying geometry factor, or g-factor, which depends on a number of factors, such as coil geometry, scan orientation, number of coils, and overall reduction factor. We will discuss the limitations that SNR loss places on parallel imaging acceleration as well as other practical considerations in the usage of parallel MRI techniques. We will finish with examples of some pulse sequence types and clinical applications that can benefit from the use of parallel MRI.

Acknowledgements

No acknowledgement found.

References

No reference found.