Analysis of signal-to-noise behavior in Cartesian continuous sampling sequences: predictions and experimental validation of opportunities for improved image SNR.

Image signal-to-noise ratio (SNR) limits many MRI applications. Here we propose a method to improve SNR based on continuous sampling (CS) during each TR without significant increases in acquisition time. The general concept of CS is defined here as sampling the NMR signal immediately after slice excitation including ramp times, both the dephase and rephase lobes, the phase-encoding (PE) gradient application, and the slice refocusing gradient. This study analyzes several cases of CS and demonstrates a specific case where sampling occurs during an isolated and balanced readout gradient in order to increase SNR in a rectilinear Cartesian sequence without significantly increasing overall acquisition time. The noise correlation consequences of rectilinear CS are mathematically derived and proven through simulation. The SNR improvement of up to approximately 40% measured in both phantom and asymptomatic human volunteer images is comparable to theoretical prediction of increased SNR proportional to the increase in the square root of the sampling time.