Spread-Spectrum Magnetic Resonance Imaging

Experimental setup of spread-spectrum imaging. The MR scanner (top left) controls the timing and execution of the gradient echo sequence (top right) and sends a trigger before each MR signal acquisition period to the current waveform controller. Eight independent current waveforms are generated, amplified by current amplifiers and applied to the eight local coils during the MR acquisition period, i.e. during the flat top of the read gradient. For example, current modulations can be phase shifted sine waves (like in the measurements performed in this study) or even noise patterns. The modulated and phase encoded gradient echoes together with the knowledge of the magnetic field induced by the local coils and its corresponding current time courses are used for image reconstruction. Like in conventional parallel imaging, acceleration of acquisition time is achieved by measuring only every n-th phase-encoded echo.

Experimental setup of spread-spectrum imaging. The MR scanner (top left) controls the timing and execution of the gradient echo sequence (top right) and sends a trigger before each MR signal acquisition period to the current waveform controller. Eight independent current waveforms are generated, amplified by current amplifiers and applied to the eight local coils during the MR acquisition period, i.e. during the flat top of the read gradient. For example, current modulations can be phase shifted sine waves (like in the measurements performed in this study) or even noise patterns. The modulated and phase encoded gradient echoes together with the knowledge of the magnetic field induced by the local coils and its corresponding current time courses are used for image reconstruction. Like in conventional parallel imaging, acceleration of acquisition time is achieved by measuring only every n-th phase-encoded echo.

In conventional imaging, the scan of k-space (which is the Fourier transformed image space) is achieved by applying linear gradients along the principal axes. In most applications, k-space is acquired line by line on a Cartesian grid, or in some implementations along projections or spirals, to name just a few. The application of additional rapid magnetic field modulations during scan of k-space has already been demonstrated in the wave-CAIPI and the FRONSAC approach. In wave-CAIPI, which is an extension of bunched phase encoding in combination with CAIPI, additional phase-shifted sinusoidal modulations are applied to phase encoding gradients during the readout. This converts the original k-space scan along a straight line into an extended corkscrew trajectory. The resulting improved distribution of k-space sampling points can be used to accelerate image acquisition if combined with parallel imaging. A similar principle is used in FRONSAC, except that the additional field modulation is achieved with global second order non-linear shim gradients superimposed to the underlying k-space trajectory produced by linear gradients. The simulated results achieved with FRONSAC demonstrate a faster coverage of k-space (along a similar corkscrew or oscillating trajectory as in wave-CAIPI) and thus increased imaging speed.
Here we present a novel concept to further boost MR imaging speed. Spread-spectrum MRI is based on the rapid dynamic modulation of local magnetic fields produced by an array of local current loop fields instead of using global field modulations via gradient or shim coils as done in wave-CAIPI or FRONSAC. These fields are modulated dynamically during signal acquisition to imprint local and unique signal characteristics into the spin distribution, which can be interpreted as a unique fingerprint onto confined regions within the object. Spread-spectrum MRI distributes or spreads the basic bandwidth of gradient-encoded spin frequencies using distinct modulation frequencies (or even orthogonal noise patterns) originating from a certain spatial location of the object. This spatially unique information is then utilized to disentangle different parts of the object, and thus to drastically boost imaging speed. In other words, spread-spectrum MRI combines local non-linear encoding with rapid modulation of k-space trajectories.

Klaus Scheffler, Alexander Loktyushin, Jonas Bause, Ali Aghaeifar, Theodor Steffen, Bernhard Schölkopf:

Spread-spectrum magnetic resonance imaging.

Magn Reson Med. 2019 Sep;82(3):877-885.

DOI