Diffusion Acceleration with Gaussian process Estimated Reconstruction (DAGER).

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Medical Imaging Technology
• Diffusion MRI

Background:

• Image acceleration in diffusion MRI offers benefits like reduced distortion and increased direction acquisition.
• High acceleration factors can lead to ill-conditioned reconstruction problems, especially with combined in-plane and simultaneous multislice imaging.

Purpose of the Study:

• To develop a novel reconstruction method for in vivo MRI acquisition that achieves higher acceleration than conventional techniques.
• To address the ill-conditioned reconstruction problem in accelerated diffusion MRI.

Main Methods:

• Constraining reconstruction in the spatial (k) domain by incorporating angular (q) domain information.
• Utilizing Gaussian processes to exploit signal smoothness in q-space.
• Employing a Bayesian framework with automatic estimation of smoothness hyper-parameters.

Main Results:

• Demonstrated in-plane undersampling exceeding parallel imaging limits.
• Achieved simultaneous multislice imaging combined with in-plane undersampling at a total acceleration factor of 12.
• Simulations and in vivo results showed superior performance compared to conventional parallel imaging methods.

Conclusions:

• The proposed method improves upon existing techniques for diffusion MRI acceleration.
• Significant improvements are observed for high simultaneous multislice acceleration combined with in-plane undersampling.
• The method achieves these improvements without compromising spatial or angular resolution and requires minimal modification to standard pulse sequences.