Ultrafast compartmentalized relaxation time mapping with linear algebraic modeling

Purpose: To dramatically accelerate compartmental-average longitudinal (T₁) and transverse (T₂) relaxation measurements using the minimal-acquisition linear algebraic modeling (SLAM) method, and to validate it in phantoms and humans. Methods: Relaxation times were imaged at 3 Tesla in phantoms, in the abdomens of six volunteers, and in six brain tumor patients using standard inversion recovery and multi–spin-echo sequences. k-space was fully sampled to provide reference T₁ and T₂ measurements, and SLAM was performed using a limited set of phase encodes from central k-space. Anatomical compartments were segmented on scout images post-acquisition, and SLAM reconstruction was implemented using two algorithms. Compartment-average T₁ and T₂ measurements were determined retroactively from fully sampled data sets, and proactively from SLAM data sets at acceleration factors of up to 16. Values were compared with reference measurements. The compartment's localization properties were analyzed using the discrete spatial response function. Results: At 16-fold acceleration, compartment-average SLAM T₁ measurements agreed with the full k-space compartment-average results to within 0.0% ± 0.7%, 1.4% ± 3.4%, and 0.5% ± 2.9% for phantom, abdominal, and brain T₁ measurements, respectively. The corresponding T₂ measurements agreed within 0.2% ± 1.9%, 0.9% ± 7.9%, and 0.4% ± 5.8%, respectively. Conclusion: SLAM can dramatically accelerate relaxation time measurements when compartmental or lesion-average values can suffice, or when standard relaxometry is precluded by scan-time limitations. Magn Reson Med 79:286–297, 2018.