Magnetic Resonance Spin TomogrAphy in Time-domain (MR-STAT) is a fast, multi-parametric technique which reconstruct all parameters of interest from a single short acquisition. In MR-STAT, the patient’s body is considered a large dynamical system which is probed in the time-domain by radiofrequency coils. Since the dynamic equations are well known (Faraday’s law and Bloch equation), reconstruction of the system parameters is carried out through the solution of a large-scale non-linear inversion problem directly from the time-domain data (Figure 1).

This approach allows for the spatial localization and the parameter quantification steps to be solved simultaneously. In standard qMRI, these steps are carried out subsequently, thus the spatial encoding scheme needs to fulfill either to Nyquist criteria (standard qMRI) or to certain types of incoherence (e.g. MR Fingerprinting) for a reliable reconstruction. The direct implication is that, in MR-STAT, acquisition is free from design constraints thus it can be carried out also with standard, reliable Cartesian encoding schemes. Another practical advantage of the MR-STAT approach is that all signal is fitted by a deterministic model (only the thermal noise is stochastic), thus large under-sampling artifacts are not present in MR-STAT (Figure 2). In addition, data reconstruction is performed iteratively, without the need of the pre-computed dictionary that hinders user adaptations in image protocols.

For more info check:



Luuk Jacobs
Master student, 2022
Jordi Kleinloog
Postdoc, 2022
Stefano Mandija
Assistant Professor
Alessandro Sbrizzi
Associate Professor