Quantification of T1, T2 and Pseudo Spin Density using bSSFP with Correction for Non-Ideal Flip Angle Profiles

The quantification of T1 (longitudinal relaxation time) and T2 (transverse relaxation time) has become an important issue in the field of MRI during the last decades in many clinical and scientific applications. Conventional methods for relaxometry suffer from very long acquisition times in the order of several hours. A very promising approach to quantify both relaxation parameters and the pseudo spin density rho0 by sampling the transient of an IR-bSSFP sequence was shown in previous works. The RF-pulses used in slice selective bSSFP sequences are usually very short, which results in rather non-ideal flip angle profiles far away from the desired rectangular shape. Hence, the quantification accuracy, especially of T2, is altered significantly by these non-ideal profiles. The aim of this work was to investigate the influence of non-ideal flip angle profiles on the quantification accuracy and to implement an algorithm which is able to correct the flip angle profile effects. Further, this algorithm should be included in an existing framework to perform the quantification of under-sampled data.



For this purpose the real slice profile of the exciting RF-pulse had to be determined. This was done by simulations according to the Bloch-equations and by measurements inside a phantom. A forward model was established, based on rotation and relaxation matrices as a result of the Bloch-equations, which models the magnetization vector recursively over one period of TR. Two correction algorithms were implemented, one performs pixel wise quantification based on already reconstructed images, and the second algorithm performs the quantification of k-space data, which is also known as the model-based nonlinear inverse reconstruction. This approach is able to reconstruct under-sampled data.



The algorithms were tested on simulated data, on measured data inside a phantom and on in vivo measurements inside the human brain of healthy volunteers. The results achieved on simulated data and on phantom measurements are very promising. The quantification error, which is most dominant in T2, could be significantly reduced. However these promising results could not be reproduced for the in vivo measurements. It turned out that the reasons for that are on-resonant magnetization transfer effects. Further investigations are necessary to prove if it is possible to either correct these effects or to gain more information about the magnetization transfer out of the transient time decay.