Fast Estimation of the Bound Pool Fraction in the Brain

The bound pool fraction (BPF) is a quantitative magnetization transfer (MT) parameter that has drawn interest in clinical research as a biomarker for the myelin content. Hence,it allows to quantify changes within the nervous system of the brain for example in a neurodegnerative disease like Multiple Sclerosis (MS). The estimation of high resolution BPF images in clinical acceptable scan times is still a challenging task. One approach that seems to overcome this issue is the fast BPF estimation from a single off-resonance measurement, which is based on constraints and assumptions of the pulsed MT-model.
In the present Master’s thesis this method was implemented and validated in three stages with the pulsed MT-model and numerical simulations.
In the first stage of the thesis, simulations with the pulsed MT-model and the coupled Bloch equations were carried out to validate the approximations of the pulsed MT-model.Furthermore a new numerical error model was developed to estimate the bias and error
due noise of the fast BPF estimation for a specific sets of constraints.
In the second stage the constraints were estimated by fitting the parameters of the pulsed MT-model from bovine serum albumin (BSA) phantom measurements and post mortem measurements. With the derived constraints the fast BPF estimation was carried out and the estimated error was compared with the predicted error from the numerical simulations.
Finally, in the last stage, the model parameters of the pulsed-MT have been estimated in vivo. With the estimated parameters numerical error simulations were performed to find the best off-resonance saturation frequency and amplitude for the acquisition of highresolution fast BPF maps.
The results of the BSA-phantom and the post mortem measurements point out that the fast BPF is possible within a certain error-tolerance. For predicting the error of the fast BPF estimation it has been shown that the numerical error simulations are better suited than the proposed error model. However, the results of the in vivo measurements indicate an overestimation of the parameters compared to the reported values in the literature.A likely explanation for the difference is that in our measurements the Magnetization Prepared 2 Rapid Acquisition Gradient Echoes (MP2RAGE) sequence was used instead of the variable flip angle (VFA) method for the T1-mapping. The VFA could bias the solution of the BPF estimation due to the short repetition rate of the sequence, which may result in a bias of the relaxation time T1,obs. Therefore further research is necessary to study the influence of different T1-mapping techniques on the fast BPF estimation.