Magnetic resonance velocimetry, as a non-invasive contactless method, is the central core of this thesis. In addition to mean flow measurement, this method can obtain a three-dimensional three-component (3D-3C) velocity map to visualize flow behavior in clinical and industrial applications. In this method, the measurement signal comes directly from the sample's nuclei, making it a very precise and reliable technique. However, some limitations and instances of inaccurate results are reported. Based on the available literature, three main issues are reported for NMR/MRI flow measurement. ... mehrThe first is the obstacles to using the NMR technique in commercial flowmeters. The second topic is the length of echo time, the main bottleneck for fast flow measurement. The last one is the velocity encoding parameter ($V_{\text{enc}}$), whose small values result in aliasing, and big values reduce phase-SNR for slow-moving voxels.
Despite extremely high precision in measuring flow rate by the available NMR spectrometers, this technique is not broadly used in the flowmeter industry. The main barriers are the high cost and large volume. The main limitation of a compact NMR magnet is the instability, more precisely, the \textit{drift} in working frequency. In the current study, the pulse sequence and measurement parameters of an affordable and portable NMR-based flowmeter are optimized to counteract the drift issue. A pulse sequence is designed based on the flow encoding bipolar gradient and the \textit{Job Acquisition} technique. The former encodes the flow in the MR phase, and the latter enables us to have two acquisitions after each excitation. The first acquisition is before, and the second one is after the bipolar gradient. The first is used as the reference, and the second contains the injected phase due to flow. Having both acquisitions in a single measurement reduces the consequences of the drift. In addition, all other parameters, including the bandwidth of the excitation pulse, bipolar gradient, and acquisition times, are optimized, and the adjustment is limited to the \textit{Drift Adjustment}. The designed pulse sequence is tried on a compact and affordable NMR-based flow meter, and the results show robust measurements.
Available studies show that the maximum measurable velocity is directly related to the length of the echo time. Conventionally, flow imaging is based on a slice selective excitation pulse and then applying a flow encoding gradient (preceding spatial encoding gradients). In this study, these two steps are merged. Instead, a slice selective flow encoding excitation pulse is designed using optimal control theory. A constraint is added to make the pulse slice selective. Due to being over-constrained, the optimal control problem is divided into two steps. In the first one, a mathematical model is proposed to encode the velocity of the central plane of the excited slice into the MR phase. The Pontryagin Maximum Principle is used to solve the optimal control problem, and the GRadient Ascent Pulse Engineering (GRAPE) algorithm is employed to maximize the Hamiltonian. Despite being nonlinear, a bijective relationship between phase and velocity is established. However, the evaluations show that the phase over the thickness of the excited slice is nonuniform. Therefore, the second step of the optimal control problem is initiated, whose target is to minimize the nonuniformity. Even though some nonuniformity is still left, the assessments show significant improvement in the uniformity. The results show that the increase in length of echo time caused by flow encoding is 1.58 times shorter in the optimal control method compared to the bipolar gradient technique. Furthermore, the slice refocusing gradient is not needed, which makes this method much more time-efficient.
The last topic is about the velocity encoding parameter $V_{\text{enc}}$, which plays a central role in phase contrast flow measurement. The value of this parameter should be slightly bigger than the maximum velocity expected in the flow system. The problem is that the phase-velocity resolution for the voxels moving significantly slower than $V_{\text{enc}}$ is too low. Also, if the $V_{\text{enc}}$ is smaller than needed, aliasing happens, and the imaging goes through wrapping, which results in considerable errors. In the current thesis, selective excitation is employed to tackle this issue. In the presence of several chemical groups, selective excitation enables us to define a different value of $V_{\text{enc}}$ for each group. For the present research, this technique is applied to flow NMR, and the number of groups is limited to two (water and sodium acetate). The bigger value of $V_{\text{enc}}$, representing the maximum velocity in the system, is assigned to sodium acetate, and the smaller one is defined for water. The injected phase in the signal coming from water goes through the wrapping. However, that of sodium acetate can be used to unwrap it reliably. The ratio between values of $V_{\text{enc}}$ is 7, which means that the phase-SNR is improved by the same factor. In flow NMR, the whole volume is treated as one voxel, which results in phase dispersion and, hence, loss of signal. It confines the maximum achievable ratio between the values of $V_{\text{enc}}$. In flow imaging, much higher ratios are expected.
