As MRI systems have increased in static magnetic field strength, the radiofrequency
(RF) fields that are used for magnetisation excitation and signal reception have become
significantly less uniform. This can lead to image artifacts and errors when performing
quantitative MRI. A further complication arises if the RF fields vary substantially in time.
In the first part of this investigation temporal variations caused by respiration were
explored on a 3T scanner. It was found that fractional changes in transmit field
amplitude between inhalation and expiration ranged from 1% to 14% in the region of
the liver in a small group of normal subjects. This observation motivated the
development of a pulse sequence and reconstruction method to allow dynamic
observation of the transmit field throughout the respiratory cycle. However, the
proposed method was unsuccessful due to the inherently time-consuming nature of
transmit field mapping sequences.
This prompted the development of a novel data reconstruction method to allow the
acceleration of transmit field mapping sequences. The proposed technique posed the RF
field reconstruction as a nonlinear least-squares optimisation problem, exploiting the
fact that the fields vary smoothly. It was shown that this approach was superior to
standard reconstruction approaches.
The final component of this thesis presents a unified approach to RF field calibration.
The proposed method uses all measured data to estimate both transmit and receive
sensitivities, whilst simultaneously insisting that they are smooth functions of space.
The resulting maps are robust to both noise and imperfections in regions of low signal.