MRI (magnetic resonance imaging) is an important 
modern imaging modality due, in part, to its 
variety of contrast mechanisms and control 
parameters. However, this same wealth of control 
and contrast mechanisms poses a difficult problem 
for the design of MRI acquisition strategies. To 
date, most sequence design has been accomplished 
using experience and heuristic techniques. Despite 
the advances made in the field using these 
strategies, they are inherently limited by the 
skill of the designer. Many individuals who have 
a sequence-design need lack the necessary experience, 
and even experts may develop sub-optimal protocols. 
In addition, some areas of MRI, such as k-space 
trajectories, seem to offer advantages, but are 
still poorly understood even by the experts. In 
order to overcome these limitations it would be 
beneficial to use modern optimization algorithms 
in the design of acquisition techniques. To date, 
such optimal design has been quite limited in both 
objectives and parameters, and has led to techniques 
with limited benefit and applicability. This work 
overcomes these limitations by investigating a 
variety of algorithms and applications that should 
prove important in MRI. First, an optimally precise 
protocol for DCE-MRI (dynamic contrast enhanced MRI) 
was developed. Second, a software-based fluoroscopic 
gridding-reconstruction technique was developed. 
This makes a variety of pulse sequences, such as 
spiral imaging, available for iMRI (interventional 
magnetic resonance imaging) applications and is a 
necessary step for the remainder of this work. Third, 
time-optimal k-space trajectories were developed 
using the calculus of variations and multi-objective 
GAs (genetic algorithms) were used to develop optimal 
trajectories with respect to time, aliasing energy, 
flow-artifact energy, and off-resonance artifact energy. 
Fourth, these trajectories were tested for improvements 
in image quality using both objective image-quality 
measures for experimental phantom images and subjective 
image-quality measures for in-vivo images. Finally, 
the multi-objective algorithms were adapted to optimize 
images acquired using rectilinear sequences with 
respect to acquisition speed, resolution, and BW 
(bandwidth). The accomplishment of these goals resulted 
in improvements for several MRI acquisition techniques 
and also resulted in the development of techniques that 
should have wider and more general utility for obtaining 
similar improvements for other MRI techniques.