Practical imaging and analyses for qMRI

Abstract

Magnetic resonance imaging (MRI) is a powerful tool to gain insight into physiological structures and processes, molecular behavior, and psychological conditions. It has a broad range of applications, including, but not limited to, medical imaging, such as the insight it can provide towards medicinal chemistry regarding the appropriate targeting and efficiency of new drug candidates. While the fundamentals of MRI are thoroughly understood at this point of its development, the ability to extrapolate results beyond the qualitative data often seen with anatomical imaging to reach new levels of insight regarding the aforementioned uses has gained burgeoning attention from physicists, chemists, and mathematicians. Through exploitation of the meta-data and manipulation of multi-image acquisition, it has been demonstrated that ultra-high-resolution imaging, quantitative imaging, and real-time data analyses are possible. This provides critical information for clinicians and fellow researchers towards understanding disease progression, treatment efficacy, and treatment behavior in infinitesimal steps, providing new methods to investigate mode of action in combination with other monitoring methodology. Quantitative MRI (qMRI) requires the application of rigorous acquisition methods, sophisticated post-processing techniques, and fastidious analysis to be relevant and replicable. However, should all these qualifications be executed appropriately, it can provide verifiable, repeatable, and pivotal information regarding enigmatic maladies and treatment pathways. Investigation into the quantitative mapping of longitudinal relaxation times to investigate tumor structures and attempted treatment methods through T₁-mapping will provide insight into the specificity and sensitivity of qMRI. The use of T₁-mapping has been shown to be a non-invasive methodology to gain deeper insight and understanding into cellular level changes as the result of inflammation and stromal barrier formation. It is sensitive enough to distinguish between ascites and tumor formations, can provide information regarding tumor heterogeneity, and is a quick non-invasive method in which to do so. Sensitivity is demonstrated by the differences of T₁-mapping behavior based upon different cell lines compared to normal tissue, and its quick acquisition time makes it a promising method for clinical applications. Diffusion tensor imaging, one of the leading modalities under study with qMRI, will be assessed for faster image acquisition. Through prioritizing different parameters in order to monitor the impact alterations have on image quality and information available, a template of key characteristics that can be monitored in varying degrees of depth will be investigated. This will provide insight into how alteration of parameters affect final data acquisition and potential remedies to condense imaging time that can be altered based upon study design. Furthermore, the inverse of this investigation will determine which key parameters will yield the highest quality data possible and examine the impacts of obtaining ultra-high-direction diffusion tensor images and optimal resolutions, as well as the consequences to signal when resolutions are too high. Additionally, a new method of analysis through an integrated pipeline that will accommodate artifact correction, tensor calculation, and yield multiple diffusion metrics will be presented that is compatible with various forms of instrumentation in order to correct for limited compatibility, un-intuitive application, and difficulty of use seen in many commercially available platforms available today. Finally, investigation into next generation steps for qMRI which look to further increase the depths of information gained through imaging will be examined through iterative back projection to calculate ultra-high-resolution images at a fraction of the time cost seen with traditional imaging. This provides insight into the application of super resolution techniques to gain higher level information regarding diffusion tensors and anatomical structures without suffering from signal loss due to nominal voxel sizes. The ultimate goal is to generate data sets which no longer hinge upon subjective interpretation and progress the field towards machine learning techniques which can be used to identify anomalies present post-imaging.