Dynamic Chemical Shift Imaging for Image-Guided Thermal Therapy

Magnetic resonance temperature imaging (MRTI) is recognized as a noninvasive means to provide temperature imaging for guidance in thermal therapies. The most common method of estimating temperature changes in the body using MR is by measuring the water proton resonant frequency (PRF) shift. Calculation of the complex phase difference (CPD) is the method of choice for measuring the PRF indirectly since it facilitates temperature mapping with high spatiotemporal resolution. Chemical shift imaging (CSI) techniques can provide the PRF directly with high sensitivity to temperature changes while minimizing artifacts commonly seen in CPD techniques. However, CSI techniques are currently limited by poor spatiotemporal resolution. This research intends to develop and validate a CSI-based MRTI technique with intentional spectral undersampling which allows relaxed parameters to improve spatiotemporal resolution. An algorithm based on autoregressive moving average (ARMA) modeling is developed and validated to help overcome limitations of Fourier-based analysis allowing highly accurate and precise PRF estimates. From the determined acquisition parameters and ARMA modeling, robust maps of temperature using the k-means algorithm are generated and validated in laser treatments in ex vivo tissue. The use of non-PRF based measurements provided by the technique is also investigated to aid in the validation of thermal damage predicted by an Arrhenius rate dose model.

Application of stellar photometry to the analysis of microarray images

Improvements in the analysis of microarray images are critical for accurately quantifying gene expression levels. The acquisition of accurate spot intensities directly influences the results and interpretation of statistical analyses. This dissertation discusses the implementation of a novel approach to the analysis of cDNA microarray images. We use a stellar photometric model, the Moffat function, to quantify microarray spots from nylon microarray images. The inherent flexibility of the Moffat shape model makes it ideal for quantifying microarray spots. We apply our novel approach to a Wilms' tumor microarray study and compare our results with a fixed-circle segmentation approach for spot quantification. Our results suggest that different spot feature extraction methods can have an impact on the ability of statistical methods to identify differentially expressed genes. We also used the Moffat function to simulate a series of microarray images under various experimental conditions. These simulations were used to validate the performance of various statistical methods for identifying differentially expressed genes. Our simulation results indicate that tests taking into account the dependency between mean spot intensity and variance estimation, such as the smoothened t-test, can better identify differentially expressed genes, especially when the number of replicates and mean fold change are low. The analysis of the simulations also showed that overall, a rank sum test (Mann-Whitney) performed well at identifying differentially expressed genes. Previous work has suggested the strengths of nonparametric approaches for identifying differentially expressed genes. We also show that multivariate approaches, such as hierarchical and k-means cluster analysis along with principal components analysis, are only effective at classifying samples when replicate numbers and mean fold change are high. Finally, we show how our stellar shape model approach can be extended to the analysis of 2D-gel images by adapting the Moffat function to take into account the elliptical nature of spots in such images. Our results indicate that stellar shape models offer a previously unexplored approach for the quantification of 2D-gel spots. ^

Extension of k-ratio methods to rank-based analyses

An extension of k-ratio multiple comparison methods to rank-based analyses is described. The new method is analogous to the Duncan-Godbold approximate k-ratio procedure for unequal sample sizes or correlated means. The close parallel of the new methods to the Duncan-Godbold approach is shown by demonstrating that they are based upon different parameterizations as starting points.^ A semi-parametric basis for the new methods is shown by starting from the Cox proportional hazards model, using Wald statistics. From there the log-rank and Gehan-Breslow-Wilcoxon methods may be seen as score statistic based methods.^ Simulations and analysis of a published data set are used to show the performance of the new methods. ^

Increased geranylgeranylated K-Ras contributes to antineoplastic effects of farnesyltransferase inhibitors.

The Ras family of small GTPases (N-, H-, and K-Ras) is a group of important signaling mediators. Ras is frequently activated in some cancers, while others maintain low level activity to achieve optimal cell growth. In cells with endogenously low levels of active Ras, increasing Ras signaling through the ERK and p38 MAPK pathways can cause growth arrest or cell death. Ras requires prenylation – the addition of a 15-carbon (farnesyl) or 20-carbon (geranylgeranyl) group – to keep the protein anchored into membranes for effective signaling. N- and K-Ras can be alternatively geranylgeranylated (GG’d) if farnesylation is inhibited but are preferentially farnesylated. Small molecule inhibitors of farnesyltransferase (FTIs) have been developed as a means to alter Ras signaling. Our initial studies with FTIs in malignant and non-malignant cells revealed FTI-induced cell cycle arrest, reduced proliferation, and increased Ras signaling. These findings led us to the hypothesis that FTI induced increased GG’d Ras. We further hypothesized that the specific effects of FTI on cell cycle and growth result from increased signal strength of GG’d Ras. Our results did show that increase in GG’d K-Ras in particular results in reduced cell viability and cell cycle arrest. Genetically engineered constructs capable of only one type of prenylation confirmed that GG’d K-Ras recapitulated the effect of FTI in 293T cells. In tumor cell lines ERK and p38 MAPK pathways were both strongly activated in response to FTI, indicating the increased activity of GG’d K-Ras results in antiproliferative signals specifically through these pathways. These results collectively indicate FTI increases active GG’d K-Ras which activates ERK and p38 MAPKs to reduced cell viability and induce cell cycle arrest in malignant cells. This is the first report that identifies increased activity of GG’d K-Ras contributes to antineoplastic effects from FTI by increasing the activity of downstream MAPKs. Our observations suggest increased GG’d K-Ras activity, rather than inhibition of farnesylated Ras, is a major source of the cytostatic and cytotoxic effects of FTI. Our data may allow for determination of which patients would benefit from FTI by excluding tumors or diseases which have strong K-Ras signaling.