Combining Molecular and Imaging Biomarkers to Enhance MALDI Biomarker Analysis

This thesis presents a three-step method to predict prostate cancer (PCa) regions on biopsy tissue samples based on high confident, low resolution PCa regions marked by a pathologist. First, a prediction model is designed to predict PCa regions using matrix-assisted laser desorption mass spectrometry (MALDI-MS) tissue imaging data from one prostate tissue slice. Second, a texture analysis technique is applied to a high magnification optical image for the same purpose from an adjacent tissue slice. Finally, those two results are fused to obtain the PCa regions that will assist MALDI imaging biomarker analysis. Experiments show that the texture analysis based prediction is sensitive but not specific, and the prediction based on the MALDI-MS data is specific while less sensitive. By combing those two results, a much better prediction for PCa regions on the adjacent slice can be achieved. This thesis focuses on the MALDI-MS based PCa region prediction and fusion of prediction from texture analysis and that from MALDI-MS data process mg.