Recurrent non-random genomic alterations are the hallmarks of cancer and the characterization of these imbalances is critical to our understanding of tumorigenesis and cancer progression. We are performing Array-comparative genomic hybridization (A-CGH) on cDNA microarrays containing 42,000 elements in neuroblastoma. We are developing a novel probabilistic algorithm, called topological statistics, to increase the sensitivity of cDNA A-CGH for detecting single-copy alterations. Our method not only overcomes the shortcoming of relative low sensitivity of cDNA A-CGH but also enables a direct visualization of the statistical confidence for the observed alterations. Furthermore using probabilistic approaches and machine learning algorithms (ANNs and Support Vector Machines) to estimate the frequency of genomic imbalances in tumors of different stage and MYCN amplification status, we are attempting to identify unique patterns of genomic imbalance in all three categories. Using these approaches we identifying models of the evolution and progression of neuroblastoma.