LESION SPECTRA - RADIATION SIGNATURES AND BIOLOGICAL GATEWAYS

We describe an integrative approach to the modeling of biophysical rad iation effects. The model takes aim at practical applications of the k nowledge provided by molecular studies of radiation-matter interaction s in DNA. The central proposition is the ideat that the distribution o f molecular lesions (i.e., a molecular lesion spectrum, MLS) generated in DNA by exposure to a particular radiation is a characteristic of t hat causal radiation (i.e., is a radiation signature, RS). We have fou nd that adaptive neural networks (ANN's) provide an efficient way to v alidate that proposition and that ANN's are also likely to be invaluab le in any attempt to correlate cancers with radiation types (i.e., wit h RS's), to use RS's for evaluating individual carcinogenic susceptibi lities, and to develop a low-dose personalized monitoring capability. Although efforts to identify products of radiation that are specific t o radiation type and to link those with biological responses are almos t a century old, the RS concept has provided the first quantitative co nfirmation of such causal relations. That is, RS's and radiation marke rs have been identified for various types of radiation, electromagneti c (EM) and particulate, and these signatures and markers may constitut e a new way for fast radiation exposure estimates, risk assessment, an d cumulative low-dose evaluation. In this work, while we will present a short review of the concepts and methods related to both RS's and ma rkers, almost the entire effort will relate to the modeling and interp retation of RS's using ANN processing.