Lesion spectra: radiation signatures and biological gateways

We describe an integrative approach to the modeling of biophysical radiation effects. The model takes aim at practical applications of the knowledge provided by molecular studies of radiation-matter interactions in DNA. The central proposition is the idea that the distribution of molecular lesions (i.e., a molecular lesion spectrum, MLS) generated in DNA by exposure to a particular radiation is a characteristic of that causal radiation (i.e., is a radiation signature, RS). We have found that adaptive neural networks (ANN's) provide an efficient way to validate that proposition and that ANN's are also likely to be invaluable in any attempt to correlate cancers with radiation types (i.e., with RS's), to use RS's for evaluating individual carcinogenic susceptibilities, and to develop a low-dose personalized monitoring capability. Although efforts to identify products of radiation that are specific to radiation type and to link those with biological responses are almost a century old, the RS concept has provided the first quantitative confirmation of such causal relations. That is, RS's and radiation markers have been identified for various types of radiation, electromagnetic (EM) and particulate, and these signatures and markers may constitute a new way for fast radiation exposure estimates, risk assessment, and 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 markers, almost the entire effort will relate to the modeling and interpretation of RS's using ANN processing.