1. Neural Net Approaches for Event Location in the Detector Modules
- Author
-
John R. Macri, Roland Diehl, Helmut Steinle, J. W. den Herder, B. Swanenberg, K. Bennett, J. A. Lockwood, V. Schönfelder, V. Zeitlmeyer, H. Aarts, G. G. Lichti, James M. Ryan, C. Winkler, Alanna Connors, Mark L. McConnell, G. Simpson, W. E. Webber, A. Deerenberg, K. Reinhard, M. Varendorff, W. Hermsen, Andrew W. Strong, D. Morris, M. Loomis, and C. de Vries
- Subjects
Discrete mathematics ,Physics ,Photomultiplier ,COSMIC cancer database ,Position (vector) ,Astrophysics::High Energy Astrophysical Phenomena ,Compton telescope ,Detector ,Neutron ,Event (particle physics) ,Energy (signal processing) ,Computational physics - Abstract
From the description of the Compton telescope given previously (Schonfelder et al., this volume), one can see that the accuracy with which one determines the position of a cosmic gamma-ray source depends not only on the measurements of the energy deposited in the upper (D1) and lower (D2) detectors, but also on how accurately one estimates the (X, Y, Z) positions of each gamma-ray or neutron interaction (an event). If nothing were known about the position of each event except in which module it occured, it would increase the uncertainty in the position of a source by on the order of 10°. Within each COMPTEL module, one extracts position information from comparisons of relative intensities of signals in the photomultipier tubes. This technique was introduced in the 1950’s for medical imaging by Anger (1958), and later was adapted to astrophysical applications (Zych et al. 1983; Schonfelder et al. 1984; Stacy 1985).
- Published
- 1992
- Full Text
- View/download PDF