Ideal observer approximation using Bayesian classification neural networks | IEEE Journals & Magazine | IEEE Xplore