Neural network restoration of gamma camera bremsstrahlung images

Abstract

Planar gamma camera Images of spherical phantoms filled with phosphorous-32 (32Ρ) are filtered with a new restoration technique which uses an order-statistic non-linear filter for noise suppression and a neural network for image restoration. The proposed method reduces Image noise effectively while preserving Image content and counts and restores the data consistently with source-to-detector distance or source size resulting in an effective attenuation coefficient for 32P. The new filter is compared to the Wiener filter for restoration in terms of error Images and attenuation measurements.