ON THE DENOISING OF NUCLEAR MEDICINE CHEST REGION IMAGES
Mitko Kostov, Cvetko D. Mitrovski · 2004
In the paper we present our approach on pre-processing of chest-region NM images. The proposed method combines Discrete Wavelet Transform realized via QMF filters with a specific strategy for selecting an appropriate threshold. Due to the signal-dependence of the Poisson noise, the Anscombe variance-stabilizing transformation is applied. As a result, the transformed coefficients are characterized with a white Gaussian noise model for which appropriate threshold selecting is easier. The performance of the proposed method is demonstrated on real NM images.