Pyramidal multiresolution image segmentation in SPECT

Abstract

A median filtered pyramidal multiresolution (MFPM) image segmentation method for detecting and delineating compact objects was applied to single-photon emission computed tomography (SPECT) images. These SPECT images were obtained by scanning spheres from 1 to 54 ml in size and from 50% to 100% contrast settings. The algorithm performed accurately for large sphere (20 and 54 ml) and high contrast (90% and 100%) cases. As the size and contrast decrease, the accuracy of the method also decreases. Comparison of the MFPM method with adaptive thresholding with edge-preserving smoothing (ATEPS) indicated superior performance of the MFPM method.<>