Title :
A count dependent reconstruction filter optimized to match a lesion size and an object contrast in SPECT
Author :
Kim, Jung Hee ; Kim, Kwang Ick
Author_Institution :
Dept. of Math., Pohang Inst. of Sci. & Technol., South Korea
fDate :
31 Oct-6 Nov 1993
Abstract :
A new formulation method for a reconstruction filter optimized to improve lesion detectability is presented, based on a physical limit of SPECT lesion detection capability. To determine the optimal characteristics of reconstruction filter, the authors focused on improving the local detectability of the threshold lesion corresponding to a physical limit. The selection of SPECT spatial resolution has been set at which the detectable limit size is matched to twice the resolution (SPECT system FWHM=limitsize/2). Assuming a gaussian SPECT system PSF, the reconstruction filter has been derived as GAUSS filter with the selected resolution, upto whose rolloff point the inverse filter for detector system resolution recovery is combined in the frequency domain. The truncation of the filter at the rolloff point has been designed to increase the local S/N ratio of the threshold lesion in the reconstructed SPECT image. The reconstruction filter has shown a flexible preprocessing characteristics of resolution recovery and smoothing, depending on influence of main limitation effects on detectability of the threshold detectable lesion. In a phantom-simulation study with change in major physical factors of photon counts collected, object contrast and detector system resolution, the proposed filter produced a significantly improved images in sense of lesion detectability over the conventional reconstruction filters
Keywords :
image reconstruction; medical image processing; single photon emission computed tomography; GAUSS filter; SPECT lesion detection capability; SPECT spatial resolution; count dependent reconstruction filter; filter truncation; frequency domain; inverse filter; lesion detectability; lesion size; local S/N ratio; object contrast; phantom-simulation study; physical limit; resolution recovery; smoothing; Detectors; Frequency domain analysis; Gaussian processes; Image reconstruction; Lesions; Matched filters; Object detection; Optimization methods; Smoothing methods; Spatial resolution;
Conference_Titel :
Nuclear Science Symposium and Medical Imaging Conference, 1993., 1993 IEEE Conference Record.
Conference_Location :
San Francisco, CA
Print_ISBN :
0-7803-1487-5
DOI :
10.1109/NSSMIC.1993.373553