skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Comparison of Channelized Hotelling and Human Observers inDetermining Optimum OS-EM Reconstruction Parameters for MyocardialSPECT

Journal Article · · IEEE Transactions on Nuclear Science
OSTI ID:918548

The performance of the Channelized Hotelling Observer (CHO)was compared to that of human observers for determining optimumparameters for the iterative OS-EM image reconstruction method for thetask of defect detection in myocardial SPECT images. The optimumparameters were those that maximized defect detectability in the SPECTimages. Low noise, parallel SPECT projection data, with and without ananterior, inferior or lateral LV wall defect, were simulated using theMonte Carlo method. Poisson noise was added to generate noisyrealizations. Data were reconstructed using OS-EM at 1&4subsets/iteration and at 1, 3, 5, 7&9 iterations. Images wereconverted to 2D short-axis slices with integer pixel values. The CHO used3 radially-symmetric, 2D channels, with varying levels of internalobserver noise. For each parameter setting, 600 defect-present and 600defect-absent image vectors were used to calculate the detectabilityindex (dA). The human observers rated the likelihood that a defect waspresent in a specified location. For each parameter setting, the AUC wasestimated from 48 defect-present and 48 defect-absent images. Thecombined human observer results showed the optimum parameter settingcould be in the range 5-36 updates ([number of subsets]/iteration enumber of iterations). The CHO results showed the optimum parametersetting to be 4-5 updates. The performance of the CHO was much moresensitive to the reconstruction parameter setting than was that of thehuman observers. The rankings of the CHO detectability values did notchange with varying levels of internal noise.

Research Organization:
Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States)
Sponsoring Organization:
USDOE Director, Office of Science; National Institutes ofHealth
DOE Contract Number:
DE-AC02-05CH11231; NIHEB00121
OSTI ID:
918548
Report Number(s):
LBNL-54935; IETNAE; R&D Project: 864K2B; BnR: 400412000; TRN: US0805404
Journal Information:
IEEE Transactions on Nuclear Science, Vol. 53, Issue 3pt2; Related Information: Journal Publication Date: 06/2006; ISSN 0018-9499
Country of Publication:
United States
Language:
English