Location-known-exactly human-observer ROC studies of attenuation and other corrections for SPECT lung imaging
Lehovich, Andre ; Gifford, Howard C. ; King, Michael A
Citations
Student Authors
Faculty Advisor
Academic Program
UMass Chan Affiliations
Document Type
Publication Date
Subject Area
Embargo Expiration Date
Link to Full Text
Abstract
We use receiver operating characteristic (ROC) analysis of a location-known-exactly (LKE) lesion detection task to compare the image quality of SPECT reconstruction with and without various combinations of attenuation correction (AC), scatter correction (SC) and resolution compensation (RC). Hybrid images were generated from Tc-99m labelled NeoTect clinical backgrounds into which Monte Carlo simulated solitary pulmonary nodule (SPN) lung lesions were added, then reconstructed using several strategies. Results from a human-observer study show that attenuation correction degrades SPN detection, while resolution correction improves SPN detection, even when the lesion location is known. This agrees with the results of a previous localization-response operating characteristic (LROC) study using the same images, indicating that location uncertainty is not the sole source of the changes in detection accuracy.
Source
Lehovich A, Gifford HC, King MA. Location-known-exactly human-observer ROC studies of attenuation and other corrections for SPECT lung imaging. IEEE Nucl Sci Symp Conf Rec (1997). 2006;6:3229-3232. doi: 10.1109/NSSMIC.2006.353696. PMID: 19194523; PMCID: PMC2633939. Link to article on publisher's site