UNLABELLED: Mathematic "model" observers that predict human performance are of interest in medical imaging as substitutes in psychophysical studies. We have examined the correlations between human observers and several forms of the channelized Hotelling observer (CHO) for a tumor detection task with simulated SPECT liver images that were used to study the effects of scatter and scatter correction on detection. METHODS: A receiver operating characteristic (ROC) study was devised to investigate the relative value of a scatter-subtraction strategy in SPECT imaging. The study used simulated images of the biodistribution of 99mTc-labeled FO23C5 anticarcinoembryonic antigen antibodies within the liver. Projection data for 3 separate tumor locations and 5 strategies for handling scatter were obtained using Monte Carlo software applied to an anthropomorphic phantom. The strategies were (a) perfect scatter rejection, (b) no scatter correction, (c) no scatter correction under an assumption of an elevated amount of scatter, (d) an energy-spectrum-based scatter compensation of the normal-scatter case (b), and (e) similar scatter compensation for the elevated-scatter case (c). Image reconstruction approximated current clinical procedures at the University of Massachusetts Medical School. Human performance for each combination of location and strategy was based on averaging the areas under the ROC curve for 7 individuals. A set of 15 signal-to-noise ratios (SNRs) was derived from these averages for comparison with SNRs for CHO models featuring constant-Q and difference-of-gaussian (DOG) filters. RESULTS: The Spearman rank correlation coefficient was 0.92 (P = 0.000001) when comparing task performances for the average human and a constant-Q CHO using 4 square-profile channels. For the DOG version of the CHO, comparison with the average human found a coefficient of 0.84 (P = 0.00005). CONCLUSION: The significant positive correlations found between the rankings of the average human observer and the CHOs for our detection task indicate that a channelized model observer could eventually serve as a replacement for human observers. The specific CHO models we have used are best suited to screen for significant differences between strategies before a human psychophysical study.
UNLABELLED: Mathematic "model" observers that predict human performance are of interest in medical imaging as substitutes in psychophysical studies. We have examined the correlations between human observers and several forms of the channelized Hotelling observer (CHO) for a tumor detection task with simulated SPECT liver images that were used to study the effects of scatter and scatter correction on detection. METHODS: A receiver operating characteristic (ROC) study was devised to investigate the relative value of a scatter-subtraction strategy in SPECT imaging. The study used simulated images of the biodistribution of 99mTc-labeled FO23C5 anticarcinoembryonic antigen antibodies within the liver. Projection data for 3 separate tumor locations and 5 strategies for handling scatter were obtained using Monte Carlo software applied to an anthropomorphic phantom. The strategies were (a) perfect scatter rejection, (b) no scatter correction, (c) no scatter correction under an assumption of an elevated amount of scatter, (d) an energy-spectrum-based scatter compensation of the normal-scatter case (b), and (e) similar scatter compensation for the elevated-scatter case (c). Image reconstruction approximated current clinical procedures at the University of Massachusetts Medical School. Human performance for each combination of location and strategy was based on averaging the areas under the ROC curve for 7 individuals. A set of 15 signal-to-noise ratios (SNRs) was derived from these averages for comparison with SNRs for CHO models featuring constant-Q and difference-of-gaussian (DOG) filters. RESULTS: The Spearman rank correlation coefficient was 0.92 (P = 0.000001) when comparing task performances for the average human and a constant-Q CHO using 4 square-profile channels. For the DOG version of the CHO, comparison with the average human found a coefficient of 0.84 (P = 0.00005). CONCLUSION: The significant positive correlations found between the rankings of the average human observer and the CHOs for our detection task indicate that a channelized model observer could eventually serve as a replacement for human observers. The specific CHO models we have used are best suited to screen for significant differences between strategies before a human psychophysical study.
Authors: Se Young Chun; Timothy G Reese; Jinsong Ouyang; Bastien Guerin; Ciprian Catana; Xuping Zhu; Nathaniel M Alpert; Georges El Fakhri Journal: J Nucl Med Date: 2012-06-28 Impact factor: 10.057