Performance of the relative difference prior for hot lesion detection in whole-body PET/CT: an evaluation with numerical and real observers

For PET (positron emission tomography) imaging, different reconstruction methods can be applied, including ML (maximum likelihood) and MAP (maximum a-posteriori) reconstruction. Post-smoothed ML images have approximately position and object independent spatial resolution, which is advantageous for (semi) quantitative analysis. However, the complex object dependent smoothing obtained with MAP might yield improved noise characteristics, beneficial for lesion detection. In this contribution, MAP and post-smoothed ML are compared for hot spot detection by the channelized Hotelling observer (CHO) and by human observers (using a "multiple alternative forced choice" method). For MAP, a prior penalizing relative rather than absolute differences was applied. MAP resulted in markedly better performance for the CHO. However, no significant performance difference was detected with the human observer study. Addition of internal noise improved the agreement between the CHO and the human observers. The prior can be adjusted to obtain a smaller penalty for large differences, resulting in improved tolerance for strong edges. An additional CHO study indicated that this edge tolerance decreases observer performance. Similarly, performance was better for the quadratic prior than for the Huber prior.