Fig. 3 | Scientific Data

Fig. 3

From: Exploring de-anonymization risks in PET imaging: Insights from a comprehensive analysis of 853 patient scans

Fig. 3

Experiment results. (a) Training/validation loss across all folds. (b) Percentage of landmarks placed across all patients for each fold on CT and PET modalities as well as the percentage of well-placed PET landmarks that are congruent to CT placement. Error bars indicate the variability among folds. (c) Mean absolute distance normalized by interocular distance distributions across all folds for denoised and nondenoised PET images. We note that the median is systematically below for the denoised PET images.

Back to article page