Modeling Annotation Uncertainty with Gaussian Heatmaps in Landmark Localization
This work addresses annotation uncertainty in medical landmark localization, which is crucial for improving diagnostic reliability in clinical settings, though it is incremental as it builds on existing heatmap-based methods.
The paper tackles the problem of uncertain landmark annotations in medical imaging by modeling annotation ambiguities and prediction uncertainty using anisotropic Gaussian heatmaps, achieving state-of-the-art results on hand radiograph and lateral cephalogram datasets with correlations to localization accuracy and observer variability.
In landmark localization, due to ambiguities in defining their exact position, landmark annotations may suffer from large observer variabilities, which result in uncertain annotations. To model the annotation ambiguities of the training dataset, we propose to learn anisotropic Gaussian parameters modeling the shape of the target heatmap during optimization. Furthermore, our method models the prediction uncertainty of individual samples by fitting anisotropic Gaussian functions to the predicted heatmaps during inference. Besides state-of-the-art results, our experiments on datasets of hand radiographs and lateral cephalograms also show that Gaussian functions are correlated with both localization accuracy and observer variability. As a final experiment, we show the importance of integrating the uncertainty into decision making by measuring the influence of the predicted location uncertainty on the classification of anatomical abnormalities in lateral cephalograms.