Fig. 3From: High precision localization of pulmonary nodules on chest CT utilizing axial slice number labelsUnsupervised clustering of Retinanet raw inferences. Visualization of unsupervised clustering of raw Retinanet inferences using DBSCAN. Diameter of each production corresponds to the Retinanet inference confidence score. Higher density clusters containing both axial and coronal predictions with high confidence scores are more likely to be real nodules. The 3 large clusters in (b)—green, purple, and dark teal—were true nodulesBack to article page