This algorithm fully automatically performs a volumetric segmentation of the spleen on contrast-enhanced thorax-abdomen CT scans. The underlying algorithm is a 3D U-Net model trained and validated on a large set of contrast-enhanced CT thorax abdomen scans from one academic center in the Netherlands. A large proportion of the training scans came from oncological follow-up and hence contained a wide variety of abnormalities and pathology. The algorithm and its validation is described in this publication in Radiology: Artificial Intelligence .
For questions about this algorithm, please contact Colin Jacobs.