Datasets used to evaluate pyKNEEr.

I. Acquisition parameters: Parameters of the protocols used to acquire the images. Images of OAI1-DESS, OAI2-BL, and OAI2-FU were acquired with the same DESS protocol, consisting of 2 echos, although only their average was available (◇). Images of one subject of the dataset OAI1 had different slice spacing and thickness (⋆). Data queries to obtain acquisition parameters are in a Jupyter notebook (here). The original identification numbers (IDs) of the OAI images are in a Jupyter notebook used as a metafile (here). II. Ground truth segmentation: The datasets OAI1 and OAI2 have ground truth segmentations. They differ for computational method, segmented anatomy, and label type. III. Experimental results: Details of the steps in pyKNEEr for each dataset. Full circle (•) indicates processing of the dataset, while empty circle (∘) indicates processing of ground truth segmentations. The numbers in “Find reference” indicated the ID of the seed images used in the convergence study. Links are to the executed notebooks on GitHub.