Fig. 5
From: Dimension reduction and outlier detection of 3-D shapes derived from multi-organ CT images
Bagplots applied to the separate 2-dimensional projection (left: MCIA, right: PCA) of feature matrices describing the geometry of multiple organs from the same individuals. Here, a section is shown for the liver, pancreas and spleen from the AbdomenCT-1k dataset. A data point is either located inside the ‘bag’ region (dark grey), inside the ‘fence’ region (light grey) or an outlier (area that is not enclosed). Each individual is labeled by a number. The full images for both datasets, displaying the projection of all organs and comparing all three dimension reduction methods, MCIA, PCA and t-SNE, are provided in Supplementary Figs. 1 and 2