Fig. 1: Micro-computed tomography (µCT) images of mouse fetuses and phenotyping of congenital heart defects.
From: Deep learning-based detection of murine congenital heart defects from µCT scans
a μCT scans of the thorax and abdomen of mouse fetuses at 18.5 days of development (E18.5, left) and at birth (P0, right) displayed as 3D renderings (top) or stacks of 2D slices (bottom). Blue bounding boxes show the set of 2D thoracic slices containing the heart. Red bounding boxes are framed around the heart. The corresponding 2D image stacks are shown with matching color frames below. Yellow contours on each slice outline the automatically segmented heart. b UpSet plot53 showing the distribution of five types of heart malformations for n = 37 mice with heterotaxy. SD septal defects, MGA malposition of the great arteries, ASD atrial situs defects, AM apex malposition, VM ventricle malposition. Horizontal bars show the total number of cases for each type of malformation. Because many mice have multiple malformations, the total (n = 129) exceeds the number of samples (n = 37). Vertical bars show the number of cases that exhibit the combination of malformations shown in the corresponding matrix column below. For example, 5 cases have both SD and MGA but none of the other malformations. Out of the 37 cases, 36 cases have at least two anomalies, and 8 cases exhibit all 5 types of anomalies simultaneously. c Contingency table for diagnostic label (CHD or normal) and developmental stage (E18.5 or P0) for n = 139 fetuses. Scale bars in (a): 0.5 mm for 3D renderings (top), 4 mm (left) or 1.5 mm (right) for 2D slices (bottom). Source data is available in Supplementary Data 1.
