Towards causal understanding of phenotypes and developmental anomalies

W. Tao, C. Keller, M. Hurtt, G. Srinivasa, R. Whitaker, M. Langham
University of Utah,
United States

Keywords: MicroCT, mice, phenotype, statistical shape analysis


Reproduction is fundamental to healthy populations of all species. Understanding phenotypic development both normal and abnormal is key to decreasing environmental and health risks of drugs and chemicals. High-resolution imaging and multimodal integrative computational analysis are promising tools in enhancing our understanding of genotype-phenotype relationships. We leveraged the synergistic potential of Lightweight Analysis of Mouse Anatomy (LAMA)1 for segmentation and ShapeWorks2 for shape representation to elucidate phenotypic distinctions between normal and mutant E18 mouse fetus microscopic, computed tomograms (uCT). This study provides a comprehensive proof of concept that could pave the way for novel insights into developmental biology and genetics. Materials and Methods Three meticulously curated datasets were used; Dataset #1 encompassed 49 IMPC-KOMP*3 wildtype mouse fetus scans and an additional cohort of 7 Foxj3_BCM_185 normals. Dataset #2 comprised 7 mutant Foxj2-/- specimens. These iodine-stained samples were uniformly down sampled to a resolution of 14 µm to facilitate analysis. Dataset #3, pilot scans done at our lab on 11 wild type E18 mice, provided a higher resolution of 12 µm and featured PTA staining, scanned without orientation protocol of IMPC-KOMP. These datasets underwent a rigorous preprocessing protocol, including reorientation and down sampling, to standardize input for subsequent computational analysis. Results LAMA's segmentation capabilities were robust for IMPC images, yet the tool encountered limitations when applied to our pilot scans, indicating a potential avenue for scan optimization. Using ShapeWorks, our pipeline identified statistically significant morphological differences (p