LAMA: automated image analysis for the developmental phenotyping of mouse embryos

Area of Science:

• Developmental biology
• Computational biology
• Medical imaging

Background:

• The International Mouse Phenotyping Consortium (IMPC) generates extensive 3D embryo imaging data using micro-computed tomography (micro-CT).
• Processing this large data volume requires significant resources and expertise, hindering immediate data exploitation.
• Automated annotation is essential for integrating 3D imaging data with other phenotyping datasets.

Purpose of the Study:

• To present an automated computational pipeline for mouse embryo phenotyping.
• To leverage wild-type control data for analyzing mutant samples with issues like incomplete penetrance and variable expressivity.
• To investigate the impact of developmental substage on automated phenotyping outcomes.

Main Methods:

• Development of an automated computational pipeline for 3D mouse embryo image analysis.
• Image pre-processing, registration, statistical analysis, and segmentation.
• Creation of a novel anatomical E14.5 mouse embryo average atlas.
• Utilizing the LAMA tool with the new atlas to identify dysmorphology.

Main Results:

• The pipeline effectively processes large volumes of 3D embryo imaging data.
• The automated system addresses challenges associated with low mutant sample numbers and variable data.
• Known and novel dysmorphologies were identified in two IMPC knockout mouse lines using the developed atlas and LAMA.
• The study assessed the influence of developmental substage on automated phenotyping accuracy.

Conclusions:

• The automated computational phenotyping pipeline facilitates efficient analysis of 3D mouse embryo imaging data.
• This approach enhances the integration of multi-dimensional phenotyping data within the IMPC.
• The novel E14.5 embryo atlas and LAMA integration enable robust identification of developmental abnormalities.