Ontology-guided segmentation and object identification for developmental mouse lung immunofluorescent images

Area of Science:

• Biomedical imaging
• Computational biology
• Developmental biology

Background:

• Immunofluorescent confocal microscopy is vital for cell type discrimination in developmental mouse lung images.
• Complex anatomical structures in these images pose segmentation challenges due to context, probe variability, data costs, and dense packing.
• Application ontologies can provide biological context to aid human experts in histological structure identification.

Purpose of the Study:

• To develop a semi-supervised approach for analyzing complex anatomical structures in immunofluorescence images.
• To leverage application ontologies for simplified image segmentation and object identification.
• To demonstrate the utility of ontology-guided processing for complex biological images.

Main Methods:

• Utilized an application ontology to provide simplified context for image segmentation and object identification.
• Employed semi-supervised analysis for complex and densely packed anatomical structures.
• Integrated image data with metadata to provide meaningful biological context.

Main Results:

• Demonstrated ontology-guided segmentation and object identification in mouse developmental lung images.
• Showcased how logical organization of biological facts in an ontology facilitates automatic processing of complex images.
• Successfully applied the approach to data from the Molecular Atlas of Lung Development (LungMAP) program.

Conclusions:

• Introduced a novel ontology-guided approach for segmentation and classification of complex immunofluorescence images.
• The ontology automatically generates image-specific constraints based on biomedical context, simplifying segmentation and classification.
• This method enhances the analysis of intricate histological structures in developmental biology research.