Constructing and optimizing 3D atlases from 2D data with application to the developing mouse brain

Area of Science:

• Neuroscience
• Computational Biology
• Medical Imaging

Background:

• Accurate 3D reference atlases are crucial for interpreting 3D imaging data.
• Existing methods for 3D atlas generation from 2D data produce artifacts, and manual methods are time-consuming.

Purpose of the Study:

• To develop a computational approach for constructing artifact-reduced 3D brain atlases.
• To improve the accuracy and comprehensiveness of 3D atlases for quantitative analysis.

Main Methods:

• A novel computational pipeline was developed to identify anatomical boundaries in imaging data.
• These boundaries guide 3D transformations, reducing artifacts and enabling atlas extension to edge regions.
• The method was applied to the Allen Developing Mouse Brain Atlas (ADMBA) across eight developmental stages.

Main Results:

• The new method generated more comprehensive and accurate 3D atlases.
• Validation using 15 whole mouse brains showed qualitative and quantitative improvements.
• A 37% greater alignment between the atlas and anatomical boundaries was observed.

Conclusions:

• The developed computational approach significantly reduces artifacts in 3D atlas construction.
• The MagellanMapper software and 3D ADMBA atlases provide valuable resources for quantitative analysis.
• These resources facilitate whole-organ analysis across samples and developmental stages.