Spatiotemporal 3D image registration for mesoscale studies of brain development

Area of Science:

• Neuroscience
• Developmental Biology
• Medical Imaging

Background:

• Accurate registration of 3D brain images is crucial for comparing samples across different developmental stages.
• Existing registration tools struggle with perinatal brains due to rapid growth and variable developmental rates.
• Perinatal brain registration requires specialized algorithms to handle dynamic morphological changes.

Purpose of the Study:

• To introduce CORGI (Customizable Object Registration for Groups of Images), an algorithm designed for accurate perinatal brain registration.
• To address the limitations of current methods in handling developmental variations in perinatal brain imaging.
• To provide an automated solution for mapping, comparing, and analyzing perinatal brain development.

Main Methods:

• Optimized image preprocessing to enhance sensitivity to registration mismatches.
• Developed an attention-gated simulated annealing procedure to focus on perinatal brain differences.
• Applied classical multidimensional scaling (CMDS) for temporal alignment and synchronization of brain samples based on individual development paces.
• Tested CORGI on 28 whole-mounted perinatal mouse brains (P0-P9).

Main Results:

• CORGI demonstrated accurate registration of perinatal brains, outperforming existing algorithms.
• The algorithm achieved a runtime of minutes per brain on a standard laptop.
• CORGI successfully automated tasks including atlas mapping, experimental group comparison, and monitoring brain development dynamics.

Conclusions:

• CORGI offers a robust and efficient solution for perinatal brain registration.
• The algorithm overcomes key challenges associated with developmental variability in perinatal neuroimaging.
• CORGI facilitates advanced analysis of brain development, enabling more precise comparative studies.