The Brain/MINDS 3D Digital Marmoset Brain Atlas Version 2.0: Population-based Cortical Region Parcellations with Multi-Modal Standard Templates

Read the full article See related articles

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

We present the Brain/MINDS population-based 3D digital brain atlas version 2.0 (BMA2.0), a population-based 3D digital brain atlas of the common marmoset (Callithrix jacchus), designed to overcome limitations of previous single-subject atlases that are prone to structural biases arising from individual variation. This release integrates manually delineated cortical regions from 10 myelin-stained brains to generate a generalized cortical parcellation. Subcortical regions from a previous version (BMA2017) and a state-of-the-art cerebellum atlas (University of Pittsburgh) were also incorporated, resulting in a comprehensive whole-brain parcellation of 636 regions. To enhance cross-modal registration accuracy, we implemented artificial intelligence (AI)-based techniques, including CycleGAN and Pix2Pix for image translation and segmentation. These methods enabled accurate alignment across myelin, Nissl, block-face, and MRI modalities by providing both visual similarity and anatomically meaningful constraints. The atlas package includes population-average templates for myelin and Nissl staining (based on 10 individuals), a new ex vivo MRI standard space constructed from 91 brains, and an in vivo average brain from 446 individuals. Additionally, flat map projections and surface models (outer, mid, and inner cortical layers) are provided, facilitating multi-modal integration and spatial analysis. While applications such as quantification of laminar thickness are demonstrated, BMA2.0 is designed as a general-purpose platform for cross-modal registration, data integration, and comparative neuroscience. It supports spatially grounded analysis of marmoset brain structure and function and offers broad compatibility with existing datasets and software.

Article activity feed