Cell-type-specific computational neuroanatomy, simulations from the sagittal and coronal Allen Brain Atlas

TL;DR

This paper introduces a method to estimate variability and error margins in computational neuroanatomy analyses using the Allen Brain Atlas by simulating different image series, enhancing the reliability of brain region and cell type identification.

Contribution

It presents a novel approach to quantify uncertainty in neuroanatomical data analysis by leveraging multiple image series from the Allen Brain Atlas, including sagittal and coronal views.

Findings

Provides error estimates for neuroanatomical region definitions

Demonstrates variability in cell type specificity analysis

Enhances robustness of computational neuroanatomy results

Abstract

The Allen Atlas of the adult mouse brain is a brain-wide, genome-wide data set that has been made available online, triggering a renaissance in neuroanatomy. In particular, it has been used to define brain regions in a computational, data-driven way, and to estimate the region-specificity of cell types characterized independently by their transcriptional activity. However, these results were based on one series of co-registered (coronal) ISH image series per gene, whereas the online ABA contains several image series per genes, including sagittal ones. Since the sagittal series cover mostly the left hemisphere, we can simulate the variability of results by repeatedly drawing a random image series for each gene and restricting the computation to the left hemisphere. This gives rise to an estimate of error bars on the results of computational neuroanatomy.