Virtual mouse brain histology from multi-contrast MRI via deep learning
- New York University School of Medicine, United States
- University of Maryland, United States
- Memorial Sloan Kettering Cancer Center
- University at Buffalo, State University of New York, United States
Abstract
1H MRI maps brain structure and function non-invasively through versatile contrasts that exploit inhomogeneity in tissue micro-environments. Inferring histopathological information from MRI findings, however, remains challenging due to absence of direct links between MRI signals and cellular structures. Here, we show that deep convolutional neural networks, developed using co-registered multi-contrast MRI and histological data of the mouse brain, can estimate histological staining intensity directly from MRI signals at each voxel. The results provide three-dimensional maps of axons and myelin with tissue contrasts that closely mimics target histology and enhanced sensitivity and specificity compared to conventional MRI markers. Furthermore, the relative contribution of each MRI contrast within the networks can be used to optimize multi-contrast MRI acquisition. We anticipate our method to be a starting point for translation of MRI results into easy-to-understand virtual histology for neurobiologists and provide resources for validating novel MRI techniques.
Data availability
All data and source codes used in this study are available at https://www.github.com/liangzifei/MRH-net/. The data can also be found at datadryad.org
-
Data fromMulti-contrast MRI and histology datasets used to train and validate MRH networks to generate virtual mouse brain histologyDryad Digital Repository, doi:10.5061/dryad.1vhhmgqv8.
-
Allen Mouse Brain AtlasThe reference data at http://connectivity.brain-map.org/static/referencedata.
Article and author information
Author details
Funding
Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01HD074593)
- Jiangyang Zhang
National Institute of Neurological Disorders and Stroke (R01NS102904)
- Jiangyang Zhang
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Ethics
Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to approved institutional animal care and use committee (IACUC) protocols (s16-00145-133) of the New York University.
Copyright
© 2022, Liang et al.
This article is distributed under the terms of the Creative Commons Attribution License permitting unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 2,510
- views
-
- 347
- downloads
-
- 3
- citations
Views, downloads and citations are aggregated across all versions of this paper published by eLife.
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Virtual mouse brain histology from multi-contrast MRI via deep learning
eLife 11:e72331.
https://doi.org/10.7554/eLife.72331
Further reading
-
- Medicine
- Neuroscience
Interactions between top-down attention and bottom-up visceral inputs are assumed to produce conscious perceptions of interoceptive states, and while each process has been independently associated with aberrant interoceptive symptomatology in psychiatric disorders, the neural substrates of this interface are unknown. We conducted a preregistered functional neuroimaging study of 46 individuals with anxiety, depression, and/or eating disorders (ADE) and 46 propensity-matched healthy comparisons (HC), comparing their neural activity across two interoceptive tasks differentially recruiting top-down or bottom-up processing within the same scan session. During an interoceptive attention task, top-down attention was voluntarily directed towards cardiorespiratory or visual signals. In contrast, during an interoceptive perturbation task, intravenous infusions of isoproterenol (a peripherally-acting beta-adrenergic receptor agonist) were administered in a double-blinded and placebo-controlled fashion to drive bottom-up cardiorespiratory sensations. Across both tasks, neural activation converged upon the insular cortex, localizing within the granular and ventral dysgranular subregions bilaterally. However, contrasting hemispheric differences emerged, with the ADE group exhibiting (relative to HCs) an asymmetric pattern of overlap in the left insula, with increased or decreased proportions of co-activated voxels within the left or right dysgranular insula, respectively. The ADE group also showed less agranular anterior insula activation during periods of bodily uncertainty (i.e. when anticipating possible isoproterenol-induced changes that never arrived). Finally, post-task changes in insula functional connectivity were associated with anxiety and depression severity. These findings confirm the dysgranular mid-insula as a key cortical interface where attention and prediction meet real-time bodily inputs, especially during heightened awareness of interoceptive states. Furthermore, the dysgranular mid-insula may indeed be a ‘locus of disruption’ for psychiatric disorders.
-
- Medicine
Glucagon-like peptide 1 (GLP-1) is a gut-derived hormone secreted by intestinal L cells and vital for postprandial glycemic control. As open-type enteroendocrine cells, whether L cells can sense mechanical stimuli caused by chyme and thus regulate GLP-1 synthesis and secretion is unexplored. Molecular biology techniques revealed the expression of Piezo1 in intestinal L cells. Its level varied in different energy status and correlates with blood glucose and GLP-1 levels. Mice with L cell-specific loss of Piezo1 (Piezo1 IntL-CKO) exhibited impaired glucose tolerance, increased body weight, reduced GLP-1 production and decreased CaMKKβ/CaMKIV-mTORC1 signaling pathway under normal chow diet or high-fat diet. Activation of the intestinal Piezo1 by its agonist Yoda1 or intestinal bead implantation increased the synthesis and secretion of GLP-1, thus alleviated glucose intolerance in diet-induced-diabetic mice. Overexpression of Piezo1, Yoda1 treatment or stretching stimulated GLP-1 production and CaMKKβ/CaMKIV-mTORC1 signaling pathway, which could be abolished by knockdown or blockage of Piezo1 in primary cultured mouse L cells and STC-1 cells. These experimental results suggest a previously unknown regulatory mechanism for GLP-1 production in L cells, which could offer new insights into diabetes treatments.
