DeepMapi: a Fully Automatic Registration Method for Mesoscopic Optical Brain Images Using Convolutional Neural Networks

Connecto-informatics at the mesoscale: current advances in image processing and analysis for mapping the brain connectivity

Mapping neural connections within the brain has been a fundamental goal in neuroscience to understand better its functions and changes that follow aging and diseases. Developments in imaging technology, such as microscopy and labeling tools, have allowed researchers to visualize this connectivity through high-resolution brain-wide imaging. With this, image processing and analysis have become more crucial. However, despite the wealth of neural images generated, access to an integrated image processing and analysis pipeline to process these data is challenging due to scattered information on available tools and methods. To map the neural connections, registration to atlases and feature extraction through segmentation and signal detection are necessary. In this review, our goal is to provide an updated overview of recent advances in these image-processing methods, with a particular focus on fluorescent images of the mouse brain. Our goal is to outline a pathway toward an integrated image-processing pipeline tailored for connecto-informatics. An integrated workflow of these image processing will facilitate researchers' approach to mapping brain connectivity to better understand complex brain networks and their underlying brain functions. By highlighting the image-processing tools available for fluroscent imaging of the mouse brain, this review will contribute to a deeper grasp of connecto-informatics, paving the way for better comprehension of brain connectivity and its implications.

The cytoarchitectonic landscape revealed by deep learning method facilitated precise positioning in mouse neocortex

Neocortex is a complex structure with different cortical sublayers and regions. However, the precise positioning of cortical regions can be challenging due to the absence of distinct landmarks without special preparation. To address this challenge, we developed a cytoarchitectonic landmark identification pipeline. The fluorescence micro-optical sectioning tomography method was employed to image the whole mouse brain stained by general fluorescent nucleotide dye. A fast 3D convolution network was subsequently utilized to segment neuronal somas in entire neocortex. By approach, the cortical cytoarchitectonic profile and the neuronal morphology were analyzed in 3D, eliminating the influence of section angle. And the distribution maps were generated that visualized the number of neurons across diverse morphological types, revealing the cytoarchitectonic landscape which characterizes the landmarks of cortical regions, especially the typical signal pattern of barrel cortex. Furthermore, the cortical regions of various ages were aligned using the generated cytoarchitectonic landmarks suggesting the structural changes of barrel cortex during the aging process. Moreover, we observed the spatiotemporally gradient distributions of spindly neurons, concentrated in the deep layer of primary visual area, with their proportion decreased over time. These findings could improve structural understanding of neocortex, paving the way for further exploration with this method.

Detecting the effect of genetic diversity on brain composition in an Alzheimer's disease mouse model

Alzheimer's disease (AD) is broadly characterized by neurodegeneration, pathology accumulation, and cognitive decline. There is considerable variation in the progression of clinical symptoms and pathology in humans, highlighting the importance of genetic diversity in the study of AD. To address this, we analyze cell composition and amyloid-beta deposition of 6- and 14-month-old AD-BXD mouse brains. We utilize the analytical QUINT workflow- a suite of software designed to support atlas-based quantification, which we expand to deliver a highly effective method for registering and quantifying cell and pathology changes in diverse disease models. In applying the expanded QUINT workflow, we quantify near-global age-related increases in microglia, astrocytes, and amyloid-beta, and we identify strain-specific regional variation in neuron load. To understand how individual differences in cell composition affect the interpretation of bulk gene expression in AD, we combine hippocampal immunohistochemistry analyses with bulk RNA-sequencing data. This approach allows us to categorize genes whose expression changes in response to AD in a cell and/or pathology load-dependent manner. Ultimately, our study demonstrates the use of the QUINT workflow to standardize the quantification of immunohistochemistry data in diverse mice, - providing valuable insights into regional variation in cellular load and amyloid deposition in the AD-BXD model.

Subregion preference in the long-range connectome of pyramidal neurons in the medial prefrontal cortex

BACKGROUND

The medial prefrontal cortex (mPFC) is involved in complex functions containing multiple types of neurons in distinct subregions with preferential roles. The pyramidal neurons had wide-range projections to cortical and subcortical regions with subregional preferences. Using a combination of viral tracing and fluorescence micro-optical sectioning tomography (fMOST) in transgenic mice, we systematically dissected the whole-brain connectomes of intratelencephalic (IT) and pyramidal tract (PT) neurons in four mPFC subregions.

RESULTS

IT and PT neurons of the same subregion projected to different target areas while receiving inputs from similar upstream regions with quantitative differences. IT and PT neurons all project to the amygdala and basal forebrain, but their axons target different subregions. Compared to subregions in the prelimbic area (PL) which have more connections with sensorimotor-related regions, the infralimbic area (ILA) has stronger connections with limbic regions. The connection pattern of the mPFC subregions along the anterior-posterior axis showed a corresponding topological pattern with the isocortex and amygdala but an opposite orientation correspondence with the thalamus.

CONCLUSIONS

By using transgenic mice and fMOST imaging, we obtained the subregional preference whole-brain connectomes of IT and pyramidal tract PT neurons in the mPFC four subregions. These results provide a comprehensive resource for directing research into the complex functions of the mPFC by offering anatomical dissections of the different subregions.

Connecting single-cell transcriptomes to projectomes in mouse visual cortex

The mammalian brain is composed of diverse neuron types that play different functional roles. Recent single-cell RNA sequencing approaches have led to a whole brain taxonomy of transcriptomically-defined cell types, yet cell type definitions that include multiple cellular properties can offer additional insights into a neuron's role in brain circuits. While the Patch-seq method can investigate how transcriptomic properties relate to the local morphological and electrophysiological properties of cell types, linking transcriptomic identities to long-range projections is a major unresolved challenge. To address this, we collected coordinated Patch-seq and whole brain morphology data sets of excitatory neurons in mouse visual cortex. From the Patch-seq data, we defined 16 integrated morpho-electric-transcriptomic (MET)-types; in parallel, we reconstructed the complete morphologies of 300 neurons. We unified the two data sets with a multi-step classifier, to integrate cell type assignments and interrogate cross-modality relationships. We find that transcriptomic variations within and across MET-types correspond with morphological and electrophysiological phenotypes. In addition, this variation, along with the anatomical location of the cell, can be used to predict the projection targets of individual neurons. We also shed new light on infragranular cell types and circuits, including cell-type-specific, interhemispheric projections. With this approach, we establish a comprehensive, integrated taxonomy of excitatory neuron types in mouse visual cortex and create a system for integrated, high-dimensional cell type classification that can be extended to the whole brain and potentially across species.

D-LMBmap: a fully automated deep-learning pipeline for whole-brain profiling of neural circuitry

Recent proliferation and integration of tissue-clearing methods and light-sheet fluorescence microscopy has created new opportunities to achieve mesoscale three-dimensional whole-brain connectivity mapping with exceptionally high throughput. With the rapid generation of large, high-quality imaging datasets, downstream analysis is becoming the major technical bottleneck for mesoscale connectomics. Current computational solutions are labor intensive with limited applications because of the exhaustive manual annotation and heavily customized training. Meanwhile, whole-brain data analysis always requires combining multiple packages and secondary development by users. To address these challenges, we developed D-LMBmap, an end-to-end package providing an integrated workflow containing three modules based on deep-learning algorithms for whole-brain connectivity mapping: axon segmentation, brain region segmentation and whole-brain registration. D-LMBmap does not require manual annotation for axon segmentation and achieves quantitative analysis of whole-brain projectome in a single workflow with superior accuracy for multiple cell types in all of the modalities tested.

DeepSlice: rapid fully automatic registration of mouse brain imaging to a volumetric atlas

Registration of data to a common frame of reference is an essential step in the analysis and integration of diverse neuroscientific data. To this end, volumetric brain atlases enable histological datasets to be spatially registered and analyzed, yet accurate registration remains expertise-dependent and slow. In order to address this limitation, we have trained a neural network, DeepSlice, to register mouse brain histological images to the Allen Brain Common Coordinate Framework, retaining registration accuracy while improving speed by >1000 fold.

Detecting the effect of genetic diversity on brain composition in an Alzheimer's disease mouse model

Alzheimer's disease (AD) is characterized by neurodegeneration, pathology accumulation, and progressive cognitive decline. There is significant variation in age at onset and severity of symptoms highlighting the importance of genetic diversity in the study of AD. To address this, we analyzed cell and pathology composition of 6- and 14-month-old AD-BXD mouse brains using the semi-automated workflow (QUINT); which we expanded to allow for nonlinear refinement of brain atlas-registration, and quality control assessment of atlas-registration and brain section integrity. Near global age-related increases in microglia, astrocyte, and amyloid-beta accumulation were measured, while regional variation in neuron load existed among strains. Furthermore, hippocampal immunohistochemistry analyses were combined with bulk RNA-sequencing results to demonstrate the relationship between cell composition and gene expression. Overall, the additional functionality of the QUINT workflow delivers a highly effective method for registering and quantifying cell and pathology changes in diverse disease models.

Three-dimensional mapping in multi-samples with large-scale imaging and multiplexed post staining

Dissection of the anatomical information at the single-cell level is crucial for understanding the organization rule and pathological mechanism of biological tissues. Mapping the whole organ in numerous groups with multiple conditions brings the challenges in imaging and analysis. Here, we describe an approach, named array fluorescent micro-optical sectioning tomography (array-fMOST), to identify the three-dimensional information at single-cell resolution from multi-samples. The pipeline contains array embedding, large-scale imaging, post-imaging staining and data analysis, which could image over 24 mouse brains simultaneously and collect the slices for further analysis. With transgenic mice, we acquired the distribution information of neuropeptide somatostatin neurons during natural aging and compared the changes in the microenvironments by multi-component labeling of serial sections with precise co-registration of serial datasets quantitatively. With viral labeling, we also analyzed the input circuits of the medial prefrontal cortex in the whole brain of Alzheimer's disease and autism model mice. This pipeline is highly scalable to be applied to anatomical alterations screening and identification. It provides new opportunities for combining multi-sample whole-organ imaging and molecular phenotypes identification analysis together. Such integrated high-dimensional information acquisition method may accelerate our understanding of pathogenesis and progression of disease in situ at multiple levels.

Progressive 3D biomedical image registration network based on deep self-calibration

Three dimensional deformable image registration (DIR) is a key enabling technique in building digital neuronal atlases of the brain, which can model the local non-linear deformation between a pair of biomedical images and align the anatomical structures of different samples into one spatial coordinate system. And thus, the DIR is always conducted following a preprocessing of global linear registration to remove the large global deformations. However, imperfect preprocessing may leave some large non-linear deformations that cannot be handled well by existing DIR methods. The recently proposed cascaded registration network gives a primary solution to deal with such large non-linear deformations, but still suffers from loss of image details caused by continuous interpolation (information loss problem). In this article, a progressive image registration strategy based on deep self-calibration is proposed to deal with the large non-linear deformations without causing information loss and introducing additional parameters. More importantly, we also propose a novel hierarchical registration strategy to quickly achieve accurate multi-scale progressive registration. In addition, our method can implicitly and reasonably implement dynamic dataset augmentation. We have evaluated the proposed method on both optical and MRI image datasets with obtaining promising results, which demonstrate the superior performance of the proposed method over several other state-of-the-art approaches for deformable image registration.

SMART: An Open-Source Extension of WholeBrain for Intact Mouse Brain Registration and Segmentation

Mapping immediate early gene (IEG) expression across intact mouse brains allows for unbiased identification of brain-wide activity patterns underlying complex behaviors. Accurate registration of sample brains to a common anatomic reference is critical for precise assignment of IEG-positive ("active") neurons to known brain regions of interest (ROIs). While existing automated voxel-based registration methods provide a high-throughput solution, they require substantial computing power, can be difficult to implement and fail when brains are damaged or only partially imaged. Additionally, it is challenging to cross-validate these approaches or compare them to any preexisting literature based on serial coronal sectioning. Here, we present the open-source R package SMART (Semi-Manual Alignment to Reference Templates) that extends the WholeBrain R package framework to automated segmentation and semi-automated registration of intact mouse brain light-sheet fluorescence microscopy (LSFM) datasets. The SMART package was created for novice programmers and introduces a streamlined pipeline for aligning, registering, and segmenting LSFM volumetric datasets across the anterior-posterior (AP) axis, using a simple "choice game" and interactive menus. SMART provides the flexibility to register whole brains, partial brains or discrete user-chosen images, and is fully compatible with traditional sectioned coronal slice-based analyses. We demonstrate SMART's core functions using example datasets and provide step-by-step video tutorials for installation and implementation of the package. We also present a modified iDISCO+ tissue clearing procedure for uniform immunohistochemical labeling of the activity marker Fos across intact mouse brains. The SMART pipeline, in conjunction with the modified iDISCO+ Fos procedure, is ideally suited for examination and orthogonal cross-validation of brain-wide neuronal activation datasets.

High-Throughput Strategy for Profiling Sequential Section With Multiplex Staining of Mouse Brain

The brain modulates specific functions in its various regions. Understanding the organization of different cells in the whole brain is crucial for investigating brain functions. Previous studies have focused on several regions and have had difficulty analyzing serial tissue samples. In this study, we introduced a pipeline to acquire anatomical and histological information quickly and efficiently from serial sections. First, we developed a serial brain-slice-staining method to stain serial sections and obtained more than 98.5% of slices with high integrity. Subsequently, using the self-developed analysis software, we registered and quantified the signals of imaged sections to the Allen Mouse Brain Common Coordinate Framework, which is compatible with multimodal images and slant section planes. Finally, we validated the pipeline with immunostaining by analyzing the activity variance in the whole brain during acute stress in aging and young mice. By removing the problems resulting from repeated manual operations, this pipeline is widely applicable to serial brain slices from multiple samples in a rapid and convenient manner, which benefits to facilitate research in life sciences.

Mesoscale microscopy and image analysis tools for understanding the brain

Over the last ten years, developments in whole-brain microscopy now allow for high-resolution imaging of intact brains of small animals such as mice. These complex images contain a wealth of information, but many neuroscience laboratories do not have all of the computational knowledge and tools needed to process these data. We review recent open source tools for registration of images to atlases, and the segmentation, visualisation and analysis of brain regions and labelled structures such as neurons. Since the field lacks fully integrated analysis pipelines for all types of whole-brain microscopy analysis, we propose a pathway for tool developers to work together to meet this challenge.