Quantifying the distribution of specific neurons throughout the whole brain is crucial for understanding physiological actions, pathological alterations and pharmacological treatments. However, the precise cell number and density of specific neurons in the entire brain remain unknown because of a lack of suitable research tools. Here, we propose a pipeline to automatically acquire and analyse the brain-wide distribution of type-specific neurons in a mouse brain. We employed a Brain-wide Positioning System to collect high-throughput anatomical information with the co-localized cytoarchitecture of the whole brain at subcellular resolution and utilized the NeuroGPS algorithm to locate and count cells in the whole brain.

• Zeiss Atlas 9000
• Zeiss Atlas 995 Topographer

NISCA All-America. 2018-2019 NISCA/Swim Outlet.com Girls Scholar Team Award: 7/9/2019: GIRLS GOLD TEAM AWARD. Them for teenage secondary school-kids who want decent microscopes at home, they will be clued up to. All the above pictures were taken with an old 3.3MP Coolpix 995 camera held by hand against the compound. Zeiss's 'The clean microsope', i.e. How to keep your microscope clean. Download The clean microscope. Software All Software latest This Just In Old School Emulation MS-DOS Games Historical Software Classic PC Games Software Library. Internet Arcade. Top Community Software Kodi Archive and Support File MS-DOS CD-ROM Software APK Vintage Software CD-ROM Software Library. Full text of 'School document'.

We evaluated the data continuity of the 3D dataset and the accuracy of stereological cell counting in 3D. To apply this pipeline, we acquired and quantified the brain-wide distributions and somatic morphology of somatostatin-expressing neurons in transgenic mouse brains. The results indicated that this whole-brain cell counting pipeline has the potential to become a routine tool for cell type neuroscience studies. Validation of the SOM-Cre mouse line We first confirmed the accuracy of using the SOM-IRES-Cre:Ai3-EYFP transgenic mouse (Fig. ) to genetically target SOM neurons. We performed immunostaining using an antibody against SOM in sections from the primary motor area (MOp), the hippocampal formation (HPF) and the caudoputamen (CP) and collected images using a Zeiss 710 confocal microscope with a 20× objective (Fig. ). The overlapping images of fluorescently labelled somas and immunostained molecules showed that the majority of the enhanced yellow fluorescent protein (EYFP)-labelled neurons were SOM-positive (SOM+), consistent with previous studies. These results indicated that SOM-IRES- Cre driver lines reliably target SOM neurons and can be used to study the distribution of all SOM neurons in the whole brain.

Distribution of EYFP-labelled interneurons and specific co-localization with SOM. ( a) Gene elements of the Ai3 Cre-reporter mouse and SOM-Cre-line mouse used in the experiments. ( b) Representative examples of specific co-localization of EYFP-labelled neurons (top, green) from a SOM-IRES-Cre:Ai3-EYFP mouse with immunofluorescence for somatostatin (middle, red).

Merged images are shown at the bottom. Scale bars, 100 µm. Nearly all of the EYFP-labelled neurons from SOM-IRES-Cre:Ai3-EYFP mice co-localized with somatostatin in MOp, HPF and CP. Abbreviations: MOp, Primary motor area; HPF, Hippocampal formation and CP, Caudoputamen. SOM-positive rates in MOp, HPF and CP were 74.27 ± 1.31% (n = 3), 80.80 ± 1.39% (n = 3) and 77.53 ± 2.19% (n = 3), respectively. Whole-brain optical imaging We acquired the whole-brain datasets from three SOM-IRES-Cre:Ai3-EYFP mice, including EYFP-labelled SOM neurons and PI-stained anatomical reference, in coronal planes (Fig. ). Figure show typical SOM distribution and co-localized cytoarchitecture in the hippocampal coronal plane, respectively.

The PI-stained landmarks enabled us to easily identify the distinct laminar distribution of SOM neurons with different cell densities in different layers in a typical six-layer structure in the cortex in Fig. The overlap of GFP- and PI-channels suggested that a voxel resolution of 0.32 × 0.32 × 2 μm enables orientation of the neuronal cell bodies based on counterstained nuclei at single cell resolution (inset in Fig. ).

The high resolution and data continuity of BPS facilitated brain reconstruction along the sagittal direction by resampling original coronal images at 0.32 × 0.32 × 2 μm. SOM cells were primarily distributed in layers IV and V, with a small number of cells detected in layers II/III and VIa. Figure shows the largest sagittal plane of a SOM-IRES-Cre:Ai3-EYFP mouse brain including almost all major brain regions. This result also demonstrated the data integrity of the whole-brain dataset and the variable density of SOM neuron distribution in different brain regions. Enlarged views of white square boxes in Fig. Illustrated that BPS enabled imaging of the entire brain in high enough detail to distinguish individual cell bodies in brain regions with variable cell densities (insets in Fig. ). These results demonstrate the ability of BPS to visualize single neurons for cell localization and counting in entire brains.

Whole-brain imaging of a SOM-IRES-Cre:Ai3-EYFP mouse brain. Images of EYFP-labelled neurons ( a) and PI-stained cytoarchitecture ( b) in the hippocampal coronal plane. ( c) Enlarged views of white rectangular boxes indicated in ( a) and ( b) and their merged image. Image size is 900 × 400 μm. Projection thickness in the EYFP channel was 20 μm.

Zeiss Atlas 9000

( d) Sagittal reconstruction of maximum intensity projections of the SOM-IRES-Cre:Ai3-EYFP mouse brain. The projection thickness shown in ( d) is 20 μm. The insets (d1)–(d5) are enlarged views of white square boxes shown in ( d). The sizes of (d1)–(d5) are 400 × 400 × 20 μm. Scale bar: ( a) and ( b) 1 mm, ( c) 25 μm, (c1)–(c3) 50 μm, ( d) 1 mm and (d1)–(d5) 100 μm.

Stereological cell counting Accurately and automatically segmenting, locating and counting cell bodies is a prerequisite for quantifying the whole-brain cell distribution. To validate the accuracy of stereological cell counting, we compared NeuroGPS cell detection with manual identification (Fig. ).

Figure depicts cell localization in a 100 × 100 × 100 μm data cube in the superior colliculus (SC) from the SOM-IRES-Cre:Ai3-EYFP mouse brain dataset. A total of 59 soma centres were automatically identified (indicated by red dots) using NeuroGPS, whereas 57 neurons were manually distinguished. Two background signals were mistakenly identified as neurons (indicated by yellow arrows in Fig. ). The recall and precision of automatic localization were calculated as 100% and 96.7%, respectively, for the overlap of automatically identified and manually detected cell bodies.

We also generated a maximum intensity projection of this data cube to simulate traditional cell quantification in histological tissue sections (Fig. ). Due to no axial resolving power, some disjunctive but aligned cells in the axial direction of the data cube (indicated by red arrows in Fig. ) were partially overlapping in the projection image (indicated by red arrows in Fig. ), resulting in segmentation difficulties and potential counting errors. We compared the accuracy of stereological counting in the full volume and planar cell counting in the z-projection image of five randomly selected data cubes (Fig. ). The results showed that the cell numbers obtained by counting in 3D full volumes were higher than those obtained by counting z-projection images. We performed a Wilcoxon signed-rank test and acquired a two-sided P value was 0.043, demonstrating a significant difference between 3D counting and Z-projection counting. This result indicates potential erroneous omission resulting from the overlap of individual neurons along the z-axis, demonstrating that automated stereological cell counting is the only adequate method for unbiased cell quantification in a whole brain. Furthermore, we also estimated the accuracy of automated cell counting in 18 SOM-expressing brain regions from the three datasets (Fig. ).

We randomly selected a representative volume of 512 × 512 × 512 μm in each region in each brain. The highest and lowest correct identification recall rates in these regions were 100% and 86.5%, respectively. The highest and lowest precision rates for correct identification in these regions were 100% and 85.2%, respectively.

All three datasets exhibited identification rates higher than the standard of 85.0%. Average whole-brain recall and precision rates were 94.0 ± 0.4% and 91.8 ± 0.5% (n = 3), respectively. Furthermore, we randomly selected 20 neurons in each data cube and calculated their cell brightness. Pairwise comparisons between different datasets shows that there was no significant difference in the averaged cell brightness (p 0.5 for all comparisons).

It demonstrates slight brightness changes due to individual difference have no effect on cell counting. These results illustrated that stereological cell counting in whole-brain datasets using the NeuroGPS algorithm can accurately detect and segment soma as a potential method for quantifying brain-wide cell distribution. Stereological cell counting accuracy using the NeuroGPS algorithm. ( a) Automatically locating the neurons in an SC data cube from a SOM-IRES-Cre:Ai3-EYFP mouse brain dataset using NeuroGPS. Grey represents EYFP-labelled somas in the raw data, and red dots indicate identified soma centres using the NeuroGPS algorithm. Overlapping grey somas and red dots indicate the correct identification of neurons. Yellow arrows represent erroneous commission.

( b) The 100 μm-thick Z-projection image of the data cube in ( a). Red arrows represent some disjunctive cells in the z-direction in ( a) that partially overlap in ( b). ( c) Comparison between stereological and planar cell counting in 3D data cubes and their own z-projection images, respectively. ( d) Accuracy of automated stereological cell counting in the regions of SOM-IRES-Cre:Ai3-EYFP mice brains (n = 3). Abbreviations: MOs, Secondary motor area; LS, Lateral septal nucleus; BST, Bed nuclei of the stria terminalis; cc, Corpus callosum; PIR, Piriform area; SSp, Primary somatosensory area; SSs, Supplemental somatosensory area; BLA, Basolateral amygdala nucleus; LA, Lateral amygdala nucleus; VISp, Primary visual area; VISl, Lateral visual areas; HY, Hypothalamus; CEA, Central amygdala nucleus; SC, Superior colliculus; CB, Cerebellum. Brain-wide cell distribution of SOM neurons Figure shows the typical distribution of SOM-expressing neurons in a mouse brain.

Zeiss Atlas 995 Topographer

SOM was ubiquitously expressed in the brain, as shown in the volume rendering in Fig., in brain regions consistent with previous studies. Figure show a series of fluorescently labelled representative coronal plane images from the same dataset, including main SOM-expressing brain regions. The insets of the enlarged views of these regions demonstrated that different brain regions exhibit different densities of SOM-expressing neurons. SOM expression was densest in SC in the brain stem. SOM was also relatively highly expressed in cerebral nuclei in the LS, CEA and BST regions, whereas SOM neurons were widely distributed with relatively low densities in the MOp, MOs, SSp, SSs, PIR, VISp, VISl and HPF regions of the cerebral cortex, the BLA and LA in the cortex subplate and the HY in the brain stem.

There were relatively few SOM neurons in the CP in the brain stem and cc, although some densely arranged SOM neurons were observed in the CB. Some of this distribution variability within the SOM population in different regions might be functionally significant. These results provide a distribution reference for the functional study of SOM-expressing neurons. Brain-wide distribution of SOM-expressing neurons. ( a) 3D volume rendering of the whole-brain dataset of a SOM-IRES-Cre:Ai3-EYFP mouse.

Green represents the brain-wide expression of SOM. Dashed lines indicate the locations of images shown in ( b– g). ( b– g) A series of representative coronal images showing the distribution of SOM neurons in selected brain regions at a resampling resolution of 1 × 1 × 2 μm. The insets show corresponding partial enlarged images in ( b– g) at the original resolution of 0.32 × 0.32 × 2 μm. Scale bar: ( b– g) 1 mm and (insets) 50 μm.

Morphological reconstruction of SOM neurons Soma segmentation not only provides the central location of the cell body but also morphological characteristics of neuronal somas. We observed that SOM-expressing neurons in the same nuclei had similar morphologies, which might be used to classify subtypes of SOM-expressing neurons. We selected and reconstructed a typical 3D morphology of SOM-expressing neurons at the original resolution in 18 brain regions (Fig. ). Most cell bodies showed an ovoid or spindle profile with various cell volumes.

SOM neuronal cell bodies in the CB were obviously bigger than all of the other somas. Furthermore, we randomly selected a data cube of 100 × 100 × 100 μm in each brain region, measured the longest and shortest radii, surface area and cell volume of all fluorescently labelled somas in the data cube, and calculated the corresponding average radius and longest to shortest radii ratio (Fig. ). The results demonstrated that both the longest and shortest radii and the surface area and cell volume of SOM-expressing neurons in the CB were higher than in other regions. However, the longest to shortest radii ratios for neurons in the CB were similar to those in other regions.

This result indicated that SOM-expressing neurons had similar soma profiles in different regions, consistent with the reconstructions shown in Fig. Except for the CB, SOM-expressing neurons in other regions exhibited similar ranges of shortest radii and various longest radii, demonstrating similar characteristics for average radii and the longest to shortest radii ratio. The longest and shortest radii in BST and CEA were overall longer, thus the corresponding surface area and cell volume were bigger.

The morphological features of SOM-expressing neurons in the neocortex were relatively consistent. We also performed a statistical analysis on the morphological traits of SOM neuronal somas (Fig. ). There was no significant difference in the soma morphological features for the pairwise comparisons between most regions. Some regions presented differently compared with other regions. For example, CB showed a significant difference in most morphological parameters with other regions, except the longest to shortest radii ratio. It indicates that combing physiological experiment and gene expressing measurement in the future might distinguish if SOM neurons in the CB belong to specific subtype of SOM neurons. These results illustrated that the platform used in this study enabled the accurate reconstruction and extraction of morphological features for cell type studies.

Morphological features of SOM neuronal somas. ( a) Morphological reconstructions of typical SOM neuronal somas in specific brain regions. ( b) Morphological features of SOM neuronal somas in spe cific brain regions in a data cube of 100 × 100 × 100 μm (n = 3).

Error bars represent SEM. ( c) Statistical significance of morphological features of SOM neuronal somas between different regions. Corresponding to ( b), cyan, blue, yellow, grey, green and purple represent the long, short and average radii, longest to shortest radii ratio, surface area and cell volume, respectively. Quantifying the whole-brain distribution of SOM neurons Quantitative mapping of a given type of neuron in the entire brain can improve our understanding of this neuron type in specific brain functions. To explore this diversity, we comprehensively quantified the distributions of SOM-expressing neurons in each region of three brains (Fig. ). The total number of SOM-expressing neurons in a SOM-IRES-Cre:Ai3-EYFP mouse brain was 1,901,273 ± 67,096 (Fig. ). In terms of numbers, the most prominent labelling was observed in the HPF (130,431 ± 2,531 cells) due to its large volume, with denser labelling than reported in a previous study.

The numbers in the SSp were also higher than in other regions (111,365 ± 2,965 cells). SC, HY, LS and PIR showed 59 × 10 4 SOM-expressing neurons, whereas the CP, MOs, CB, MOp, VISp, CEA, SSs, BST and BLA have 15 × 10 4 SOM-expressing neurons. LA, VISl and cc exhibit less than 1 × 10 4 SOM-expressing neurons. There were only 2,996 ± 29 SOM-expressing neurons in the cc. Cell density showed distinct diversity in terms of cell number, reflecting a volumetric difference in various brain regions.

The qualitative impression of the brain-wide distribution patterns was supported after calculating the cell density. Across the whole brain, the most abundant labelling was found in the SC, LS and CEA, consistent with Fig. The cell densities in these regions were 48,130 ± 1,278, 41,140 ± 1,708 and 34,370 ± 498 cells/mm 3, respectively. The regions in the neocortex had similar cell densities, consistent with previous studies, and the cells were much denser than the 2D density reported in a previous study. The average cell density in the neocortex was 10,735 ± 422 cells/mm 3. The cell density of HY in the brain stem was similar to the neocortex.

CP and CB showed low SOM expression densities of 4385 ± 233 and 3882 ± 91 cells/mm 3, respectively, reflecting their large volumes. SOM was sparsely expressed in the cc (2431 ± 116 cells/mm 3).

We also compared the cell distributions between different regions (Fig. ). Different from the statistical results of the soma morphology (Fig. ), both cell number and density presented a significant diversity among different regions. There was a significant difference in the cell number for the pairwise comparisons of most regions. For cell density, there was no significant difference for the pairwise comparisons of sub-regions of cortex, cortex and HY, CP, cc and CB.

The other pairs presented differently. This distribution information could potentially be used as a reference for studying SOM neurons, and these results demonstrated that this platform could be used for the quantitative analysis of cell distribution in the whole brain. Understanding the principles of brain functions requires the systematic characterization of participating cell types. Establishing methods to efficiently identify and quantify specific neurons will facilitate our understanding of how neurons are involved in specific functions. Advances in whole-brain optical imaging and stereological cell localization provide a method to answer fundamental questions concerning cell distribution and cell number. BPS enabled the acquisition of fine structural information with the co-located anatomical reference of labelled neurons in the whole brain at subcellular resolution. NeuroGPS is capable of automatically detecting, locating and counting cell bodies in 3D.

In this study, we combined these two technologies to constitute a novel platform to quantify a comprehensive distribution of type-specific neurons. To demonstrate the utility of this platform, we obtained high-resolution brain-wide datasets of SOM-IRES-Cre:Ai3-EYFP mouse brains with co-located cytoarchitecture and quantitatively analysed the distribution features of SOM expression and the morphological features of SOM neural somas. The results provide a platform with the capability of precise orientation, fine soma morphological reconstruction and accurate cell counting of a given type of neuron in the whole brain. The automatic whole-brain optical imaging and stereoscopic cell counting and reconstruction demonstrated an unprecedented ability to accurately and quantitatively detect and analyse the distributional features of specific neurons in the whole brain. Acquiring a 3D brain-wide structural dataset is a prerequisite for accurately locating and counting the cells throughout the whole brain in the stereological approach. In conventional 2D approaches, information overlap between axial projections easily leads to cell omission and increases the difficulty of segmenting and counting. Interval sampling decreases the workload of manual data acquisition but fails to depict the real topological organization of the neurons.

In contrast, visualizing neurons with full-volumetric whole-brain imaging overcomes these limitations and records realistic brain-wide cell organization in a reliable and efficient manner. Whether or not to precisely segment axial-aligned or adjacent somas in 3D is the key factor influencing automatic locating and counting accuracy, and this step requires micron- or even submicron-scale imaging resolution in the whole brain, although the sizes of neuronal somas are at the scale of tens of microns. The strategies of combing optical imaging and mechanical sectioning for whole-brain imaging.