Dec 18, 2024
Assessing histological image series for errors in serial order, flipping and mounting
This protocol is a draft, published without a DOI.
- 1University of Oslo
Protocol Citation: Ingvild Elise Bjerke 2024. Assessing histological image series for errors in serial order, flipping and mounting. protocols.io https://protocols.io/view/assessing-histological-image-series-for-errors-in-dpan5ide
License: This is an open access protocol distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: October 10, 2024
Last Modified: December 18, 2024
Protocol Integer ID: 109614
Funders Acknowledgements:
EU Horizon Europe
Grant ID: 101147319
Research Council of Norway
Grant ID: 269774
Abstract
An optimal histological series contains sections that are in the right order (along the axis of sectioning), oriented the same way (with the left and right hemisphere on the same side across all sections), with the midline straight and without mounting errors. This protocol outlines the steps taken to ensure this quality in a series of brain images.
Preparations
Preparations
A note on automation options: To facilitate efficient processing of large data collections, we provide Python code to create Nutil Transform files and Transform templates for evaluation. See the functions under "create_nut_file_functions.py" in https://github.com/ingvildeb/brain_section_scripts/
Many Nutil Transform files can also be run in batch using Nutil as a command line tool (see https://nutil.readthedocs.io/en/latest/NutilCmd.html)
The first step is to prepare the necessary materials, in this case small versions of the images to be evaluated and spreadsheets to note down any changes to be made. In the example used here, we will have one series stained for perineuronal nets and parvalbumin, and one cresyl violet stained series from the same brain.
Create small thumbnail images of all the images you will assess. This will make it faster to open and flip through the series. This can be achieved with the Transform function in the Nutil Software (https://nutil.readthedocs.io/en/latest/Transform.html).
Prepare spreadsheet to make notes. Use a spreadsheet to make notes on what should be done to each section. The Transform template provided with Nutil will be used in this example. This makes it so any changes indicated in the sheet can be automatically applied to the batch of images using Nutil Transform.
To get the Transform template, download the Nutil software from the NITRC homepage (https://www.nitrc.org/projects/nutil/) and find the template in the Transform folder of the downloaded package. Alternatively, create on automatically for a folder of files using our Python scripts (see Step 1).
Image setup
Image setup
To ease the process of comparing sections, open at least two consecutive images from the series. If you have more than one series from the same brain, open consecutive sections for each of them at the same time. In the current example, four images (two fluorescent stained and two neighbouring cresyl violet-stained ones) have been opened with the Windows Photos viewer and placed side-by-side on the screen. Holding the WIN key while using the arrow buttons makes it easier to place the different images on the screen.
Example setup to evaluate consecutive sections. Sections are placed on the screen according to their anteroposterior position (from top left to bottom right: s144, s145, s150, s151).
Evaluating the series
Evaluating the series
Once all the materials are ready, and images are open, there are three main things to consider for each section:
- Is the name correct according to the serial order?
- Is the image rotated correctly?
- Is the section oriented correctly?
- Are there mounting errors that needs to be corrected?
These questions will all be assessed simultaneously for each image. It is recommended to start with a section where the brain is mounted in one piece, as the evaluation is more complex in parts where different pieces make up a single section. For coronal sections, start approximately at the level of the anterior striatum and move anteriorly towards the bulbs. Once all the rostral sections have been evaluated, go back to the starting section and move posteriorly.
For all of the steps below, our guide to anatomical landmarks might be a useful resource, providing illustrations of major landmarks in the rodent brain.
Is the name correct according to the serial order?
Images should contain the section number in the file name. See our protocol on "Exporting and renaming raw microscopy images" for advice on using a consistent naming convention and numbering sections.
Depending on the measures taken to maintain the serial order of the sections while staining them, it will be more or less likely that errors in the serial order occurs. Make sure to check that sections are in the right order while evaluating them, by using anatomical landmarks to determine relative positions.
Is the image rotated correctly?
Ensure that:
- Images are rotated the same way.
- For horizontal and coronal images, that the midline is straight.
- For sagittal sections, that the ventral base of the brain is straight.
In the example below, the bottom right image is slightly tilted. A rotation of 15 degrees counterclockwise will give a straight midline.
If an image is rotated, enter the degrees of rotation needed (in counterclockwise degrees) in the "Rotation CCW" column of the Nutil transform sheet.
Note
The free software IrfanView (https://www.irfanview.com/) is useful to estimate rotations. Open the image by drag-and-drop, go to Image > Custom/Fine rotation, and try out different angles. Note that the angle in IrfanView is clockwise and that the opposite one will have to be entered in the Nutil transform sheet. For example, for an estimated rotation of -15 in IrfanView, enter 15 in the transform sheet.
Is the section oriented correctly?
During mounting, horizontal and coronal sections may be flipped so that the left and right hemispheres are on different sides across sections. There are at least three ways to figure out whether the sections are oriented correctly.
Method 1: Use a mark that was made to the brain before sectioning.
The easiest way to determine whether sections are oriented correctly is to mark one of the hemispheres of the brain before sectioning. In the example below, a shallow cut was made to the right side of the brain before sectioning, which can be seen across the images. If none of the hemispheres were marked prior to cutting, use one of the other methods.
A shallow cut in the right cortical hemisphere is used to evaluate whether sections have been flipped during mounting. In this example, the two sections on the bottom have the cut on the left side and are therefore flipped.
Method 2: Use anatomical landmarks that appear different on left and right side
If the brain has been cut with an angle across the hemispheres, the different appearance of landmarks on the left and right side can be used to evaluate whether sections are oriented correctly. This evaluation can be very efficient when there is a sizable cutting angle across hemispheres, but requires neuroanatomical knowledge.
Asymmetries across hemispheres to evaluate whether sections have been flipped during mounting. In the top right image, we can see that the posterior end of the olfactory tubercle is larger on the left side, indicating that this side is more anterior (since the tubercle gradually disappears towards posterior). Similarly, in the top right section, the septum is more detached on the right side, indicating this side is more posterior. In contrast, the bottom left section has a more prominent olfactory tubercle on the right side.
Method 3: Use other markers such as damage, blood vessels or shape of the brain
If there is no mark to any of the hemispheres and no noticeable angle across sides, evaluating whether sections are oriented correctly will be challenging. However, it is still possible to use other indications such as tears, folds or other damage, blood vessels or shape features of the different sides.
Blood vessels to evaluate whether sections have been flipped during mounting. The red arrows point to an elongated blood vessel at the base of the brain; the purple arrow to one that extends through the cortical layers into the striatum. Note that the use of blood vessels to determine orientation of sections is difficult and that they can change shape quite rapidly. Thus, they are mostly useful to determine orientation across consecutive images and less useful the longer the gaps between sections.
If the section is flipped around the horizontal axis, write -1 in the "Scale X" column for this section in the Nutil transform sheet. If the section is flipped around the vertical axis (upside-down), write -1 in the "Scale Y" column.
Note
For fluorescent sections, you might have multiple channels for each image. In this event, it is crucial to ensure that the same transform parameters are applied to all channels. See "validation_functions.py" at https://github.com/ingvildeb/brain_section_scripts, which has functions to validate that all channels of an image are treated the same and that there are no duplicate names in the transform sheet.
Are there mounting errors that needs to be corrected?
In parts of the brain where a section consists of more than one piece, there might be mismatches in the mounting of different pieces. During evaluation, write down any mounting in the first empty column of the Nutil transform sheet. See our protocol on "Correcting mounting errors in histological images using Photoshop" for detailed instructions on how to fix such errors.
Example of poor mounting; pieces are in the correct place but mounted too far apart. Depending on the analyses planned on the tissue, this might need to be corrected, e.g. if images are to be spatially registered to atlas and used for automatic analyses.
Running the transformations
Running the transformations
Once the transform sheet is completed, all changes indicated in the "Scale X", "Scale Y" and "Rotations" columns can be applied in batch using Nutil Transform. Detailed instructions on the parameter settings and running of Nutil Transform can be found here: https://nutil.readthedocs.io/en/latest/Transform.html. You can also use our Python code to create the Nutil file automatically (see Step 1), which is particularly useful for large data collections.