Visualizes a heatmap of distribution of nuclear protein c-Fos expression spots in images of mouse brain physical slices acquired by widefield transmission microscopy.

Plugin for ImageJ/Fiji.

As an input the plugin takes a 2-channel 32-bit image data.

In the 1st channel, left - in grey, there is a multi-tile picture of a physical coronal slice of the mouse brain captured by Leica DM6000 widefield microscope in the transmission mode (the picture created and provided by Dr. Helena Janíčková, Laboratory of Neurochemistry, Institute of Physiology of the Czech Academy of Sciences, Prague, Czech Republic). c-Fos expression was visualized with immunohistochemistry and DAB staining. Nuclear expression of c-Fos protein is depicted by bright spots. In the 2nd channel, middle, there is a result of a segmentation procedure with the original bright spots of the 1st channel now labeled by colors. Each segmented (i.e. identified as c-Fos-positive) spot has been labeled uniquely by a different color. On the right there is the input for the plugin - both channels, the original image and labels, together in one composite image.

Protein expression (PE) spots were segmented with the help of StarDist project. A new deep-learning-based model was created and trained with sub-images of PE spots manually annotated by Labkit. Left-a sub-image, right-an example of manual annotation.

When the plugin is applied, then three image windows appear. Left - an original composite image showing all segmented PE spots (i.e. all c-Fos-positive nuclei in the given area); middle - an image showing only the PE spots passing a predefined mean intensity threshold (here, 0-221); right - the resulting heatmap of the predefined size (here, 20x20) showing filtered either absolute numbers or ratios of PE spots related to the numbers in the original image with mean intensities above chosen threshold. In the very right picture I added a LUT used.

The images above were obtained with the plugin dialog setup, please, see below. Each label in the 2nd channel correspond to a single PE spot in the 1st channel. Each PE spot area in the 1st channel has got its mean intensity (MI).

Parameters:

Visualization Method: In the heatmap it is possible to visualize either PE Spot Ratios, i.e. ratio of filtered PE spots with respect to all PE spots in the given grid area, or PE Spot Absolute Numbers, i.e. just a number of PE spots remaining after filtration.

Mean PE Spot Intensity <0, Max>: Threshold - for visualization in the filtered image, middle, we choose only labels that correspond to PE spots with their MI greater than the defined threshold. Max here is the maximum MI of a PE spot found in the open data. The first message below sliders, in black, shows the selected value of MI threshold normalized to Max and expressed in percentage.

Grid Density <1-30>: Defines number and size of individual grids of the heatmap. The size of the heatmap image matrix is the same as the size of the original image. If the selected value is 1, the heatmap contains only 1 real value expressing an absolute value/ratio of labels with MI above the threshold remaining in the filtered image to all labels in the original image. If the value is 2, the heatmap is split to a 2x2 grid areas, original and filtered images as well, and these numbers are computed for all four areas separately, etc. When moving with a cursor in such heatmap area, the corresponding ratio in percentage or absolute number is shown in the status area of the Fiji window.

For each combination of the two parameters described above, Min and Max values in the heatmap are visualized in the red message in the plugin dialog, to indicate which colors in the heatmap, according to LUT applied, correspond to minimum and maximum, respectively.

The blue message diplays plugin status: Computing... or Ready.

The last black message informs that when pressing OK button, a Log file appears that contains data about all labels in the filtered image together with positions of their centroids (X, Y) and their corresponding mean intensity, then the dialog closes. When pressing X, the dialog closes only, without showing the Log file.

Remarks

• The plugin is used for visualization of spatial distribution of "super bright" PE spots (showing strong c-Fos expression) and will be used for comparison of these distributions in normal and treated mouse brains.
• It is more efficient, firstly, to apply the threshold for label removing (Mean PE Spot Intensity) and, secondly, to vary with Grid Density, since computation of multiple heatmap areas is more intensive. For this reason the computation was implemented in parallel threads.
• An example image in ZIP is available for downloading with the plugin file as well.

Installation

Download the Protein_Expression.jar file and put it into the plugins directory of ImageJ/Fiji installation. After restarting the program the plugin appears in Plugins->LMCF-IMG->Protein_Expression menu item.

Sources

Rename Protein_Expression.jar file to Protein_Expression.zip and uncompress to retrieve the source codes.