Analysis

Filters

To start image analysis, first You need to select the filter of interest in the upper right window.

Examples of available filters

Examples of filters include:

  1. Color thresholding (auto/manual) using different color mods.
  2. K-means
  3. Nuclei counting (solution by HistomicsTK)

And others.

The basic principles of starting analysis are quite similar and are demonstrated in the examples below.

Quantification of collagen amount inside annotations (ROI) using color thresholding.

Task: quantify the percentage of collagen fibers (green) in MassonĖ‡s trichrome stained tissue inside the (yellow) annotation.

Task: quantify the amount of collagen (green) inside the annotation

The simple and reproducible way is to apply color thresholding or K-Means. In this example, we will stick with color thresholding as a more flexible tool.

Expand filter “1” and select color_mode, Color.ModeHSV

Selecting filter

Right-click on hsv_range and Add filter.

Adding new subfilter

Double-click on the range, and select the colors of interest (greenish-blue in our example)

HSV color range

As the filter is set, go to the annotation list in the lower right angle, select the annotation to apply the filter, select “filter_id” and enter “1” and press enter.

Applying filter inside the annotation

Voila! The analysis is displayed in the main windows as a percentage of selected HSV dispose inside the annotation. The annotation can be freely moved – analysis will be done automatically inside the annotation.

Greenish collagen fibers are now highlighted as white

A similar result can be achieved by using K-mean clustering (filter_id: 2) as in the example below

Quantification of stromal components using K-means clustering

K-means clustering is a simple yet powerful and difficult to speculate tool for color quantification.

By default, K- means is presented as filter 2.

K-means filter menu

In the list of filter properties, the most frequent to use is the number of clusters (n_clusters), defying the number of colors for segmentation: if set to 3, all the colors inside the annotations will be combined into 3 colors.

Using Masson`s trichrome staining and n_clusters set to 4 (nuclei dark red, cytoplasm red, collagen green, empty field greyish)

K-means: n-clusters set to 4.

The annotation can be freely moved and color analysis performed live.

Hint: for faster results in large areas, in an annotation property set filter_level to 1 or 2.

Quantifying Ki67 using an integrated HistomiscTK solution.

Pathadin has integrated HistomicsTK solution for nuclear quantification, by default presented as filter 3 and can be used in calculating the number of nuclei in ROI. The precise description is available on their homepage.

Ki67 in prostate adenocarcinoma.

In filter properties, set the parameters of the cells of interest. You might need to do several tests before getting proper results.

Example of Nuclei filter setting.

Apply the filter “3” inside annotations. Setting the filter level may also influence the result. Apply filter to get the results.