Cluster Analysis of Nuclei Tool

Analyze the clustering behavior of nuclei in DAPI stained images. The nuclei are detected as the maxima in the image. Using a threshold intensity value, maxima below the threshold are eliminated. The resulting points are clustered using the DBSCAN algorithm. The nearest neighbor distances between all nuclei, and those outside and inside of the clusters are calculated. You can download an example data-set:

The source code in git-hub can be found here.

Getting started

To install the tool save the three files cluster_analysis_of_nuclei_tool.ijm, dbscan_clustering.py and find_nearest_neighbors.py into the folder macros/toolsets of your FIJI installation.

Select the "Cluster Analysis of Nuclei Tool" toolset from the >> button of the ImageJ launcher.

• the first button (the one with the image) opens this help page
• the c button runs the cluster analysis on the current image
• the n button creates a selection that shows the nearest neighbor relations between the points of a point roi
• the sa button can be used to select all nuclei after a cluster analysis has been run using the c-button.
• the su button can be used to select the unclustered nuclei after a cluster analysis has been run using the c-button.
• the sc button can be used to select the clustered nuclei after a cluster analysis has been run using the c-button.
• the r button runs the batch cluster analysis on a folder containing the input images.

Usage and method

The nuclei are detected as maxima in the image using the "Find Maxima..." command of ImageJ. Only maxima with an intensity above a threshold value are taken into account. Set the detection parameters by right-clicking on the 'c'-button (see below). If you select the 'no detection'-option, the nuclei detection will be skipped and the values from the ImageJ-results table are used. In this case the table must at least contain the columns X and Y. If the table also contains a column Mean the above mentioned threshold is applied, otherwise not.

Press the 'c'-button to run the clustering. For the clustering the DBSCANClusterer from apache is used.

After the clustering, the nearest neighbor relations of the clustered nuclei are shown. To display and measure individually the nearest neighbor distances between the unclustered or all nuclei, press first the 'su' or 'sa'-button to select them and then the 'n'-button. With the 'sc"-button you can display the nearest neighbor relations between the clustered nuclei again.

Options

Options of the cluster analysis

• sigma of gaussian filter - Sigma of the Gaussian filter applied before the maxima are detected.
• noise - Noise parameter of the "find maxima" tool.
• no detection - Do not detect the nuclei, use the data (X, Y, Mean) from the Results-table instead.
• threshold - The intensity threshold above which the maxima are taken into account.
• max. dist. - The maximum radius for the DBSCAN clustering algorithm, see DBSCANClusterer.html
• min. pts. - The number of points up to which it is not yet a cluster, i.e. clusters must minimally have min. pts. + 1 nuclei.

• output folder - The base of the name of the output folder. It will be created in the input folder. A timestamp will be added to the name.
• channel - The name of the channel to be processed. It must be a part of the filenames of all input-images.
• save control images - If checked, control images with the objects in an overlay will be saved in the output folder.
• save control snapshots - If checked, RGB-control images will saved in the output folder.

Results

| | |

Publications using the tool

1. Abend, A., Steele, C., Schmidt, S., Frank, R., Jahnke, H.-G., and Zink, M. (2022). Neuronal and glial cell co-culture organization and impedance spectroscopy on nanocolumnar TiN films for lab-on-a-chip devices. Biomater. Sci. 10, 5719–5730. 10.1039/D2BM01066F.