An html-based plate viewer for microscopy images using DeepZoom technology.
A demo web-viewer of a 6 gigapixel image montage created out of 12'300 1024x1024 pixel images.
The montage is an excerpt from a project between Pertz Lab at University of Bern and the Department of Pharmaceutical Sciences at the University of Basel. We aim to find compounds that inhibit cell proliferation, one of the hallmarks of cancer.
The images depict melanoma cells treated with various plant-derived compounds. The cells contain 3 fluorescent biosensors which emit light in their respective wavelengths when illuminated with specific light frequency. One of the biosensors helps to locate cells and perform automatic image segmentation. The other two biosensors sense and measure the activity of two important proteins, ERK and Akt, involved in cell proliferation. If a drug-candidate compound stops cell proliferation, we observe changes in biosensor fluorescence, which we can image and analyse.
The experiment is performed in a 384-well plate format, where each well contains cells subject to a different treatment. A camera mounted on a microscope takes the images (1024x1024 pixels) at 3 different wavelengths (3 channels) in 16 locations of a well (4x4 fields of view), and repeats the process for all 384 wells (24x16 wells).
The demo shows the result of imaging of one of such plates.
A demo web-viewer with only 2x2 wells, 4x4 FOVs per well can be accessed here.
A zip archive with a dataset used to produce that demo can be downloaded from here.
A zip archive with a full demo website can be downloaded from here.
- Python Deep Zoom Tools to generate an image montage and the
dzipyramid file/folder structure with png image tiles. - OpenSeadragon, an open-source, web-based viewer for high-resolution zoomable images, implemented in pure JavaScript, for desktop and mobile.
- OpenSeadragon Filtering plugin with image filters to adjust contrast, brightness, levels in real time.
- Flat Design Icon set for the viewer.
- Python Imaging Library and imageio to read/write images.
- OpenSeadragon CanvasOverlay plugin that adds the canvas overlay ability to OSD images.
- D3.js javascript library that allows data binding to Document Object Model (DOM) and interactive data visualization.
scripts/makePlateMontageDZI.py
A Python script to convert individualTIFFimage files into adzipyramid.HTML-template
HTML template files with a basic single-channel viewer and a multi-channel viewer with real time image adjustments.demosite2x2/openseadragon
OpenSeadragon Java Script viewer ofdzipyramids.demosite2x2/openseadragonsiltering
An OpenSeadragon plugin with image filters.demosite2x2/openseadragoncanvasoveray
An OpenSeadragon plugin for adding canvas overlay to OSD images.jscripts
Javascript files that are used for creating sliders and heatmaps based on csv data that interact with OSD images.
Image files should follow the following naming convention:
A01f00d0.TIFF
A01f00d1.TIFF
...
B12f13d2.TIFF
Where:
A01, A02, ... B12, ...is the well name,f00, f01, ...is the field of view,d01, d02, ...is the channel number.
The parameters -p and -w of the scripts/makePlateMontageDZI.py script prescribe the geometry of the plate and the well, respectively. For example, -p 2 2 -w 4 4 define 2x2 wells and 4x4 FOVs per well. With this definition, the script assumes 64 images per channel. If an image is missing, the script will fill the gap with an empty image with an appropriate label.
The parameter -r of the scripts/makePlateMontageDZI.py script determines the number of threads to be used in creating the deepzoom pyramid, default option is 4.
The script adds grid lines to the composite image; thin lines are added between FOVs, thicker lines are added between wells. In addition, wells are labeled with well names.
To generate dzi image pyramids for both channels in the ../demosite2x2 folder from data in ../demodata2x2, execute:
./makePlateMontageDZI.py -v -p 2 2 -w 4 4 -c 0 -f dzi_c0 -o ../demosite2x2 ../demodata2x2
./makePlateMontageDZI.py -v -p 2 2 -w 4 4 -c 1 -f dzi_c1 -o ../demosite2x2 ../demodata2x2