Tuesday 10 April 2012

Using ImageJ to analyse dot blots

I had a request recently from someone who had a time series of dot blots and wanted a quick way of assessing signal from nearly 50 dots over a series of images, each image representing a time point. A pretty laborious job if done manually but something pretty straightforward with ImageJ.

Step 1. Loading individual images into a series, or stack
The images were in tiff format, and each individual time point was represented by a single file. To import these as a series I went to File-Import-Image Sequence... and told the software to import every tiff image containing the word 'Movie'. This imports the images as a series.



Step 2. Identify the dots
I tried a number of different thresholding methods using the IJ plug in AutoThreshold but due to uneven illumination one single method wasn't optimal. To address this, and to save time faffing around with different threshold settings, I made a circular region of interest (ROI) that covered one of the dots then used this to generate all the regions. To do this, draw an oval ROI and position it on the dot, then press Ctrl+T and this will add the location of the ROI to the ROI Manager. Then just go to the next region and repeat. Some of the ROI's were bigger than the dots, this can be adjusted if needed by adjusting the ROI size. I made 49 dot specific ROIs and a 'background' ROI off to the side incase the user wanted to subtract or correct for background.


Step 3. Measuring the dots
I now had a list of 50 dots in my ROI Manager window which I could use to measure the images. To ensure I didn't have a whole bunch of numbers that I didn't need I went to Analyse-Set Measurements and selected Area and Mean Grey Value. I then went to my ROI Manager, selected More>> Multi Measure and ticked both boxes in the window.


This produced a table of results where the first column represented each of the time points in the experiment and the subsequent columns the intensity within each region at each time point. Save this as a *.xl and you can play with it in Excel. Simples!!

I'm pretty sure there will be some experimental issues with the way I approached this but its just an example of how ImageJ could be used beyond its usual area of analysing microscopy data.

Any thoughts..?

No comments:

Post a Comment