I am trying to analyze some images acquired using a Leica Microscope which has a 12 bit camera. All the images are exported with the Leica software to 16 bit tiff files. When I use cellprofiler to analyze the images, the files are rescaled to 8 bit resolution (based on the min, max and median values returned). To rectify this, I tried to use the RescaleIntensity module, but still the output resolution is 8 bit. How can I solve this?
A few questions:
- In NamesAndTypes, what option are you using for “Set intensity range from”?
- How exactly are you inferring that the rescaling is 8-bit? I.e, what sort of min,max and median values are you getting?
- Can you post an example image file. Our forum is acting up, so you won’t be able to attach to the post directly, but you can use Dropbox or some such and post the link.
Thanks a lot for your response. As for the NamesAndTypes settings, I have already tried to set the intensity range to image metadata, image bit depth and even manual to 4095 with the latest trunk built 2.1.2. Below you can find a link to a dropbox folder with an example pipeline and 2 example images (one FRET and CFP image). The images are actually the first images from a time series FRET measurement. In the output CSV file the max intensity values are actually different from CFP and FRET for the median I only get 3 values? Thanks a lot for your help
I took a look at the files under both FIJI (a specialized distribution of ImageJ) and MATLAB. FIJI allows for image metadata inspection via BioFormats (loci.wisc.edu/software/bio-formats), a library for reading/writing most image formats and is used by CellProfiler. MATLAB has the ‘imfinfo’ command which does a similar task.
All that to say that both of these approaches report that the images are 8 bits per channel. Can you confirm on your end?
I am sorry, by mistake I put 8 bit files in the dropbox folder as example files. However, because of this I discovered that there was a problem exporting to tiff files in the Leica software which makes the cellprofiler software to read them as 8 bit somehow. This is now solved
Ah, good to hear!