Split channel DICOM

[joe90]

Hi

I am a beginner with this sort of thing and having some confusion with a compressed DICOM series I have imported into ImageJ using Bioformats, was wondering if anyone could help? Sorry if this is the wrong place for this sort of question.

It's a file from a diagnostic ultrasound scanner (Zonare scan engine with 8MHz transducer), consisting of a simple 2D slice through a test chamber repeated at 41Hz. I want to create a time-intensity curve of the mean intensity in an ROI I will create. There is no harmonic imaging or anything like that, so I was expecting a single channel of grayscale with no colour data. But when I import the file it is split into three channels of RGB. This happens whether I select default, composite or grayscale in the colour options. When I carry out the "Plot Z-axis profile" it analyses 3606 frames instead of 1202. The channels do appear to be identical upon visual inspection, but I don't see any repetition in the curve or the intensity data points so I think they must be slightly different. How can this be the case in a simple ultrasound scan where there should just be one variable being measured - intensity of backscattered ultrasound?

Basically I want to know is can I just merge the channels and will my data still be correct, or should I just be generating a 3606-frame time-intensity curve even though there are only really 1202 frames, or is there a way I can create a curve from just one channel and will that data be correct?

Confused.

[mlinkert]

I would guess that all 3 channels are identical; depending upon which compression type was used, single channel data is sometimes stored into 3 channels to satisfy the requirements of the compressor. If that is the case, then analyzing a single one of the channels should be correct. You can open just one channel by selecting the "Specify range for each series" option in the "Bio-Formats Import Options" window, then setting the C start, end and step to 1.

However, without seeing the file itself I really can't say for sure. If you are willing to upload a file to:

http://qa.openmicroscopy.org.uk/qa/upload/

then we can double-check that this is not a bug in Bio-Formats.

-Melissa

[joe90]

Thanks Melissa,

There is nothing confidential involved so I have uploaded the file as requested. It is a file called US00000L.dcm.

Opening one channel sounds as if it should work, as I said before the channels do look identical visually. I was just concerned because I don't see any duplication in the mean intensity values. I would have expected to see three identical values for each time step, since there are 3606 data points for 1202 time steps, but each data point appears to be different. But maybe this is not how the Z-axis profile works.

Thanks again

Joe

[mlinkert]

Thank you for uploading a file.

There really are 3 channels stored in the file; you can see this by looking ("Image > Show Info") at the metadata values for "Samples per pixel" and "Lossy Image Compression Ratio". This is a result of the compression method used, i.e. lossy JPEG.

The channels are very similar, but not quite identical - various pixels in each of the channels are off by one with respect to the other channels in the same frame. I would think that you would be safe enough using any single one of the channels, but note that due to the fact that lossy compression was used, the analysis results likely will not be quantitatively meaningful no matter what.

[joe90]

OK, thanks for your advice. Hopefully I will still be able to glean some useful information on timing etc if not the actual numerical values.