We have a Philips Intera 1.5T MTRI scanner. Dynamic contrast enhanced studies do not display properly on our PACS (Ferrania Lifeweb): the different phases are merged as a single series, rather than being displayed as separate series. The same thing happens when the study is viewed on Vitrea workstations (Vitrea 2 v3.7.02). Interestingly, the studies could be read properly with earlier verions of the Vitrea (v3.1)
I have been told that this is due to changing DICOM standards (??!!??)
The solution proposed by Philips is an upgrade costing about 150 000 GBP. I know that they have a similar problem at Newcastle.
Does anyone know of a more realistic solution?
posted on Thursday, July 28, 2005 - 12:57 am
There has been no change in the DICOM standard for existing MRI objects, so this is a red herring !
DICOM does not in fact say what the relationship should be between series, so it is likely that both series organisations are strictly speaking legal under DICOM rules (for those interested in the technical reasons, fields such as TR, TE etc. are "image level" attributes, so there is no rule to say that they must be constant within a series)
There IS a new "multi-frame" MR object, but I doubt whether this is the problem....if it IS the problem (i.e. if this new object is being used for image export), then the answer would be to blame the viewer, as the new format has all the information necessary for a viewer make sensible decisions and give sensible options for image display in mutliple different ways. Even without the mutli-frame object, this is still really (like the image ordering issue a few months ago) a viewer issue - viewers should be able to sort on more fields than just series and image number!
Just for information - our GE PACS does the same thing with dynamic contrast enhanced images received from our Philips 1.5T MR machine: obviously a common problem: Dave what should we ask the companies to do about this if you think it is a viewer problem?
posted on Thursday, July 28, 2005 - 09:31 am
Your PACS should be able to display the images by the series UID, This will allow you to view them by series rather than them all grouping together.
Nicola, you can try this.
On your workstation, go to utilities, then preferences, then display preferences, then image grouping, and select "DICOM SERIES" for MR.
That might do the trick.
Failing that speak to your vendors and tell them the images should be displayed by the series UID.
posted on Thursday, July 28, 2005 - 11:32 am
Nicola & James,
The answer to all these problems is for viewing equipment vendors to accept that they cannot rely on source equipment organising the images in any particular way as regards series or image number within these series - DICOM does not specifiy how this must happen, and therefore it is the viewer vendors responsibility to cope with all legal variants.
James, I may be wrong, but I suspect that sorting by SeriesUID is already happening - it is bundling of different echo images into one series (quite legally) which is the problem.
So, what is needed is for viewing software to offer a wide range of sorting options (in fact, they could easily be user defined in a good system - you say which DICOM tag(s) you want, and give this a name - it could then become a menu option. In this particular case, we would like the viewing software to sort using the following tags in decreasing order of precedence:
StudyUID Acquision time SeriesUID TR TE Acquision time Image Position
Some people may wish to use a different order, but the important point is that the flexibility to define your OWN should be present, and in general the image number, which is commonly used within series is the LEAST useful, as it is so implementation dependent.
posted on Thursday, July 28, 2005 - 12:09 pm
"it is bundling of different echo images into one series (quite legally) which is the problem"
In my experience, which does not include Phillips (sorry) This is something the vendor can prevent by having the modality assigning a separate series UID to each acquisition or run of images. All our modalities are set to send images which allow our PACS to display them as Dave states, Study UID - Series UID - Image UID.
Sometimes this takes a bit of doing, the reason being that to your average modality engineer DICOM is rather new and not something they have ever had to worry about in the past. The result being they don't know how to set up the modalities properly as per above. Get them to escalate this within their company and to discuss this with other engineers who have been using that modality with PACS.
Disclaimer - Yes I am painting with a rather wide brush and I mean no offence to modality engineers, but like everything in life some are better than others. The best ones are those who admit their limitations and look for an answer rather than saying, "It can't be done" or "Itís a PACS thing"
We had a demo from "Dynamic Imaging" today. Their system is able to sort out the Dynamic CE MR images of the abdomen from the same series by slice location, time of imaging etc. in two mouse clicks. So we know that it is possible. Then it just behoves all the other vendors to find a solution. One just has to push harder.
I have asked Philips for a written explanation of the situation, and will ask Ferrania and Vital Images for the same, which I will post here. Perhaps Nicola might like to do the same with GE and Philips.
I have told the vendors about this thread in the hope that this might help to facilitate a solution to this extremely tedious problem.