Don't know why I didn't think of asking here before so here goes!
We have a GE PACS and MRI scanners from Philips (Intera) and GE. Philips are the only manufacturer to scan sagittal images from left to right with no option to change.All other manufacturers default to scanning from right to left.
Philips are unable to offer any software to correct this. GE can offer a PACS software patch to display images from right to left however it means that our GE MRI has to be set to scan left to right to conform to Philips which in turn restricts the hard copy production to silly little images on a large sheet of film.
We are also going to run into issues when scans from the Alliance mobile (Independent sector activity courtesy of Mr Blair) are downloaded into the PACS unless we can persuade them to scan left to right. Thus far Alliance have not been receptive to any changes from their contract with Whitehall so I am not hopeful.
We seem to be heading for a lifetime of asking non-Philips manufacturers to scan left to right as the PACS patch is not machine specific. In other words if in the future we jettison Philips (highly likely at this rate!) because of displaying the 'old' Philips acquired images we will have to continue with the current display protocols
This was never a problem with hard copy as the laser imager gave the option to print the in the correct order regardless of order acquired. Any inconsistency in soft copy display however poses an unacceptable clinical risk - we roll out to orthopaedic out-patient this month!
What surprises me is that the Intera is a very popular MRI and that no one has apparently come up against this problem before - or so we are lead to believe.
If anyone has an answer to this we would be exceedingly grateful!
Simon Elliott Consultant Radiologist Derby Hospitals
posted on Wednesday, May 04, 2005 - 09:55 pm
This question all hangs on what is meant by "scan sagittal images from left to right with no option to change". I'll try not to bore everyone with technical detail, but basically, CT and MR images are currently exported from scanners to the PACS is INDIVIDUAL images (this might change in a few years time, as "multi-frame" CT/MR has been standardised, but it is not yet in general use). So, we have to consider the factors which might cause your DISPLAY system to put the images into any particular order - the likely candidates being:
1) The order they are sent to the PACS. Due to the way PACS& DICOM work, this is extremely unlikely to have any effect (though is in practice probably the same as (2) below)
2) The "Image number" field in each image - normally a small incrementing integer, unique within a series
3) The "Position" information, which is a set of 3 numbers representing the location in 3D space of the top left corner of the image (measured in mm from an arbitrary position which is fixed for at least one series)
Now, a proper display system should give you the choice of how you wish to sort/display/print the images within a series, and a well-designed system should allow you to use (3), the true 3D position for the sort - have you tried this option? If you have, and the order is still "wrong" (seen by you as being different direction for different scanners), then this would indicate a SERIOUS defect in the scanner which would need reporting to the manufacturer and MHRA.
However what I suspect is happening is that the "Image number" is being used instead, and it is perfectly permissible for this to be in any order the scanner manufacturer likes - there is no standard for this.
So, the "standards" answer is that the fault is likely to be in your display system, sorting by image number, which is an inappropriate "proxy" for position.
As regards a solution - try looking at the sorting options in your display - there may be an option you have not noticed. If however you have a broken display system which CANNOT sort properly on position, then you might wish to consider a "black-box" to modify images on their way from the scanner to the PACS - perhaps setting image number to 1000-[original imagenumber]......a hack, but not a difficult or dangerous one!
posted on Thursday, May 05, 2005 - 09:32 am
Dave is right, GE should be able to display the images as per their Image Instance UID and give you the desired results. At the moment they are possibly displaying by the Study Instance UID hence the problems. Ask your FE to play about with this.
Failing that completely you could try and identify the scans done on the phillips unit in some other fashion i.e. a unique order code or populate another RIS value(left or right for example). Basicaly anything that will stand out from other codes and allow Centricity to identify it. Not the best way to do things but a fix none the less.
We've just had a Philips Intera Acheiva installed linked to an Agfa PACS. The MR runs SW version R1.2-1, in the installation manual section 6.3.1 DICOM image numbering direction you can get your engineers to specify and store the image number direction for Transverse, Sagittal and Coronal scans.
This might help, or point the Philips engineers in right direction.
The only inaccuracy is that in fairness to Alliance I haven't asked them to scan in reverse order as I think this isn't really the answer here. As the national program rolls out we need the ability to import images from any scanner in any centre.
Anyhow here is GEs comment:
Thank you for this. I have read Dave Harvey's response with interest. I have spent quite some time this morning talking through this problem with various people within GE.
To deal with Dave Harvey's email first:
"Now, a proper display system should give you the choice of how you wish to sort/display/print the images within a series, and a well-designed system should allow you to use (3), the true 3D position for the sort - have you tried this option? If you have, and the order is still "wrong" (seen by you as being different direction for different scanners), then this would indicate a SERIOUS defect in the scanner which would need reporting to the manufacturer and MHRA."
GE PACS does have this sort/display/ functionality
"However what I suspect is happening is that the "Image number" is being used instead, and it is perfectly permissible for this to be in any order the scanner manufacturer likes - there is no standard for this."
GE PACS does not use the image number, it uses the table location data
It is a complex situation with a considerable number of variables which are all interrelated. So that we are all clear where we are at the moment I have tried to summarise below, the issues and possible ways forward.
Please could all addresses let me know if anything below is incorrect.
Philips MR - can only acquire L to R. GE MR normally acquires R to L but has been forced to acquire L to R so that it matches the Philips MR.
Therefore to date all MR images on PACS have been acquired L to R.
The Alliance Medical Scanner acquires R to L and they are refusing to alter it. So far no Alliance images have been sent to PACS.
A GE fix for the workstation (from a site which has a Philips MR and a GE PACS), using platinum properties to reverse slice order for body parts, means that although the GE and Philips MRs are still acquiring L to R they are displayed on the PACS Workstation R to L. This DOES NOT fix the issue on the PACS Web.
1. Philips MR - It can only acquire L to R.
2. Alliance MR - They will only allow it to acquire R to L.
3. Printing on GE MR - When GE MR is configured L to R the printing is sub-optimal.
4. Web - even with the GE Fix for the Workstation the Web displays them the opposite way round.
5. Alliance MR with GE fix - With the GE fix to correct the Philips/GE MR L to R images, it will flip the Alliance images acquired R to L around to L to R therefore the Alliance images would still be display differently from the Philips/GE Images.
6. Legacy Images - All MR images currently on PACS have been acquired L to R.
From my discussions today it seems that the fix using Platinum Properties on the Workstation cannot be replicated on the Web. Therefore the current GE Fix (Platinum Properties), displays GE and Philips MR correctly on PACS Workstation but there still is the Alliance MR Issue, Printing Issue, Web Issue, Legacy Images Issue.
The only option that would solve the majority of the issues would be to place some type of "black box" between the Philips MR and PACS so that the Images are sent to PACS R to L, and the GE MR is set back to acquiring R to L in line with the Alliance MR. This would solve all the issues other than that of all the legacy data.
Dave Harvey has a company called Medical Connections which might be able to help, there are other companies which do similar things.
I am sorry I am unable to come up will any better news.
posted on Thursday, May 05, 2005 - 08:21 pm
Simon, thanks for posting the GE reply, which seems entirely fair and consistent. In particular, I was relieved to see this line:
> [.. a well-designed system should allow you to use > the true 3D position for the sort...] > > GE PACS does have this sort/display/ functionality ... > GE PACS does not use the image number, it uses the table location data
This is what I would have hoped and expected to see. However, it does not then fit with the discussion below:
> 1. Philips MR - It can only acquire L to R.
> 2. Alliance MR - They will only allow it to acquire R to L.
It is commented that this acquision order affects the display, but this "ordering" would only be reflected in the "image number" (or perhaps other related variables such as the UID[which is never a proper bais to sort!]), which it says above are NOT used !!
So, if the GE PACS is using only the location data, and the location data is correct, then "acquision order" should be irrelevent !
We are therefore left with the 2 original ideas I mentioned last time, albeit with more data to clarify the issues. EITHER
1) One or other of the scanners is sending incorrect location data
2) The GE PACS is not using that location data to sort the images.
Thanks to Richard for the commercial "Plug", but I'm not trying to sell anything here, I'd far prefer to see this sorted using existing kit!
Part of the issue which complicates matters is that even if the PACS could display images by table position, if we have 2 scanners scanning in opposite directions, the images are displayed numerically opposite. ie GE right most image no.1, Philips left most image no.1
Clearly this is very confusing to clinicans when images go out on the web. Far better to be able to say image number 1 is always right most whichever machine the patient has been scanned on.
I'll keep in touch with any developments - hopefully it will get sorted!