Jul 7, 2023 12:20 AM
by Rachel
I have an urgent question regarding RadiAnt’s 3D cursor function. This is crucial for a MRI interpretation of pressing importance, so I’d be grateful for quick and timely clarification of confusion surrounding the issue.
**
Brief background information, so the problem with 3D cursor function can be understood better later on —
* We start out with an inherent problem of images from one sequence not appearing to be synchronised with images from other sequences.
Normally, it should be
【IMAGE 1 from Sequence A】
MATCHES
【IMAGE 1 from Sequence B】
MATCHES
【IMAGE 1 from Sequence C】
In this case it’s something like this:
【IMAGE 1 from Sequence A】
MATCHES
【IMAGE 1 from Sequence B】
DOES NOT MATCH
【IMAGE 1 from Sequence C】*
*The brain structure shown on this image does not match with brain structure shown on the other images (images from other sequences)
However,
【IMAGE 1 from Sequence A】
MATCHES
【IMAGE 1 from Sequence B】
MATCHES
【IMAGE 2 from Sequence C】*
*The brain structure on this image matches the brain structure on the other images above; so this is what’s meant by not being synchronised, in that it’s not IMAGE 1 from Sequence C that matches all the other images, but IMAGE 2 that matches them.
Not sure why, but that’s the situation on hand.
**
So, under such a situation, when we use the 3D cursor to click on【IMAGE 2 from Sequence C】, one would expect that, for any spot that we click on this particular image, that same spot should also be marked out accordingly on【IMAGE 1 from Sequence A】,【IMAGE 1 from Sequence B】, since (as explained in the notes above) these are the images that structurally match.
However, the 3D cursor still pulls out 【* IMAGE 2* from Sequence A】,【*IMAGE 2* from Sequence B】for comparison, and marks out spots on them…
But that’s the problem!!
Because we have just noted that【images from Sequence C】do NOT match images from the other sequences (which means structure-wise, IMAGE 2 from Sequence C does NOT actually match IMAGE 2 from Sequence A, nor IMAGE 2 from Sequence B).
So what the 3D cursor just pulled out are images that don’t even show similar structures, yet it’s trying to look for corresponding spots across these images.
In this case, do we trust our eyes (hmm these images do not seem to match structurally, so it’s wrong to compare across them?) or do we trust the 3D cursor (perhaps these ARE the right images to compare across, even though it doesn’t make sense why images that are supposed to correspond to one another would look so clearly different structurally / positionally).
To find out the answer, I hope to understand how【fundamentally, the 3D cursor has been programmed to function】.
*
i)
【Is the 3D cursor programmed such that, if you click on IMAGE 2 from Sequence C, the 3D cursor can ONLY PULL OUT AND COMPARE for you IMAGE 2 from Sequence A, IMAGE 2 from Sequence B…】
Because in this case, if right from the start, IMAGE 2 from Sequence C is not synchronised with images from other sequences, then 3D cursor may just be comparing across wrong images that don’t match. And any identification made on the basis of that may not be accurate either.
ii) OR is 3D cursor actually programmed to function by searching through ALL images from ALL sequences (a safer method)? Like it’s not limited to search only by IMAGE 1, IMAGE 2, or IMAGE 3… but will search and pick an image that is best matched to the image of focus (out of ALL the images in a set), for comparison.
So maybe for Sequence A, it will search through ALL the available images in this sequence, then pick the one image that matches closest to the focal image. Then it will do the same for Sequence B, C, and later compare across all the images that are picked out?
If so, that means (regardless of whether the images are synchronised between sequences or not), the images that 3D cursor uses to compare across are really the ones that correspond most closely to the image of focus.
However, because the images that 3D has cursor pulled out to compare so far look structurally and positionally different from the image of focus (but truly corresponding images should look similar in structure), I’m wondering if this hypothesis is actually true.
Hence, I have written in for clarification.
**
Summary:
We have started out with an inherent problem of Sequence C images not being synchronised with images from other sequences. Usually 3D cursor is capable of identifying the same spot across corresponding images. However, due to this problem, it now identifies spots across images that structurally do not seem to match - so I’m not sure if these picked-out spots are accurate either.
To further investigate this, I hope to understand how 3D cursor essentially works to identify corresponding images and corresponding spots.
Suppose we focus on IMAGE 2 of a certain sequence. If 3D cursor works by automatically taking IMAGE 2 from all other sequences, put them together to compare, then look for the corresponding spot only from these IMAGE 2 pictures - this could be subject to error. Because when the images are not synchronised, the wrong pictures could be put together, and when we try to identify a spot based on the wrong pictures, we may arrive at the wrong area.
But if 3D cursor works by independently comparing all the images in a sequence to our image of focus, then taking the image it finds most resembling to the focal image, and then repeating this process sequence-by-sequence. In that way, we can get a collection of comparison images that correspond to one another as closely as possible, and unaffected by whether the images themselves are unsynchronised / missing / have any problem within any sequence. And when it then locates a spot across these images, I guess there will be a greater chance that it’s a truly corresponding spot, because the pictures themselves are closely matched (?)
At the moment, because I don’t understand how the 3D cursor works, I don’t know whether the structurally-different images it compares across are pictures that rightly correspond to one another. But it so happens that we need to get the right pictures, and check how some spots appear across these images, to confirm or disconfirm a highly important medical diagnosis. Hence I’m writing in seeking help and explanation on this issue. As my knowledge is shallow in this area, I’d be grateful also for guidance on how to make sure we can get the right corresponding images from all sequences, and locate the right corresponding spot across these images — particularly when images from one sequence may not be synchronised with images from other sequences.
Thank you!!