Oh my little darlin,
Just the thought of injection in the eye makes a Spinal :Tap seem a breeze, you’re a brave kid.
Please don’t ask Dr. Google; there’s some strange things out there. Stick to A-Z of MS | MS Trust or the MS Society.
Read rizzo’s brilliant sticky A brief beginner's guide to the brain and MRI - New diagnosis and before diagnosis - MS Society UK | Forum
This is an email sent to me some time ago that explains why lesions sometimes do not show:
The last T2 FLAIR scan I did of my brain used 70 slices (on a 3T scanner). The one I had done today - 13 (on a 1.5T scanner). THIRTEEN. THIRTEEN!!!
Thankfully I wrote it out the voxels/T2 reply in Word before posting - to avoid the dreaded time-out! So here it is…
A MRI image typically consists of voxels (3D pixels). Slice thickness is one dimension (on the z-axis if you think of maths). The images you see on the CD show you the other two dimensions (on the x- and y-axes). You can set the voxel size to anything you want, in any dimension; all that happens is that it changes the time the scan takes to run (and therefore, of course, how much it costs). The smallest voxel size used in everyday MRI is typically 1mm x 1mm x 1mm. The “off the shelf” scan that I used to use for this size of voxel had 176 slices. The voxels (and slices) cover the whole brain irrespective of what the voxel size is – nothing is missed out (but see later).
If a standard T2 sequence is used for the scan, white matter gives off a poor signal and shows up as dark whereas lesions (which are full of fluid) give a strong signal and show up as bright.
However, the brightness of a voxel depends on the average of the response from the matter represented by that voxel. So a voxel that is 1mm x 1mm x 4mm will show the signal generated by all matter located in that 4mm3 cube. That is, if the voxel only contains white matter it will be dark in the image, if it only contains fluid it will be bright, but if it contains a mix of white matter and fluid it will look somewhere between dark and bright, depending on the proportion of the different matter types.
So if you have a large voxel (say 4x4x4) and a small lesion (say 1x1x1), the overall signal in the voxel will only be slightly higher than one without a lesion (and therefore look only slightly brighter, and therefore may be overlooked). [NB Small lesions would also not always be completely contained within one large voxel – it is more likely that it would be partially in at least two. So this makes it worse.]
But if you have small voxels and a large lesion, then you will get several very bright voxels (where the matter is all fluid), some intermediate voxels (where there is a mix of fluid and white matter), and some vaguely brighter voxels (that contain predominantly white matter).
In other words, small voxels are much better for detecting lesions.
So, can lesions be missed if you use thick slices? Basically, yes. It is entirely feasible. However, they would have to be much smaller than the slice thickness because if they are closer in size, they would contribute sufficient signal to make the voxels significantly brighter than the surrounding voxels and would (should!) be picked up by a decent radiologist. Saying that, it is possible that it might be missed if a small lesion, by chance, spans lots of voxels (e.g. if it is centred on where four voxels meet on that slice) and the signal is lost by the averaging with the white matter signal in those voxels.
However, there are new “pulse sequences” (the settings that programme the scanner) that are particularly sensitive to fluid. If you use one of these rather than a standard T2 sequence, you will be able to use bigger voxels and still be able to detect lesions relatively easily. And the power of the scanner makes a big difference too. A 3T scanner is much better than a 1.5T scanner.
So, if a hospital has a 1.5T scanner and a neuro is ordering a standard T2 scan, then he should be asking for a high resolution (i.e. small voxel size).
If the hospital has a 3T scanner and the neuro is ordering a FLAIR or another new type of pulse sequence that’s good for fluid, then he can get away with a lower resolution.
[NB A related point: it is possible to set gaps between slices. For example, the MRI may capture signal from 0-4mm, 8-12mm, 16-20mm etc rather than 0-4mm, 4-8mm, 8-12mm etc. This would DEFINITELY miss lesions!]
So the trick to not missing lesions is not so much about the number of slices, but about the voxel size, whether or not the slices cover the whole brain without any gaps, the power of the scanner and the choice of pulse sequence.
Me again, to save all that injection in the eye stuff; this is Optic Neuritis. mult-sclerosis.org
George
