MRI: Magnetic Resonance Imaging
When you have an MRI scanning session, they run a number of different scans. The standard ones, that are run for pretty much all medical conditions, are called T1 and T2. Scan types are constantly being developed, but the most commonly used additional scan type for MS investigations is FLAIR. You will also hear of PD, FSE, STIR, DWI and others, but I’m not covering them here.
The way MRI works is that different types of matter give off different levels of energy when they are placed in a magnetic field. The computer slices the thing being scanned and collects the energy signal from one slice at a time. Because of some of the loud banging noises the scanner makes, it can also work out the signal from small cubes of each slice. These cubes are called voxels (short for volume pixels).
Each slice provides the information for one image and each voxel provides the information for one pixel in that image. The whole process uses some very advanced mathematics, but ultimately, the higher the overall signal from a voxel, the brighter the pixel is on the final image (and the lower the signal, the darker the pixel). This is where the terminology gray and white matter is less than helpful so bear with me!
In a T1 scan, gray matter gives off a low signal and looks darker in the images than white matter which gives off a higher signal and looks pale gray. CSF meanwhile gives off the lowest signal and looks black. Everything is reversed in a T2 scan: gray matter looks pale gray, white matter looks darker gray and CSF looks white. FLAIR is a clever adjustment of a T2 scan which suppresses the signal from CSF. This means that we end up with gray matter looking pale gray, white matter looking darker gray and CSF looking black.
Different lesions give off different signals too. In general, a white matter MS lesion gives off a high signal in T2 and FLAIR scans, leading to it looking like a “white spot” against the white matter which looks relatively dark in these scan types. T1 is usually rubbish for spotting MS lesions unless the lesion has caused the area to die, or “atrophy”, in which case it is called a “black hole” – because it is a wee black hole on a T1 image. The terms “hyperintensity”, “hyperintense”, “high signal”, etc, all refer to the fact that somewhere is brighter / whiter than it should be. “Hypointensity”, “hypointense area”, etc, refers to somewhere that is darker than it should be. Exactly what relevance these have depends on their size, shape, location and the type of scan though.
Remember that I mentioned CSF bathes the brain? That means that all the sulci of the brain are full of CSF and so there is a lot of white on T2 images. Spotting a lesion in amongst lots of perfectly normal white stuff can be tricky, however it is really quite easy in FLAIR images – because there shouldn’t be very much white; all the CSF is black. (Please note that some small white spots can be perfectly normal – they are usually blood vessels.)
Sometimes, neuros ask for a scan with contrast. This is a T1 scan taken after the patient has been injected with a “contrast agent”. This is usually gadolinium which looks bright white on a T1 scan. The central nervous system, i.e. the brain and spinal cord, is protected by the blood brain barrier (bbb) which stops things that might harm it from getting in; gadolinium normally can’t get through the bbb.
In MS, cells from the immune system get through the bbb and attack the myelin coating of nerves in that area, causing inflammation and damage: a lesion. While this is happening, the lesion is called “active”, “enhancing” or “contrast enhancing”. The gap the immune system has caused in the bbb allows gadolinium to get in. If there are no breaches in the bbb, there should be no bright white signs of gadolinium inside the brain or spinal cord. If there are, these show where there are breaches, in other words, where the immune system is actively causing new damage.
Contrast is used for two main reasons: to show up very new lesions (typically lesions newer than about six weeks) because these can be difficult to see on normal MRI and to help show which lesions are active and which are not as this can be important for deciding on meds. Newer types of scans, e.g. DWI (diffusion weighted imaging) which measures water flow, can be used to do this too so contrast is being used less often these days.
Some terms you might come across:
- Sagittal: images taken from ear to ear, parallel to the nose
- Axial: images taken from front to back, at right angles to the nose
- Coronal: images taken from the front to back, parallel to the nose and so that one slice goes through both ears
- Artefact/artifact: a computer error, nothing to worry about
- 1.5T (and other numbers followed by a T): This is scanner strength, i.e. how strong the magnetic field is that the scanner produces. The T is short for Tesla, but has nothing to do with the T in T1 and T2 scans.
Most NHS scanners are 1.5T. There are a few 3T (and stronger) scanners that are much more powerful than 1.5T scanners, but these are usually run by Universities. Do not assume that private MRI scanners (or scans) are better than NHS scanners (or scans) – they are mostly the same, but can be worse.
For those of you who aren’t quite brain dead yet(!)...
“Partial volume effects” are an important factor in how good a scan is and are particularly relevant if you are told your MRI is clear or you have no new lesions when you are having new symptoms. Remember that the computer slices up whatever’s being scanned and then splits the slices into little cubes, the signal from these determining how dark or light a pixel will be in the image? Well, the size of those cubes has a major effect. Imagine a voxel that only contains white matter.
In a T2 scan, that voxel will give a low signal and the resultant pixel will look dark gray. Now imagine a voxel that is mostly white matter, but also has a bit of a lesion in it. In a T2 scan, lesions give high signals, so the pixel will now look brighter than the last one. But by how much? If it’s a small voxel and the lesion provides a decent proportion of the signal, the pixel will be obviously brighter than its neighbour and a radiologist should spot it easily.
However, if it’s a big voxel, the extra signal from the lesion might only make the pixel a wee bit paler than its neighbour and make it very easy to miss. So, the moral of the story: it’s important to have thin slices and small voxels! 3mm slices are fine. Less than that is great, more than that and you are losing a lot of definition and increasing the chances of missing small lesions.
By the time you get to 6mm and thicker, you could miss average and even bigger than average lesions too. (The average MS lesion is 7mm.) The slice thickness is not usually written in reports or on images, but the number of images is the same as the number of slices so the greater the number of images in one scan, the thinner the slices.
The best images generally come from thin slices with small voxels on a stronger scanner, but small voxels on a 1.5T scanner will be better than big voxels on a 3T scanner every time.