By Mathieu Boudreau
This latest MRM Highlights Pick interview is with Sila Dokumaci and David Carmichael, researchers at King’s College London. Their paper entitled “Simultaneous Optimization of MP2RAGE T1-weighted (UNI) and FLuid And White matter Suppression (FLAWS) brain images at 7T using Extended Phase Graph (EPG) Simulations” was chosen as this month’s Highlights pick because they shared code that reproduces their simulations, and did so by building upon another open-source software package (EPG-X).
MRMH: Tell us a little bit about yourselves.
Sila: I completed my undergraduate studies in electrical and electronics engineering at Bilkent University in Ankara (Turkey). I then moved to Switzerland for my graduate studies focusing on MRI. I did a Master’s at ETH Zurich and a PhD at the University of Bern. And after that, I came to King’s College London for a postdoc position and decided to stay. I’ve worked on many different research topics, including MR spectroscopy and MR elastography. I’ve been working with David on our 7T system since January 2020.
David: I did my PhD with Roger Ordidge and Bob Turner at UCL in London, and basically stayed at UCL for a long time after that. I then came across to King’s College in 2018 as a reader which, in our different, slightly odd, and archaic terminology, is not quite a professor! I came here for one reason, namely the nice new 7T Terra they had just installed. I was excited to try it and get my hands on that machine! In particular, I’m interested in the methodological side and also in clinical neuroscience, particularly pediatric epilepsy, and the 7T felt like a good system to use to explore this realm. And that’s what led us to this paper.
MRMH: Before we dive into the paper, could you give us an explanation of FLAWS and extended phase graph simulations, which are both mentioned in the title of your paper?
Sila: FLAWS stands for FLuid And White matter Suppression. It is one type of contrast that can be obtained using the MP2RAGE sequence and different from the conventional UNI image (which, instead, is optimized to provide good contrast between white matter and gray matter, and gray matter and CSF). Extended phase graph simulations are a specialized type of MRI signal simulation that keeps track of echoes formed throughout the repetitions of the sequence.
David: The way I think about EPG simulations is to consider what happens to magnetization when it’s dephased, and you then apply another RF pulse. This situation is quite easy to imagine, for a couple of RF pulses at least. The set of spins is often visualized as a disc of magnetization that rotates then gets flipped, and continues to undergo other manipulations like pinching of the center to form a nice kind of pasta shape or something like that (for those of you with a culinary bent! [laughs]). But try tracking that over 50 pulses; you just can’t. So that’s the basis of the phase graph algorithm — it allows you to track these different bits of the magnetization and how they get rotated/refocused/flipped over a large number of pulses. That’s the beauty of it, and using this methodology you can work out what your magnetization is doing throughout a very complex process.
Sila: Added to that, EPG calculations are conceptually doing the same thing as isochromat simulations. But one method is working in the spatial domain (isochromat simulations), and the other in the Fourier domain (EPG). In EPG, magnetization evolves as a Fourier series that gets more complex after each RF pulse and gradient, and these Fourier components are computed, directly, much more efficiently than the other way.
MRMH: That was very enlightening. Thank you! Having described these fundamental concepts, could you give us a brief overview of the paper?
Sila: As David mentioned, our end goal is to get good images in children with epilepsy at 7 Tesla to help with surgery. MP2RAGE is widely used at 7T for structural imaging because it is insensitive to receive field inhomogeneities, proton density effects, and T2* effects. MP2RAGE typically outputs a uniform image (UNI) that is the result of combining the two images acquired in the sequence, and UNI is optimized to get the best WM-GM and GM-CSF contrast. But other groups have explored different protocol parameters optimized for GM-dominant images, such as FLAWS. In this paper, our goal was to optimize the MP2RAGE protocol so as to get both FLAWS and UNI contrast images simultaneously and efficiently, and we did so by using EPG. We found that the end result only marginally compromises UNI contrast, and we tested it both in children and in adults, obtaining 0.65-millimeter isotropic resolution images that cover the whole brain in under seven and a half minutes.
David: The MP2RAGE UNI images have been around for a while, and at 7T they’re the starting point for most imaging protocols. But it still takes a significant amount of time to acquire these images, and we want to obtain them at higher resolution. Also, for clinical applications, having the range of contrasts you get from the FLAWS image is quite advantageous. Obtaining both UNI and FLAWS images at high resolution is not very feasible in the pediatric population, in which you need to keep the scan time low. What we found, and showed in this paper, is that nice UNI and FLAWS images can be obtained simultaneously in 7-8 minutes without there being too much of a disadvantage, in terms of contrast, compared with what we would have had if we had just optimized the individual images on their own.
MRMH: Your study appears to be particularly driven by the needs of the pediatric epilepsy population. Could you describe some of the challenges when imaging this population?
David: Epilepsy is the most common neurological disorder in children, and unfortunately, in a considerable proportion of cases many of the drug treatment options are quite ineffective. If you have focal epilepsy, your only fully curative treatment option is surgery, but first the focal abnormality that is starting your seizures needs to be identified. In pediatric epilepsy, the most common type of abnormality is focal cortical dysplasia (caused by cortical layer malformations that occurred during the development of the brain in the womb), which has a propensity to generate seizure activity. Focal abnormalities are quite subtle, and challenging to detect on conventional radiological images as you’re looking for regions where the GM-WM border is not quite defined enough. Other expected markers include a slightly unusual gyration and some hyperintensities. FLAWS, however, helps us get better contrast here where the white matter is nulled helping to visualize the WM/GM border. And to image these features in children, in whom the cortex is only about two millimeters thick, you need to have submillimeter resolution.
MRMH: To finish off, what do you enjoy doing when you’re not in the lab?
Sila: I love doing sports – I can get very competitive, but don’t always have the time. I also enjoy bird watching. There’s a very beautiful park next to where we live where you can spot many different birds, like kingfishers.
David: I’ve got quite an interest in football, and today I’m in a pretty good mood because the England women’s team have made it to the World Cup final, which is pretty exciting for us. I have three children, and have managed to get roped into doing some coaching for kids, including my eldest. There are actually a lot of parallels between getting a bunch of 11-year-olds to do what they’re asked on a football pitch and managing team meetings [laughs].