What is the future of MRI technology in industry?

Hi everyone!
I am a new PhD student about to enter the field of MRI. After doing extensive literature review and information collection for a few months, I found myself starting to doubt the future of MRI, especially for industry. I feel very sad for this and I know this is absolutely not a good thing for a new PhD student, so I came to this website and would really appreciate anyone that could correct my thoughts.
Here are my observations:
(1) For the acquisition and sequence development part, the sequences that are routinely used in hospitals have been there for decades. New contrasts are being created in acdemia, but it seems that few of them found their roles or key applications in diagnositics. A T1 or T2 weighted image can already fullfill the needs of radiologists in most cases.
(2) For the reconstruction part, it goes without saying that the advent of ML/DL and compressed sensing greatly improved the reconstruction speed and image qualities. However, this may be the only major development of MRI reconstruction in the past few decades.
As a student who is keen to joining the industry upon graduation, I now really fear that there would be no job opportunities for me, as MRI has become a very mature technology and R&D may no longer be needed by manufacuters. How do you think of this? These are all my humble opinions as a newbie student so any opinion would be much appreciated…

1 Like

Hi there.

I would dispute basically all of the points you’ve made. I am not going to outline all of the new developments in MRI sequence and reconstruction technology, but you can find whitepapers from vendors on clinically-available sequences being released every year, including compressed sensing, parallel imaging, MR fingerprinting, elastography, ultra-short/zero echo time imaging, etc. This is not even including the novel postprocessing research and advanced multishell diffusion tensor imaging, AI segmentation and lesion detection/diagnostics, noise-reduction analysis, etc. Look at the recent history of lung MRI for respiratory diseases (especially cystic fibrosis) for an example of how things such as UTE/ZTE have changed in the last 5 years.

You mention that T1 and T2 images make up the bulk of diagnostic imaging. That is always going to be the case, because T1 and T2 are fundamental contrasts by which MR imaging works. This is like saying that CT won’t have any new developments because diagnostic images just measure radiation absorption. Computers still use transistors and semiconductors that have been around since 1948, but I wouldn’t say computing research is “dead.”


Couldn’t agree more with @Tim_Rosenow! To add a few examples to the list of exciting new MRI areas to be explored, PDFF is playing an increasing role in NAFLD / NASH clinical trials, MR Fingerprinting and Synthetic MR are a major area of development, Hyperpolarized MRI shows a lot of potential, and there is clearly much work to be done on the increasing number of 7T scanners. That’s not to mention the worlds first 11.7T install - hopefully we’ll see in vivo images soon!

1 Like

Dear all, I think all of you are right. There is no longer much to do in the field of MRI if we take into account the theory of the method. There are more parameters than T1 and T2 which create contrast in MR images, and this is the great advantage of MRI over other diagnostic imaging methods, although it’s also rather correct that these two are sufficient to provide us with enough information for being able to see most, but not all of pathologies. Basis of synthetic MR have been known since 90’s of the XX century. MR Fingerprinting, as matter of fact, is nothing else but speeded-up synthetic MR. Thus, the MRI theory is well known and established, but there is still lot of to do in the development of MRI hardware so that we can acquire images as fast as possible, with as high SNR and CNR as possible. However, please keep in your minds, that such achievements are limited by MRI safety issues.

Category icons made by Freepik from www.flaticon.com