AI, Deep Learning and the MR technologist– it’s time to map out the realities of our future.

About six months ago I wrote a short article about adapting to survive in MRI. Although it didn’t win the Booker Prize (possibly due to the fact that year 8 English equivalence rarely does) and I stated that I was ‘no oracle’, certain predicted things have already turned out to be true. I would like to think this proves I am psycho-pathetic but the probable answer is ‘ya can’t stop progress’.

Apart from the avalanche of heavily AI saturated media releases, scientific papers and new developments involving deep learning, there have been several advances which incorporate AI in the front end of our work. These developments are tangible and therefore can’t be dismissed as having no immediate affect upon our career. People were recently up in arms on Facebook as one vendor advertised 15-minute appointments for patients as the norm. Computing power is coping with increased data size and things like compressed sense make it possible to scan much more in a condensed time frame. Anatomy recognition is becoming commonplace and touch-and-go protocolling removes more steps from our workflow.

The latest AI algorithm I have seen identifies lung nodules on chest CT and highlights them in real time to alert the radiologist of their presence as they report. Algorithms like this aren’t perfect (yet) but as humans aren’t prone to perfection either, the combination can be powerful and ultimately help patient outcomes (HR-speak alarm). These types of algorithms, and there are an innumerable amount being developed, will start to be added to our MR screen, identifying abnormalities as we scan and either instructing us of the pathology or adjusting our protocols automatically.

Each day I amuse myself by teaching my staff things like where the mammillary bodies are and how they might appear if a patient has Wernicke –Korsakoff syndrome (may be a self fulfilling prophecy). Well, its not always that esoteric, but I think a liberal dose of pathology and associated anatomy can really help a radiographer/tech and make us feel a bit more invested in our job (was going to say empowered but was hit by my anti HR-speak alarm again)

This is where we should pause for thought. Currently I only see articles on how Radiologists shouldn’t be frightened of their worlds being encroached upon, as this technology can only be a giant help. This is fantastic news – for the radiologist.

I am putting a conference together at present with a focus on AI and have some of the biggest names in radiology speaking (RADAIM with Elliot Fishman, Eliot Siegel and Meng Law). The topics seem to be more abundant on the radiologist side of the ledger.

What I want to know is – What is in store for Us?

Our world is governed as all businesses are, by financial balances. It would make little sense for us to not try to improve throughput. Your competitor across the road will do it if you don’t or your administrator will ask for more bang for their buck out of their multimillion dollar investment in your machine. As these machines get simpler to use after the initial setup, and the skills currently required to operate them are no longer needed, what happens to the bulk of us? I can see the high end users setting up the machine and duplicating protocols across the fleet. As we obviously can’t all be the high end user, and we will be overqualified to run the simplified front end of the machine, then our employers will start to look at employing less costly individuals for this task.

The next sentence makes me feel a bit queasy. The further we go into the future, the harder it will be for us to argue the justification of a degree to operate our machinery. I can’t see vendors trying to make things harder to operate and with less importance placed on anatomy, pathology and sequence parameter optimisation from normal everyday MR staff, I can only foresee lesser skilled staff doing our job. You don’t need a formula one driver to steer a taxi. This will not be everywhere of course – research sites and hospitals that do high-end work will definitely need top-tier qualified staff. The rest however – not so much.

As I have banged on about previously, one way to safeguard our positions is by increasing our MR Safety knowledge. Implants are commonplace and magnet and gradient figures steadily rise, as do patient throughput expectations. This knowledge will help us retain our jobs – it only takes one lawsuit to wreck your day and the balance sheet. I have already spoken of the ABMRS exam and the need for global MR safety standards so will leave that there as my opinion has only been bolstered.

This change will not be immediate, but it is happening faster than people think. Now is the time for our professional accrediting bodies to be openly discussing how our profession should be preparing and for our university program coordinators to look at what changes will be needed in their courses. The steady as she goes, ostrich with its head in the sand approach won’t work. Vendor input will be crucial as they are the only ones who know what truly is in store for us.

On that happy note I will cease and desist. Not a time to panic but a great time to plan our future.

p.s. For Australian MR techs who want to add level 2 MR accreditation to their resume but are struggling to reach the points quota, there is good news that ASMIRT have just added 10 points for those of us who pass the ABMRS exam.