Radiology At the Crossroads

Radiology is now at a crossroads. Let us expand the ‏metaphor. If we are to proceed we are to be mindful of ‏the traffic in our direction and be aware of the dangers ‏of crossing the divider, i.e. the median where we have been ‏trained to avoid and therefore not interfere with the flow of ‏commerce conducted by other specialties. What we must ‏also be aware of is what we may not see at first—coming at ‏us from the side streets and seeking the right of way, thereby ‏halting our progress. These side streets are not quiet lanes ‏today but have become major conduits.

Two Challenges Lie Ahead

We refer to the two initiatives, which are going to challenge ‏us now and in the medium term at least. They are artificial ‏intelligence (AI) and genomics. We must recognise the ‏profound interest of these two dynamic innovations (which ‏are most likely to be irresistible) and their strategic implications. ‏And we should also address the tactical changes ‏we must make to maintain a niche in the caregiving enterprise, ‏which involve where we physically situate ourselves and ‏how we regard our specialty vis-à-vis our clinical colleagues.

We explored in a recent talk presented by the first author at ‏the Management in Radiology meeting in Barcelona in 2015 ‏the computerisation of interpretation of head CTs and head ‏MRIs, ie the rendering of diagnosis by application of algorithmic ‏patterns to findings of normality and disease within ‏the crania, including the brain and its coverings (Health- ‏Management.org 2015). The continuing advances in deep ‏learning, i.e. the manifestation of the capabilities of AI, have ‏already been applied to these common procedures placing ‏before us a formidable competitor. We have no doubt that ‏computer-determined diagnosis will supplant us as the diagnostic ‏interpreter in this regard. A recent review of AI in the ‏25 June-1 July edition of the Economist highlights its application ‏to imaging (Economist 2016). It has become a watchword ‏in public discussions of the expected imperatives of AI ‏that the radiologist will be the proverbial “canary in the mine” ‏in this respect, as medical problem solving for the recognition ‏of the presence and severity of disease will become like ‏other forms of intellectual work, a function of the capability ‏of non-human actors.

Similarly, the incorporation of genomics into medical practice ‏will in its own way revolutionise the paradigm of diagnostic ‏investigation. Consider this taxonomic analogy: a disease ‏can be classified by the family it is in, the species it is a part ‏of, and the unique identity it possesses. If a sample of blood can identify the individuality of a disease and recognise its ‏pattern and potentiality for spread and patient survival, then ‏the macroscopic delineation of family and the pathological ‏labelling of species becomes irrelevant. Staging will have a ‏new meaning in a genomic-based rendering of classification ‏and new protocols derived from it will to a considerable ‏extent bypass radiology. Genomics is not as far along as ‏computer-based diagnosis, but both are hot topics and have ‏stimulated heavy investment by many companies with deep ‏pockets, who see in each a technological winner.

Public Aware of Cutting-Edge Science

These threats have reached the section of the public knowledgeable ‏about headlines heralding exciting applications of ‏cutting-edge science. Among this public are senior medical ‏students in the U.S. and abroad. As a residency programme ‏director the first author has been tuned into their thoughts ‏and fears. The spectre of coming technological irrelevance of ‏radiology is disturbing to them as they project 40 years or so ‏of hopefully financially and intellectually satisfying employment. ‏That is the main reason why our specialty has become ‏less attractive to them.

What to do? First of all we must recognise these worrisome ‏prospects. Radiology has had a glorious 40-year run, a ‏happy conjuncture of technological discovery in the service ‏of clinical imperialism. That, we believe, is not a negative ‏phrase, as we have been both the managers and leaders of ‏the incorporation of imaging in all aspects of contemporary ‏practice. However, imperialists tend to be imperious not so ‏much in attitude as in actions and assumptions.

The role of radiologist as it is presumed in many practices ‏in the U.S. will soon become obsolete. The radiology department ‏is distinctly defined spatially in a clinic or a hospital. For ‏the radiology group such a setting is convenient and collegial. ‏There all our machines and offices and reading areas are ‏in a conjoined agglomeration. In the days of film where the ‏viewed image was tangible as an object, such images were ‏at the same time diagnostic information, archival material ‏and in the U.S. a billing record. So our referrers came to see ‏us. Alternators made those consultations more fulfilling than ‏simple view box demonstrations. But with the advent of electronic ‏depictions our clinicians no longer needed to traipse ‏down to see us. So we have now become physically remote ‏and we dare say philosophically distant. Many of us regard ‏our function primarily to make diagnoses from afar rather ‏than to interact continually and in person with our referrers.

That has to change. Image reading venues must be relocated. ‏Paediatric radiologists must be situated in paediatrics, ‏the neuroradiologists should be situated in the neurosurgery- ‏neurology complex, the emergency room (ER ) radiologists ‏must be as close to the ER as possible. The abdominal ‏radiologist and the musculoskeletal radiologists should ‏relocate themselves and their viewing computer as close to ‏the doctors’ lounge or dining room as possible so they can ‏be sure that in-person dialogues can be fostered.

Warning Residents to Avoid Subspecialty

Still, in all these are necessary but not sufficient manoeuvres ‏to staunch the flow of intellectual capital we will forfeit ‏as new and powerful AI and genomic interventions sweep ‏over medical practice. We have warned residents, to whom ‏we have a responsibility as mentors, to avoid neuroradiology ‏and abdominal radiology (the latter at cancer-based hospitals ‏predominantly) as subspecialty choices, because we want them ‏not to pick a job for which the prospects will soon shrink.

Key Points

• Artificial intelligence and genomics are going to challenge us now and in the medium term.
  
• The spectre of coming technological irrelevance of radiology is disturbing to medical students.
  
• Radiologists have now become physically remote and philosophically distant.