A.I. in Veterinary Radiology and Diagnostic Imaging. Not Always Black and White.

“Only a Sith deals in absolutes.” Spoken by Obi-Wan Kenobi to Anakin Skywalker in the 2005 film Star Wars: Episode III – Revenge of the Sith

This is a brilliant line that is sometimes said to be ironic because the statement in itself is an absolute. However, the context of this line reveals its true meaning. Anakin Skywalker stated that if you are not with me, then you are my enemy. 

When it comes to utilizing artificial intelligence technology inside of our work in diagnostic imaging, I do not believe that it is a black and white issue. We always speak about “shades of gray" in diagnostic imaging giving us more (or less) insight into the well being of our animal patients. I believe that it is the same with utilizing A.I. in our line of work. I know veterinary radiologists who are committed to never using A.I. in any form. I also know a few who are trying to make a fortune from A.I. They wish to make it dominant in veterinary diagnostic imaging interpretations. I, like many others, have a more gray approach. I believe that A.I. has a place in veterinary medicine and also in diagnostic imaging. 

Writing (or dictating) interpretation reports in diagnostic imaging has evolved immensely since I started my career over 20 years ago. In the beginning there was just typing reports on a computer word processor which evolved into reports transcribed from the interpretation recordings by a medical transcriptionist. Then came dictation software like Dragon® Naturally Speaking and Dragon® Medical Edition. This technology changed everything for radiologists. Medical transcriptionists were no longer needed. Radiologist’s interpretation reports could be done in a fraction of the time. Turn-around time shrunk significantly. Typos did occur regularly. Nowadays dictation software (aka voice to text) has become commoditized and we still have those pesky misspellings, and other typos like inaccurate words or even accidental conversational phrases inserted into reports from time to time. Those happen when the radiologist is having a conversation while dictation is still activated. It’s even more hilarious when there are curse words. I am okay utilizing an A.I. assistant to clean up those mistakes. A radiologist will sometimes say "left" instead of "right", or "dorsal" instead "ventral". Again, I am open to utilizing an A.I. to catch these little confusing contextual errors as well. This is being done already. 

I will not participate in the use of A.I. technology in the interpretation of radiographs, ultrasound, and CT scans from animal patients. I think this way because I hold the belief that human beings are better equipped and more flexible than A.I. when it comes to interpreting multiple different modalities, diagnosing pathology in multiple different anatomies, in multiple different species and multiple different shapes and sizes within each species. A.I. cannot perform an ultrasound scan of an animal patient and has zero context for the challenges and issues that can occur during those patient scans (gas, panting, and so on). There is no technology that will ever be able to match the wisdom of experience that is gained by humans in several years of veterinary medical school, internships, residency, and years of practice experience. I do see where A.I. can assist the specialists, but A.I. will never be good enough to replace them. 

I predict that this article will be met with prejudice and disagreement from those who are proponents of replacing humans with A.I. Those who believe this way are welcome to have and share those opinions. I believe that these good people will continue to find themselves in the minority. 

When it comes to artificial intelligence in radiology, it is not black and white. There are many shades of gray. We cannot (nor should we) eliminate A.I. completely from our space, that would be silly. We cannot wholly depend on it either. 

-Robert