8 things to know about radiology and AI

When it comes to the future of artificial intelligence in healthcare, it's time to separate the fact from the fiction, says RANZCR president

Artificial intelligence (AI) continues to be a hot topic in healthcare, and this is no more so than in radiology. But sometimes there's more hype in the discussion than heat.


Dr Lance Lawler, president of the Royal Australian and New Zealand College of Radiologists (RANZCR), briefly examines the likely impact and promise of AI. And busts some myths too. All in eight key points...

1. If you aren’t at the table, you become the menu

If you think AI is a distant dot on the healthcare horizon, then think again. As Dr Stefan Harrer, from IBM Research, put it recently: "The future comes earlier these days than it used to."

AI will affect all areas of healthcare, which is why RANZCR has spent the past two years discussing AI with its members and established a dedicated working group to examine its impacts.

2. Beware the hype

Many stories about AI are propagated by technology makers. These manufacturers and developers are adept at using hype to generate interest in their latest products, which are often only incremental improvements on existing products.

While the stories may help them gain funding and influence, the hyperbole tends to distract from the real benefits new technology may bring, which are better tools to aid in faster and more accurate diagnoses.

Repeat after me: We have to concentrate on how we can best use technology in healthcare for those who will benefit from it.

3. Why what’s happening in radiology affects other health professionals

How often do GPs refer patients for a scan? Whether it's an X-ray for a child injured at Saturday morning soccer, an MRI for a middle-aged factory worker with a dodgy knee, or an ultrasound for a mother-to-be, medical imaging plays a critical role in modern healthcare.

How radiologists harness the potential benefits of AI will have flow-on effects for all doctors.

That is why RANZCR has been proactively examining this issue, including hosting the Intelligence18: AI in Healthcare Summit in Sydney in late November, to unite thought leaders on the topic.

4. Assist, NOT replace

Some of the dire predictions about AI would have us believe it will replace humans with malevolent robots. That makes for a great plot in the Terminator movies, but everywhere else the story is a little less apocalyptic.

As Dr Harrer told our Annual Scientific Meeting, which took place in Canberra in October: “AI will always be an assistance system.”

Machine learning won’t replace radiologists — rather, it will allow us to interact more directly with patients and colleagues in other medical disciplines.

AI will augment our skills and capabilities, not render them and us redundant.

5. Embrace the future

Comfort zones are called just that for a reason — they're pleasant places to hang out. But one thing is certain: when it comes to AI, there’s no going back.

While all technology has potential benefits and disbenefits, healthcare professionals must acknowledge the future and adjust their mindset.

Early adaptors of AI will become the thought leaders within their hospitals and practices. Think about that.

6. Avoid the ‘wild west’

Without a robust testing and deployment process to ensure patient safety, the implementation of AI technology may become a ‘wild west’ in medical practice.

RANZCR has argued for a national dialogue in Australia to inform the drafting of robust regulations, implement safeguards for sensitive patient data and ensure the necessary oversight of the development, testing, implementation and monitoring of AI tools.

It’s not about delaying or stubbornly standing in the way of progress, but ensuring any new technology adds to quality, efficiency and safety and improves patient outcomes.

7. Human values won’t be replaced

As machines become more intelligent, they will be allowed increasingly to choose between two (or more) non-ideal clinical outcomes, both of which involve patient preferences and carry an inherent risk of harm.

The choice will sometimes depend on human values such as dignity, respect for others and quality of life. Just because the decision is initiated by a machine won't change that.

The decision-making algorithms will need to be tested for compliance with the ethical code RANZCR is developing to guide members to make these judgements. We see this as essential to overcoming the biases that will be embedded in AI algorithms.

8. Radiologists can lead

Without intending to blow our own horns too loudly, the rise of AI emphasises how radiologists’ voices must be heard when it comes to the integration of technological advances into healthcare.

Radiologists are already adept and experienced at harnessing new technologies to benefit patients — look at MRI, for instance.

Radiologists have a history of innovation and technology application, which qualifies us as pioneers to show how AI can be used for optimal patient care. It’s a role we should relish.