Health NZ explores use of AI in radiology

NEWS - eHealthNews.nz editor Rebecca McBeth 

Health New Zealand | Te Whatu Ora is exploring the use of artificial intelligence (AI) in radiology to improve patient safety, streamline workflows, and improve access to care.

Radiologists across the country are working with Health NZ’s AI Lab to develop the evidence base to support clinical rollout of AI. They say AI will not replace any radiologists, but can save them time and enhance consistency across the health system.

Health NZ’s AI Lab is particularly interested in investigating AI applications for high-volume examinations such as chest X-rays, fracture detection, and CT brain scans, which could benefit the largest number of patients.

Stuart Barnard, clinical director of radiology at Health NZ Counties Manukau, says that AI is primarily being explored to improve patient safety.

“These tools are about improving quality, not replacing expertise,” he says.

“The initial focus has been on image interpretation: it should reduce the error rate and help us triage the scans for reporting.”

Barnard says computer aided detection tools already in use in New Zealand help to identify possible abnormalities on CT images and advances in CT image reconstruction software have led to a dramatic reduction in the radiation dose delivered from CT scans. 

AI has been used to develop MRI reconstruction algorithms which reduce patient time in the scanner while maintaining diagnostic image quality.

“It also improves efficiency of the of the use of the machines, which then helps us to scan more patients and decreases the waiting time,” he says.

Sharyn MacDonald, chief of radiology at Health NZ Waitaha Canterbury, says AI can help in the triaging of unreported images.

“There are AI solutions that could adjust the triage flags on our work list to help identify which patients need to be reported with priority,” she tells eHealthNews.

MacDonald says studies show AI is very consistent at detecting abnormalities and - acting as a co-pilot to radiologists - could reduce error rates in detection of things like lung cancer.

“These solutions never get tired of looking at X-rays or CT scans. They can improve safety, and in turn help us as radiologists to improve quality of the service that we provide, and enhance patient flow,” she says.

While Canterbury already uses some AI tools to look for one particular abnormality on some scans, the new generation of tools can look for over 100 different findings.

“We are very keen to understand how those solutions could be integrated into our workflow and enhance care,” she explains.

“Some are in use in the private sector already in New Zealand, including to provide care to Health New Zealand patients, so we have the opportunity to potentially learn from the experience of our private provider colleagues.”

Clinical director of Health NZ AI Lab Cheng Kai Jin says the AI Lab is taking a risk-based approach to validation of tools.

“If it is high risk, more upfront testing should be conducted. For lower risk, more emphasis should be placed on ongoing monitoring,” explains Jin.

“But there is always a human in the loop.”

All three leaders say the promise of AI is significant but there is a need for critical evaluation.

Barnard says: “we have got to consider performance, computing power, network load, and also security, data sovereignty, and privacy aspects.”

MacDonald says it is important to have clinicians leading on this work as it is a duty of care to ensure the solutions are delivering as intended to patient populations and do not have unintended consequences.

Listen to our latest eHealth Talk podcast episode with Sharyn MacDonald, wherever you get your podcasts or listen here.

To comment on or discuss this news story, go to the eHealthNews category on the HiNZ eHealth Forum

You’ve read this article for free, but good journalism takes time and resource to produce. Please consider supporting eHealthNews by becoming a member of HiNZ, for just $17 a month

Read more AI & Analytics news

Return to eHealthNews.nz home page