By Professor in Nuclear Medicine Geoff Currie, AM, (pictured inset) in the Charles Sturt School of Dentistry and Medical Sciences in Wagga Wagga.

What is medical imaging?

Medical imaging relates to those imaging techniques in nuclear medicine and radiology used to create detailed images of the organs and tissues of the body to help diagnose and treat diseases.

These techniques include single-photon emission computerised tomography (SPECT), positron emission tomography (PET), computed tomography (CT), magnetic resonance imaging (MRI), ultrasound, Xray techniques, and hybrid techniques like SPECT/CT, PET/CT, and PET/MRI.

The different imaging tools are used to detect and localise disease, map disease progression, identify optimal treatment approaches, monitor response to treatment, and more.

What is artificial intelligence and how is it applied in medical imaging?

AI is a term that refers to a very broad range of tools that probably should not be used to describe what we are doing in medical imaging. AI is used in social media apps, satellite navigation, shopping, clothing design, household appliances and much more.

The application of AI in medical imaging is largely referring to algorithms developed to look at data and extract features from data (or images) that is not typical of the human observer, and to perform some of the repetitive data analysis tasks.

The ‘sharp edge’ of this area is the use of artificial neural networks that can have data or images as inputs for analysis. This is referred to as machine learning or, when using images and convolutional neural networks, deep learning.

I have introduced the term ‘intelligent imaging’ to reference the application of AI in medical imaging and ‘engineered learning’ instead of AI. But AI is language people relate to.

How will artificial intelligence change medical imaging, and why is it important?

AI has been changing medical imaging for decades, helping us analyse data and images, and automate processes.

The recent momentum has come from new capabilities in ‘deep learning’ that allow very abstract features to be extracted from images.

In an era of precision medicine, these deep learning algorithms can help to identify disease, best treatment options, or improve the process.

Some applications allow deeper insights for enhancing human reporting, others improve the accuracy of radiation dosimetry (measurement of radiation exposure from Xrays etc) for radiation therapies, and some even reduce the radiation dose to patients without compromising image quality.

In essence, the value of AI will be in allowing us to do more, do it quicker, do it more accurately, and improve patient outcomes.

How can members of the public (‘the community’) become better informed, express their views, and influence the way this new technology can be used?

It is hard because the technology and applications are often beyond the expertise of health workers, and so the public may find it more difficult to source information.

But the patients are the images or data being interpreted, and the patient is the beneficiary of improved outcomes, so it is important that patients ask questions, express their opinions and are represented.

Our research group’s survey (noted below) is one way for the industry to better understand what the public and patients know, understand and want from AI in medical imaging.

Are governments, business, and the scientific community likely to be responsive to public concerns about artificial intelligence?

If there are concerns, we need to hear them. We can only optimise care of patients by knowing preferences and concerns. And there are issues to resolve around equity of access, ethics, transparency in algorithms, and diversity.

But AI can bring opportunity too, such as potentially improved outcomes, lower costs, greater accuracy, and improved remote access.

By understanding what patients need or want, their concerns and their priorities, we can shape policy, research and practice of this frontier technology.

Should the community embrace AI, or be cautious about AI?

AI is a misnomer. The learning and capability itself are neither artificial nor intelligent. It simply appears so because there is substantial overlap between the most insightful algorithms and the least insightful humans.

Basic science, in the eyes of those that do not understand it, is exciting magic. In the hands of those without command, misanthropic sorcery.

As AI makes promises of social justice and equity, the boundaries of capability need to be masterfully explored to transition from frontier technology to community beneficence.

Yet we are right to fear artificial intelligence. This is because humans gifted with superior intelligence have consistently exploited that capability at the expense of other species, the environment, and even other humans to create extinction, habitat destruction, environmental crisis, and social asymmetry and injustice.

An increasing learning rate has simply increased disparity and exploitation. There is no evidence to suggest human decision making and judgement augmented by artificial intelligence will alter that course.

Our research and other leadership provided by Charles Sturt University includes a focus on ensuring we exploit the capabilities of artificial intelligence to create parity of basic human rights and of health, education, and security-of-self in favour, and to eliminate historical and institutional bias in algorithms to positively augment human decision making.

A mixed-skills research group has recently come together within the medical radiation science discipline at Charles Sturt University integrating researchers at Wagga Wagga and Port Macquarie campuses.

We are very keen to understand what the community perceptions and preferences are around the application of artificial intelligence in medical imaging.

The public can complete the Community Perspectives on Artificial Intelligence in Health and Medicine: Survey here.

For those interested in a deeper understanding of AI and medical imaging, people can read my Harry Potter-themed paper ‘A muggles guide to deep learning wizardry’ (published in Radiography, Elsevier, Vol. 28, Issue 1, February 2022, p 240-248), or watch my four-minute animation.