Skip to main content
Home » Digital Health » The promise and challenges of imaging AI
Digital Health 2021

The promise and challenges of imaging AI

iStock / Getty Images Plus

Dr Sarim Ather

The Royal College of Radiologists’ Radiology Informatics Committee
AI Clinical Champion at the National Consortium of Intelligent Medical Imaging

AI can read scans and improve hospital workflows, but what does the imaging AI landscape look like now for radiologists?

From self-driving cars to facial recognition, artificial intelligence (AI) plays an increasing role in our daily lives. Within the highly regulated healthcare landscape, development has been slower in order to ensure patient safety. However, the clinical community is both optimistic and proactive when it comes to harnessing AI’s huge potential.

How imaging AI can and will improve care

There are many areas where AI can enhance hospital imaging. For example, automating the measurement of abnormalities will improve the speed and accuracy of disease staging and management. In future, algorithms may be able to integrate data such as blood test and genomic results with scan findings to give prognoses or recommend the best treatment regimen.

One ongoing NCIMI AI project I’m involved with is a partnership between my trust and GE Healthcare to evaluate an algorithm for detecting collapsed lungs.

AI will also have a big role in improving backroom radiology by flagging abnormal scans for urgent review and improving appointment scheduling.

What imaging doctors need

With a variety of AI solutions coming online, it’s important that radiologists understand the benefits and limitations of the algorithms at their disposal.

The Royal College of Radiologists (RCR) has a unique role in supporting the radiology community in training and understanding the use of AI, as well as supporting engagement with its development. It’s important the safety and efficacy of any new tool is proven prior to adoption, and that innovations lead to time efficiencies, rather than greater burden on a stretched NHS workforce.

AI will also have a big role in improving backroom radiology by flagging abnormal scans for urgent review and improving appointment scheduling.

Overcoming hurdles to widespread adoption

The UK has a thriving AI innovation ecosystem. However, the lack of AI integration with clinical workflow and poor efficacy evidence has been a barrier to widespread adoption.

The RCR has taken a leading role in highlighting past lessons to the new wave of innovators, as well as engaging with industry to ensure new AI integrates with existing IT infrastructure. Work is also well underway from UK regulators to develop robust regulatory pathways to ensure AI is safe and can be rolled out quickly. Even with real-world evidence and clearing regulatory hurdles, the NHS can still be slow to up-scale innovation. Organisations like NHSX have a key role in supporting adoption at a national level, as do programmes such as NCIMI by bringing the right specialists together.

Successful implementation of imaging AI depends on engaged staff, armed with the right tools. The radiology community needs AI champions to promote buy-in and pave the way, supported by the right NHS IT infrastructure.

Next article