It has been over 130 years since the discovery of the x-ray by German physicist, Wilhelm Röntgen. In November 1895, while investigating the effects of electron beams in low-pressure gases, his work resulted in him being awarded the first Nobel Prize for Physics in 1901.

One of the first medical uses of x-rays occurred in 1896 when a British doctor, John Hall-Edwards, located a needle embedded in his colleague's hand. From there, NHS data shows that 45 million imaging tests were reported in England between March 2022 to March 2023. This is an increase of 2.2% from the 44 million in the previous year. In 2024, demand for scans increased by 8%, but there is estimated to be a 30% shortage of clinical radiologists. 

In response, NHS Trusts spent £325 million to manage that demand, with £216m of that going to private reporting companies – an increase of 24% in a year. If this continues, annual spending on outsourcing could exceed £400 million by 2028, enough to fund 3,389 full-time consultant radiologists, according to the Royal College of Radiology. 

Outsourcing poses further risks to patient safety in tandem with increased claims and their associated issues; such claims often involve determining where liability for the error lies and which party is responsible for the damages and costs. The need for carefully drafted contracts reflecting related risks and liabilities becomes paramount. 

Outsourcing can be beneficial but there are pitfalls; reporting can take place in any part of the world and across different time zones. Late night scans in the UK can be reported during more social hours in other parts of the world, but conversely it can also mean that a clinician working a second job may be viewing scans after their shift has ended, adding to the risk profile. 

The case of Karen Spellman v. Portsmouth Hospitals University NHS Trust (2024) realised these pitfalls, with the claimant’s radiology expert having been found to have “made a number of errors in his interpretation of the scan including identifying pathology at the wrong level”. Given that the expert reviewed the scan without the pressure of clinical practice demonstrates the difficulty of the task at hand and, when combined with the growing volume of scans to report, the pressure on radiologists to report accurately and in a timely manner is evident.

AI is becoming increasingly popular in an effort to remedy the current predicament of increasing demand coupled with the shortfall in radiology. In 2023, 54% of NHS Trusts were noted to be using AI tools to assist with interpreting, prioritising, or reporting on radiology. In 2024, the rollout of an AI diagnostic tool called Annalise.ai to seven Trusts across the Greater Manchester region was delivered to assist doctors in detecting conditions visible on chest x-rays, like cancer, more quickly. 

This AI tool can detect up to 124 findings on a chest radiograph in seconds, thereby helping to identify points of interest on a scan for the radiologist to review, and should it identify possible lung cancer, it conveys this information to the reporting medical provider in less than a minute, allowing for a swifter prioritisation of imaging. Implementation of AI in radiology is only set to expand, with NHS England’s AI Diagnostics Fund (AIDF) providing £21 million to 11 imaging networks across 64 NHS Trusts to utilise this technology.

As healthcare providers continue to cross the AI frontier, recent statistics show that AI has not been a panacea to the issues faced by radiologists today. In the RCR Workforce Census 2024, it was reported that 56% of Clinical Directors experienced no significant change to their workloads due to the implementation of AI tools, and 37% reported an increase in their workload due to using AI software. 

While the large-scale bedding in and training for such technology will naturally take time, some are concerned that AI will adversely impact the way that radiologists examine and report imaging. Given the capacity of AI to review imaging at a relatively rapid pace, some believe this may mean that radiologists are given more specific information to consider, extending, rather than reducing, the time that they would have spent reporting on each image.

Radiology will always be an essential diagnostic tool and the risk of errors and resulting legal action will not abate, particularly where workforce challenges faced by radiologists remain. Undoubtedly, AI will impact the claims space but for all its benefits, AI cannot (yet) increase the supply of reporting radiologists’ time that is needed to meet the ever-growing demand for imaging. 

We’ve come a long way since the discovery of the x-ray but however AI supports the human race, the role of the human being remains and is ever more in demand to help balance competing challenges.