Can AI spot lung cancer and cut diagnosis times?

A new UCLH study is hoping to reduce the time taken to diagnose lung cancer by using artificial intelligence (AI) software to detect abnormalities on chest X-rays.

GPs in England request around two million chest X-rays every year. The results of most X-rays come back as normal, but for a small group of patients, there may be early indications of lung cancer. Once these signs are picked up, the next stage is referral for a CT scan of their chest, to help confirm or rule out a diagnosis of lung cancer.

The LungIMPACT trial will evaluate AI technology called qXR which is designed to spot potential abnormalities on chest X-rays. The study is co-led by Dr Nick Woznitza and Professor David Baldwin. Dr Woznitza is a UCLH consultant radiographer.

The AI system, developed by the company, is designed to flag which patients would benefit from an urgent review of their scan by a reporting radiographer, who can arrange for a same-day CT scan if their X-ray indicates possible lung cancer.

In the study, the AI system will produce a secondary image for each chest X-ray with an overlay to highlight certain abnormalities. If the AI detects a problem, this will highlight the X-ray on the reporting worklist so that the reporting radiographer can prioritise this for urgent reporting. The research team will evaluate whether use of this AI ‘triage’ can successfully bring down time diagnosis times.

Dr Woznitza, who is also a clinical academic at Canterbury Christ Church University, said: “The quicker we pick up on any potential anomalies on a chest X-ray, the better. We review X-ray results as quickly as possible, but if this technology can help us prioritise who would most benefit from a rapid review of their X-rays, this would help improve outcomes for our patients.”

Prof Baldwin, honorary professor of medicine at Nottingham University Hospitals (NUH) and the University of Nottingham, said: “Studies evaluating the clinical impact of AI are urgently needed to ensure the safe and effective implementation that is needed to help the NHS and our patients. Doing these studies is a significant challenge but a worthwhile one.”

Darren Stephens, Senior Vice President & Commercial Head UK and Europe of said: “Trust in healthcare AI as a tool for supporting clinical case prioritisation is growing. Real-world evidence from collaborative trials such as LungIMPACT is vital to help power confidence in digital health innovations and improve speed of cancer care for patients now and into the future.”

For the study, the qXR software has been integrated into imaging and health record systems by UCLH’s Digital Healthcare team. It will analyse chest X-rays of adult patients referred to UCLH by their GPs.

This work was commissioned and funded by the NHS Cancer Programme, with the support of SBRI Healthcare and the NHS Accelerated Access Collaborative. It is supported by the National Institute for Health and Care Research Biomedical Research Centre at UCLH.