Artificial Intelligence Read X-Rays

Robert Preidt of the HealthDay Reporter gives us more information on what artificial intelligence read x-rays are and how it works.

The artificial intelligence (AI) system analyzes chest X-rays and has the ability to spot patients who should receive immediate care.  Reducing backlogs in hospitals could someday also be helped with the system.  About 40% of all diagnostic imaging worldwide is accounted for by performing chest X-rays.  This can cause a large backlog, according to the researchers.

At this time there are no systematic and automated ways to triage chest X-rays and provide the critical and urgent findings to the top of the reporting.  Giovanni Montana is formerly of King’s College London and now at the University of Warwick in Coventry, England who believes this information should be at the top of the pile.

He and his colleagues used more than 470,300 adult chest X-rays to develop the AI system that identifies unusual results.  The performance of this system in prioritizing X-rays was assessed in a simulation that used a separate set of 15,887 chest X-rays, with the identifying information removed to protect patient privacy.

There was a high accuracy with the system in distinguishing abnormal from normal chest X-rays.  With the AI system, simulations showed critical findings that received an expert radiologist’s opinion within an average of 2.7 days, which was compared with an average of 11.2 days in actual practice.  The study results can be found published in the journal Radiology.

These results are very exciting as they demonstrate an AI system can be successfully trained using an extremely large database of routinely acquired radiologic data.  This technology is expected to reduce a radiologist’s workload by a significant amount by detecting all the normal exams with further clinical validation.  More time can then be spent on those that require more immediate attention.

The next step is to test a much larger number of X-rays and conduct a multi-center study to assess the AI system’s performance.

Dr Fredda Branyon