MJ Medical

Artificial intelligence and diagnostic radiology – trend or travesty?

12-Jun-2019

In a recent paper, one of our Principal Consultants, Dan Gibson, looks into the impact of artificial intelligence on diagnostic radiology technologies

Introduction

The past few years have witnessed an explosion in the development of AI in healthcare. Whilst clinical applications are widespread, this research primarily focused on diagnostic and radiology integration of AI in healthcare. We reviewed the current and developing uses of AI and machine learning in diagnostics and radiology, to understand the efficacy of the solutions, regulatory requirements, and accessibility and sharing of patients’ data. Our aim was to understand ways in which AI can support and enhance workflows for clinical teams, creating efficiency, and improving patient safety and outcomes.

Methodology

Our approach began with a literature review focused on published research on artificial intelligence and machine learning in healthcare. We then undertook semi-structured interviews  with diagnostic radiology manufacturers, AI development organisations, research institutes, and radiological associations. A small sample of informal semi-structured interviews were also undertaken with practising UK radiologists.

Results

The research identified a number of subsets of AI, which are different in function and have differing outcomes. Radiology manufacturers already use AI within imaging modalities, with diagnostic reporting being the most exciting area where AI has moved beyond recognition and computer-aided detection to sensing, reasoning, acting and adapting. New AI systems are available to review medical images, spot anomalies, and develop and iteratively refine treatment plans to achieve high success in prediction rates. In clinical pathways, where diagnosis and treatment time are critical factors for long-term prognosis and recovery, the benefits are compelling. Internationally, there is a shortage in training, recruitment and retention of radiologists, who also have among the highest levels of professional burnout in the healthcare sector. AI offers the opportunity to improve the workflow of radiologists, reduce their workload, increase throughput of patients, and enhance human interaction between patients and clinicians.

Conclusions

By infusing AI into the physical space of a health system, healthcare can become safer, offer better care and delivery for the patient, and optimise workflow for clinicians. The research suggests the future of AI is in collaboration with humans, working together to enhance healthcare. It will not replace radiologists – but radiologists who use AI may replace those who don’t.