Spending on artificial intelligence (AI) in medical imaging is expected to continually trend upward, with a new report forecasting world expenditures to total more than $2 billion by 2023.
Signify Research, an independent global healthcare technology consultancy, based in the U.K. published its report on the topic August 2.
“Despite some of the earlier market hype, it is becoming increasingly clear that AI will transform the diagnostic imaging industry, both in terms of enhanced productivity, increased diagnostic accuracy, more personalized treatment planning, and ultimately, improved clinical outcomes,” wrote Simon Harris, a principal of Signify.
The report placed 2018 revenue forecasts for deep learning alone at around $200 million. That number is expected to climb up around $1.5 billion by 2023.
Once deep learning hit the market, and affordable cloud compute (GPU) and storage followed, the pace of product development for AI-based medical image analysis tools has now started to take-off, Harris wrote.
Breaking the market down by clinical application, the report forecasted neurology and cardiovascular software to account for the largest chunk of revenue market share by 2023. Harris partially attributed this growth to products that detect and diagnose stroke and tools that measure blood flow.
Adding to the projected growth is the change in attitude toward AI in the radiology from foe to friend, which he’s seen over the last 12 to 18 months, Harris noted.
However, there remain multiple barriers before AI begins to fully change medical imaging. Those include:
- Regulatory challenges, including few approved products.
- Large-scale validation studies are needed to boost faith in AI products.
- AI tools must be able to be fully integrated into radiology workflows.
- Healthcare providers remain skeptical of purchasing AI tools due to challenges over vendor specific implementation.
As medical imaging vendors begin to enter the space once dominated by start-ups and software developers, Harris believes barriers will begin to fall, and radiologists will have plenty of AI options to choose from.