The AI Avalanche in Radiology

Recent technological advances have made machine learning tools more accessible and affordable than ever before. And radiology, as one-of most data-rich dep

post image

Recent technological advances have made machine learning tools more accessible and affordable than ever before. And radiology, as one-of most data-rich departments in healthcare, has seen the introduction of an avalanche of AI algorithms in recent months.

The number of AI start-ups that have entered the market in the past few years is nothing short of incredible. In June last year, Signify Research found 50 different AI use cases from 81 companies that currently have regulatory approval. A use case means that at least one vendor that has regulatory approval for an algorithm that detects a certain pathology, such as intracerebral hemorrhage (ICH), for example. If you compare these figures to those from even just a year earlier, only a handful of AI use cases had been approved for radiology purposes. Given the amount of data available to train and validate radiology AI algorithms, we can expect that the pace of innovation and regulatory approval will most likely only increase in the months and years ahead. Growing maturity in AI applications What we have also observed over this period is a growing maturity in the AI toolsets available to radiology departments. Many AI tools initially entered the market with a minimally viable product (MVP), providing just enough features to attract initial customers and validate a tool’s core value proposition. Today, many of these MVPs have now been in the market for a couple of years, validating assumptions and learning about user preferences. It now appears that many AI tools are entering the next major phase in product development, and are becoming minimally marketable products (MMPs). An MMP incorporates a core set of functionalities that address customer needs, deliver the desired user experience, and can start creating quantifiable value. Challenging choices However, as the number of vendors and algorithms continue to climb, healthcare providers and radiology departments are challenged to determine which algorithms will have the biggest impact for them and, ultimately, their patients. There are a host of reasons why providers may feel overwhelmed when it comes to trying to make decisions on which AI applications to move forward with. They may be skeptical of claims made by AI firms based on previous experiences, and they may have concerns about the maturity of the product or the company behind it. Or they may fear adding new technologies will hinder rather than help a radiology department’s efficiency. And, while the concerns of IT departments are different, they are no less valid – whether it’s limited resources, security concerns, numbers of integrations, complexity of implementation, or concerns about ongoing maintenance and updates. All of these concerns and more are addressed directly by Blackford Platform and its curated approach to effective selection, deployment, orchestration and use of market-leading medical imaging applications and AI. Stay tuned for our next blog when we’ll talk about how.