Alan Karthikesalingam, clinical research scientist at Google Health in London, in an article on AI for medicine in Nature.

In this episode of the Imaging Wire Show, we talked to Kristina Lång, MD, about new research into AI for mammography screening. Lång is associate professor in radiology diagnostics at Lund University in Sweden and is the principal investigator on the landmark MASAI trial on the use of AI for mammography.

Imaging Wire Sponsors

Bayer  •  Blackford  •  CARPL.ai  •  Change Healthcare  •  Clearpath  •  Enlitic  •  GE HealthCare  •  Intelerad  •  Medality  •  Merge by Merative  •  Nuance  •  PocketHealth  •  Riverain Technologies  •  Siemens Healthineers  •  Subtle Medical  •  Tempus Radiology  •  United Imaging  •  Us2.ai  •  Visage Imaging

Artificial Intelligence

Predicting the Future of Radiology AI

Making predictions is a messy business (just ask Geoffrey Hinton). So we’re always appreciative whenever key opinion leaders stick their necks out to offer thoughts on where radiology is headed and the major trends that will shape the specialty’s future. 

Two of radiology’s top thought leaders on AI and imaging informatics – Curtis Langlotz, MD, PhD, and Paul Chang, MD – gaze into the crystal ball in two articles published this week in Radiology as part of the journal’s centennial celebration. 

Langlotz offers 10 predictions on radiology AI’s future, briefly summarized below:

• Radiology will continue its leadership position when it comes to AI adoption in medicine, as evidenced by its dominance of FDA marketing authorizations
• Virtual assistants will help radiologists draft reports – and reduce burnout
• Radiology workstations will become cloud-based cockpits that seamlessly unify image display, reporting, and AI
• Large language models like ChatGPT will help patients better understand their radiology reports
• The FDA will reform its regulation of AI to be more flexible and speed AI authorizations (see our article in The Wire below)
• Large databases like the Medical Imaging and Data Resource Center (MIDRC) will spur data sharing and, in turn, more rapid AI development

Langlotz’s predictions are echoed by Chang’s accompanying article in Radiology in which he predicts the future of imaging informatics in the coming age. Like Langlotz, Chang sees the new array of AI-enabled tools as beneficial agents that will help radiologists manage growing workloads through dashboards, enhanced radiology reports, and workflow automation. 

The Takeaway

This week’s articles are required reading for anyone following the meteoric growth of AI in radiology. Far from Hinton’s dystopian view of a world without radiologists, Langlotz and Chang predict a future in which AI and IT technologies assist radiologists to do their jobs better and with less stress. We know which vision we prefer.

New Guidelines on Heart Failure Management

The ESC has updated its heart failure guidelines, for the first time including a guideline-directed medical therapy for heart failure with preserved ejection fraction (HFpEF). Learn how to trust echo AI to find these cases earlier in this video from Us2.ai.

Using AI to Assess Coronary Artery Calcification

Learn about the power of the platform to deliver multiple AI solutions in this October 26 webinar sponsored by Nuance. Discover how to optimize your existing data with AI to get more out of routine imaging.

The Wire

• The AI Revolution in Medicine: In the same vein, a new article in Nature explores how AI is revolutionizing medicine, seven years after Geoffrey Hinton’s controversial prediction that AI would replace radiologists. The article sees promise in generalist AI models that can be used for many tasks, and also the potential of foundational models as a better way to train AI algorithms. The story concludes that despite the great promise of AI in medicine, it’s still (currently) no match for radiologists. 

• AI Triages Chest X-Rays: A homegrown AI algorithm developed by South Korean researchers was able to triage from the radiology worklist 40% of chest X-rays most likely to be normal. In a paper in Radiology, researchers described how their algorithm analyzed chest X-rays acquired at two different time points, and triaged the ones that had no changes while detecting the pairs that had interval changes. The research is in line with studies like MASAI, in which AI was able to remove 44% of normal mammograms from the worklist.

• Unilabs Adds Annalise AI: European imaging services provider Unilabs has adopted the Annalise Enterprise CTB AI solution from Annalise.ai for analyzing non-contrast CT brain exams. The algorithm can identify 130 different radiology findings and can alert emergency radiologists to pathologies that may need immediate attention. Unilabs will use the algorithm for acute and trauma care. The partnership adds to other agreements Annalise has been signing in Europe, such as a recent deal to enter the Nordic region. 

• FDA Issues Guidelines for AI Changes: How do you regulate a medical product that’s constantly learning – and thus changing – as it goes along? The FDA and regulatory agencies from the UK and Canada this week issued guidelines for predetermined change control plans for AI/machine learning-enabled devices – advice that developers should follow to determine when a device or software application has changed significantly enough to require new authorization. The guidelines are particularly relevant given the recent surge in new AI/ML marketing authorizations. 

• GE Tops FDA’s AI Authorization List: While we’re on the topic of the FDA’s AI regulation, GE HealthCare says it came out on top of the agency’s recently published list of regulatory authorizations as the company with the largest number of products on the list. The FDA released the list last week, identifying 155 new authorizations for the year through July 30, 2023 and almost 700 since the agency began tracking them. GE said it has 58 products on the list, the most of any medical technology company. 

• HAP Wins New Client in New England: Revenue cycle management company Healthcare Administrative Partners (HAP) continues to add to its stable of clients in New England. The company has signed up the department of radiology at York Hospital, which operates community sites and five walk-in care centers in southern Maine and New Hampshire. The deal came as current HAP client Commonwealth Radiology Associates secured a contract to operate York Hospital’s department of radiology. HAP earlier this month brought on Radiology Associates of Keene in New Hampshire.

• Patients Prefer Structured Reports: Patients are increasingly getting access to their own radiology reports. That begs the question, what do patients like in a report? In Current Problems in Diagnostic Radiology, researchers found that in a survey of 135 CT and MRI reports from 2019 to 2021, the reports that patients rated highly were more likely to be structured (94% vs. 7%). On the other hand, highly rated reports had lower readability scores on the Flesch Reading Ease scale (20 vs. 29). 

• Esaote Launches New Open MRI: Italian medical device vendor Esaote has launched S-scan Open, a new configuration of the company’s S-scan system, a 0.25T C-shaped open scanner targeted at physician offices. S-scan Open has been optimized for MSK exams and includes the company’s eXP package with image optimization algorithms and streamlined clinical protocols. The scanner also includes workflow optimization and a new open knee coil. Esaote recently secured a group purchasing agreement with Premier for its O-scan dedicated extremity MRI scanner. 

• High-Res MRI Predicts Stroke Recurrence: High-resolution brain MRI with a 3T scanner predicted stroke recurrence in a new study in European Radiology. Researchers from China scanned 132 patients from 2017 to 2020 and focused on intracranial plaque burden and composition. They found that patients had nearly triple the hazard ratio for stroke recurrence (HR=3.2) for every 10% increase in plaque burden, and nearly double for gadolinium enhancement ratio (HR=2.2). Such quantitative estimates could be more accurate than qualitative assessment for predicting stroke recurrence.

• PE Firm to Take Akumin Private: Imaging center company Akumin has agreed to be taken private in a buyout from Stonepeak, a private equity firm specializing in infrastructure and real estate. The transaction will start with a “prepackaged” Chapter 11 filing, and Akumin’s existing $470M debt to Stonepeak will be canceled and converted into common shares. Stonepeak will also invest an additional $130M into the company. Akumin is still grappling with the effects of a ransomware attack earlier this month that forced it to cancel scans. 

• iCAD to Sell Xoft Brachytherapy to Elekta: iCAD has inked a deal to sell its Xoft brachytherapy business to radiation therapy vendor Elekta for $5.5M and assumed liabilities. iCAD acquired Xoft in 2010; its flagship product is the Axxent brachytherapy system, which delivers targeted radiation to cancer sites while sparing healthy tissue. With radiation therapy its primary business, Elekta seems to be a better fit for Xoft, and the deal will enable iCAD to focus on its core AI software for breast imaging.

• AI Predicts Breast Screening Non-Compliance: Value-based care software developer Cedar Gate Technologies says its predictive AI model was able to determine that 55% of the 2.4M women in its database won’t get a screening mammogram. The model generates a probability score for breast screening compliance, but found that 1.3M women fell below its predictive scoring threshold and were classified as “unlikely” to get mammography. Such predictive scores could be used with targeted interventions to boost screening compliance.

• Lucida Gets CE Mark for Prostate AI: Lucida Medical has received the CE Mark allowing European marketing of its AI-based Prostate Intelligence (Pi) prostate cancer detection software. Pi analyzes prostate MRI scans and automatically produces 3D segmentations, volumes, and risk scores for cancer. Lucida touted recent studies indicating that Pi can cut undetected cancers by 6% while reducing unnecessary biopsies to 43%. The company plans to launch Pi in Europe and the UK in Q1 2024.

The Benefits of an AI Accelerator

What is an AI accelerator, and how can it impact your practice? Learn about the Visage AI Accelerator program and its impact on research and clinical practice in this video from Visage Imaging.

What Exams Are Radiologists Reading?

What kinds of medical imaging exams are radiologists reading today, and how confident do they feel across different subspecialties? Find out in the 2023 Radiology Practice Development Report from Medality.

The Resource Wire

• Discover the Power of the AI Platform: Do you want to streamline your diagnosis, improve your efficiency, and measure your ROI? Then reserve a slot at the Bayer Experience Center at RSNA 2023 to discover the power of the AI platform with Calantic Digital Solutions and Bayer’s ROI calculator.

• Imaging Data Management for PACS and IT: Effective data management policies can help PACS and IT administrators get their jobs done more efficiently. Find out how data standardization practices like good data governance and standardized descriptions can help in this Enlitic white paper.

• Advanced Imaging Access for All: Patients’ zip codes shouldn’t determine the healthcare they receive. Siemens Healthineers aims to make advanced diagnostic imaging accessible for all patients. Register today for a launch event on November 16 at 12:00 p.m. ET to meet the company’s new Somatom CT family member. 

• Delivering the Quadruple Aim Through Image Exchange: Learn how a redesigned enterprise image exchange can enhance patient experiences, improve population health, generate cost savings, and increase staff satisfaction in this white paper from PocketHealth.

• Solving AI’s Adoption Challenges: The flow of new AI applications makes it hard for radiology groups to determine which tools would help them. In this Blackford Analysis white paper, radiology and IT leaders from NYU and Canopy Partners share how to solve these challenges.

• Learn about Subtle Solutions: How can Subtle Medical’s advanced solutions like SubtleMR and SubtlePET – and upcoming SubtleSYNTH launch – improve image quality, workflow efficiency, and patient care? Find out in this RSNA 2023 Lunch & Learn session at Noon Tuesday November 28. 

• Real patients. Real stories: Meet the women who inspire GE HealthCare to keep pushing for progress and pioneering mammography technology that improves health and can save lives. Hear their stories in these videos.

• The Power of Enterprise Image Exchange: Exchange medical data across the enterprise and grow your referrals and patient transfers. Share images in real-time – no VPN or CDs required – with Intelerad’s Enterprise Image Exchange.

• The Multitenant Cloud Advantage: Check out this Change Healthcare video explaining the difference between single-tenant and multitenant cloud architecture, and how multitenant solutions can improve your efficiency and flexibility.

• Power at the Point of Care: Are you ready to upgrade your X-ray technology? Check out United Imaging’s high-performance X-ray systems at RSNA 2023 at Booth #4100 and see the uDR 380i Pro Mobile X-ray system navigate its own obstacle course.

• Focus on Actionable Data: Riverain Technologies’ ClearRead technology is the first FDA-cleared Clear Visual Intelligence (CVI) system that improves both detection and reading time for lung abnormalities. Learn more in this article. 

• A Better Way to Deploy AI: CARPL.ai’s end-to-end modular platform for AI deployment is changing how University Hospitals in Cleveland uses AI to enhance patient care. Hear from UH clinicians and administrators as they share their experiences.

• Automated Patient and Provider Sharing: What’s the latest news from Clearpath? Find out how its fully automated patient and provider solutions for sharing medical records and images are making waves in healthcare.

• Address Your Imaging Needs Today: Looking for a partner to help guide your digital transformation? Face your future with confidence with enterprise imaging solutions from Merge by Merative. Learn more in this white paper.