Van Leeuwen K et al, in a new study from Project AIR in Radiology comparing AI algorithms.

Imaging Wire Sponsors

Bayer  •  Blackford  •  CARPL.ai  •  Change Healthcare  •  Clearpath  •  Enlitic  •  GE HealthCare  •  Gleamer  •  Intelerad  •  Medality  •  Merge by Merative  •  Nuance  •  PocketHealth  •  Riverain Technologies  •  Siemens Healthineers  •  SpinTech MRI  •  TeraRecon  •  United Imaging  •  Us2.ai  •  Visage Imaging

Artificial Intelligence

AI Models Go Head-to-Head in Project AIR Study

One of the biggest challenges in assessing the performance of different AI algorithms is the varying conditions under which AI research studies are conducted. A new study from the Netherlands published this week in Radiology aims to correct that by testing a variety of AI algorithms head-to-head under similar conditions. 

There are over 200 AI algorithms on the European market (and even more in the US), many of which address the same clinical condition. 

• Therefore, hospitals looking to acquire AI can find it difficult to assess the diagnostic performance of different models. 

The Project AIR initiative was launched to fill the gap in accurate assessment of AI algorithms by creating a Consumer Reports-style testing environment that’s consistent and transparent.

• Project AIR researchers have assembled a validated database of medical images for different clinical applications, against which multiple AI algorithms can be tested; to ensure generalizability, images have come from different institutions and were acquired on equipment from different vendors. 

In the first test of the Project AIR concept, a team led by Kicky van Leeuwen of Radboud University Medical Centre in the Netherlands invited AI developers to participate, with nine products from eight vendors validated from June 2022 to January 2023: two models for bone age prediction and seven algorithms for lung nodule assessment (one vendor participated in both tests). Results included:

• For bone age analysis, both of the tested algorithms (Visiana and Vuno) showed “excellent correlation” with the reference standard, with an r correlation coefficient of 0.987-0.989 (1 = perfect agreement)
• For lung nodule analysis, there was a wider spread in AUC between the algorithms and human readers, with humans posting a mean AUC of 0.81
• Researchers found superior performance for Annalise.ai (0.90), Lunit (0.93), Milvue (0.86), and Oxipit (0.88)

What’s next on Project AIR’s testing agenda? Van Leeuwen told The Imaging Wire that the next study will involve fracture detection. Meanwhile, interested parties can follow along on leaderboards for both bone age and lung nodule use cases. 

The Takeaway

Head-to-head studies like the one conducted by Project AIR may make many AI developers squirm (several that were invited declined to participate), but they are a necessary step toward building clinician confidence in the performance of AI algorithms that needs to take place to support the widespread adoption of AI. 

Two Questions about AI for Radiology Leaders

Are today’s radiology AI solutions solving the right problems? And are there other solutions available for AI of brain MRI? Read this article from SpinTech MRI to learn how its STAGE solution can optimize MRI utilization. 

The Present and Future of Lung Cancer Screening

What is the latest evidence to support lung cancer screening, and what role will AI play? Join this Bayer webinar on January 17 to learn about evidence-based recommendations for the diagnosis and management of lung cancer. 

The Wire

• GE to Acquire MIM Software: GE HealthCare is expanding its position in advanced visualization software for radiation oncology and other therapy-based applications with a deal to acquire MIM Software of Cleveland. MIM’s focus has been on software for radiation therapy treatment planning as well quantitative analysis of nuclear medicine brain and cardiac images; GE plans to integrate MIM’s offerings into its own AV portfolio to facilitate AI-based image segmentation and contouring, as well as dosimetry analysis. The deal dovetails well with GE’s recent emphasis on theranostics and precision medicine. 

• RSNA to Stay in Chicago through 2032: RSNA has tacked another four years on to its contract to hold its annual meeting at McCormick Place in Chicago. The extension through 2032 is seen as a boon to Chicago’s recovering tourist industry, but it’s also good news for fans of the Windy City (including those of us at The Imaging Wire). Last year, RSNA announced that starting in 2027 it would move meetings in odd-numbered years to mid-November rather than the traditional Thanksgiving weekend slot.

• AI of CCTA Cuts Processing Time: Using the CoronaryDoc AI algorithm from Shukun Technology of China, researchers slashed CCTA image processing and report turnaround times without a negative impact on accuracy. In European Radiology, Chinese researchers found that in a population of 1.8k patients, CoronaryDoc cut image processing times compared to conventional semi-automated mode for CCTA (121 vs. 434 seconds) as well as scan-to-report-release times (6.4 vs. 10.5 hours); metrics like sensitivity, specificity, and patient outcomes at two years were largely the same.

• Butterfly to Launch Third-Gen US Scanner: Butterfly Network is launching the third generation of its semiconductor-based handheld ultrasound technology after receiving FDA clearance for its Butterfly iQ3 device. The new scanner sports a new ergonomic design and will deliver double the data processing speed, which will produce better image resolution, sensitivity, and penetration, as well as faster 3D capabilities. Butterfly CEO Joe DeVivo gave us more details on iQ3 at this week’s JP Morgan Healthcare Conference. 

• Is Same-Day Biopsy Really Better for Women? In the quest to improve breast cancer screening, same-day biopsy of suspicious breast lesions has been offered as an option. But is it really better? Results were mixed in a survey of 974 women in JACR, about half of whom got same-day biopsy. Women surveyed prior to biopsy reported higher satisfaction levels, but they also reported higher anxiety, and there was no difference in satisfaction when surveyed after biopsy. Authors believe better radiologist communication can alleviate patient anxiety before biopsy.

• Qure Gets Another FDA Clearance: Qure.ai is expanding its portfolio of AI solutions for chest imaging with the FDA clearance of its qXR-LN algorithm for analyzing lung nodules from 6-30mm on chest X-rays. The algorithm is intended for use by radiologists, pulmonologists, and ER physicians, and can also act as a second reader for doctors. This is the sixth clearance for Qure in chest X-ray, and is the company’s 13th clearance overall; the company has seen growing support from recent clinical studies for its chest X-ray solutions. 

• Study Supports AI Double Reading for Breast Screening: Another clinical study supports using Transpara’s ScreenPoint AI instead of a first reader in breast screening programs that perform double reading. In a study of 159k women in Denmark in Cancer Imaging, researchers saw a sensitivity improvement when AI replaced the first reader compared to the first reader alone (74% vs. 64%), while specificity was largely the same (97.8% vs. 97.9%). The findings support real-world use of AI as a first reader, but setting the right AI threshold is key.

• Us2.ai’s Home Heart Failure Advantage: Access to echocardiography has long been a hurdle for effective heart failure management, and results from the CUMIN study suggest that echo AI-enabled home exams might help overcome these barriers. Researchers trained seven echo-novice nurses on performing Us2.ai-enhanced cardiac ultrasound exams for one day, and then had them examine 94 Tunisian patients without previous heart failure diagnosis. The novice at-home AI-POCUS exams achieved higher sensitivity (92% vs. 87%) and AUC (0.86 vs. 0.64) for detecting the primary endpoint (LVEF <50% or LAVI mL/m2) than NT-proBNP testing.

• 4DMedical Partners with Slovakian Hospital: 4DMedical has signed a contract to install its XV Technology for lung imaging at facilities in Slovakia operated by Penta Hospitals Slovensko. XV transforms medical images into visual maps of a patient’s lungs to detect perfusion issues that can be signs of lung disease; the company has begun rolling XV out globally after receiving regulatory clearances. The agreement with Penta will enable 4DMedical to access a framework for data collection to help the company’s expansion into new markets in the European Union. 

• Cleerly’s ISCHEMIA Clearance: Cleerly significantly expanded its cardiovascular AI portfolio with the FDA clearance and launch of its new Cleerly ISCHEMIA solution. Available via prescription to users of Cleerly LABS software, Cleerly ISCHEMIA analyzes CCTAs to determine the presence of coronary vessel ischemia based on quantitative measures of atherosclerosis, stenosis, and significant vascular morphology. The launch appears to mark Cleerly’s transition from a focus on plaque analysis to supporting diagnosis across the cardiovascular care continuum.

• Mayo Signs Cerebras for Generative AI: Mayo Clinic has signed up Cerebras Systems of Silicon Valley as a corporate partner for developing generative AI algorithms for healthcare. In an announcement at the JP Morgan Healthcare Conference, Mayo said the partners would develop models for tasks like summarizing complex medical records or analyzing images for patterns that humans might miss. Technology developed through the partnership will be made available on Mayo Clinic Platform, Mayo’s AI deployment infrastructure. Cerebras develops hardware and software for AI development.

• GANDALF Works Magic on AI Data: We don’t totally understand this paper, but it gets the award for best radiology acronym of 2024 so far: GANDALF, or Graph-based TrANsformer and Data Augmentation Active Learning Framework. In Medical Image Analysis, researchers describe GANDALF as a tool they developed for augmentation of chest X-ray datasets with synthetic images to support the development of more robust AI algorithms that can detect multiple diseases. Can FRODO be far behind?

A New Era for fMRI

Reducing acoustic noise from MRI scanners is particularly important in rapid functional MRI exams of the brain. In this article, learn what GE HealthCare is doing with Looping Star, a novel noise reduction method for whole-brain fMRI scanning.

Why One Healthcare Cloud Doesn’t Fit All

When it comes to the cloud for healthcare providers, one size does not fit all. Healthcare organizations are adopting a mix of hybrid and multi-cloud operating models; learn more about this emerging trend in this article by Intelerad Chief Product Officer AJ Watson.

Turning Medical Imaging into Great Medicine

Imaging precision is at the core of positive patient outcomes. Learn how the Intuition 4.7 advanced visualization solution from TeraRecon is powering better patient care across the spectrum. 

The Resource Wire

• Improving Patient Outcomes in Lung Cancer: Early detection is key to improving treatment pathways and patient outcomes in lung cancer. Learn more facts about lung cancer screening on this resource page from Riverain Technologies. 

• Change Healthcare’s Secure Cloud: Did you know one-quarter of healthcare organizations have experienced a cyber-attack in the last year? This Change Healthcare animation explains how third party-certified cloud-native enterprise imaging can help secure IT infrastructure that might be exposed with re-platformed imaging systems.

• An Update from Merge by Merative at RSNA 2023: At RSNA 2023, we caught up with Merge by Merative’s Julie Pekarek, head of product and strategy, for an update on the company’s cloud solutions, their new digital pathology and 3D printing announcements, and how Merge is tackling challenges around workflow and burnout.

• Unlocking the Potential of RadiologyAI: How is AI revolutionizing radiology to enhance patient outcomes and streamline workflow? Find out in this video discussion between CARPL.ai CEO Vidur Mahajan and Mitchell Goldburgh of NTT Data USA.

• Overcoming Challenges to Achieving Quadruple Aim Goals: Watch a highlight from RSNA explaining how to overcome barriers to improving population health, patient experience, avoiding revenue loss and increasing staff wellness. Presented by PocketHealth Co-Founder & CEO Rishi Nayyar.

• Hot Topics from RSNA 2023: What were some of the hot topics at RSNA 2023? In this edition of The Imaging Wire Show, we caught up with Dave Wilson, vice president of marketing and communications at Enlitic, to discuss major trends at the meeting and new technologies the company was highlighting.

• Clinical Excellence on the Move: Healthcare providers are striving to deliver faster, better care for patients, no matter where they are. Learn how the MAGNETOM Viato.Mobile 1.5T MRI scanner from Siemens Healthineers delivers high-quality care where your services are needed most.  

• Solutions to Transform Medical Image Delivery: Medical providers and health systems are looking to ditch the disc and modernize their patients’ journey. Learn how Clearpath helps them reduce the cost of retrieving and fulfilling patient requests for imaging and medical records.

• Revolutionizing Chest X-Ray Reporting: A new study in Radiology shows how Gleamer’s ChestView AI solution is revolutionizing chest X-ray reporting by helping radiologists detect lung abnormalities with greater accuracy while also reducing interpretation times. 

• Monitoring AI Solutions in Healthcare: Clinical applications for healthcare AI are rapidly expanding, but many barriers are still preventing widespread adoption. This Nuance post explores a critical set of questions: what happens after an AI model goes into production, and how to know if it continues to perform as expected?

• The Journey to Cloud-Based PACS: Radiology is looking closely at cloud-based PACS as an alternative to on-premises digital image management. Learn about providers who have made the switch with Visage Imaging in this white paper from Signify Research. 

• Reducing the Risk of Medical Errors and Misreads: What are the factors that contribute to errors and misreads in radiology – and how can they be prevented? Find out how to improve your practice’s quality and accuracy in this white paper from Medality. 

• Echo AI and Mobile Health: Powered by AI-automated software, mobile ultrasound devices are taking echo exams out of the conventional clinical setting and on the road. In this article, learn how Singapore is using Us2.ai’s echo AI software to complete the full mobile exam pathway while maintaining high clinical standards. 

• United Imaging’s RSNA 2023 Highlights: What were the highlights in United Imaging’s booth at RSNA 2023? Watch this video interview with CEO Jeffrey Bundy, PhD, at the conference to learn about the company’s 510(k)-pending 5T MRI scanner, mobile digital radiography system, and more. 

• Unlocking the Value of AI: Despite significant interest, there’s still confusion about the value of imaging AI. This Blackford Analysis white paper explores the key cost considerations and ROI factors that radiology groups can use to figure out how to make AI valuable for them.