|
Lung Screening’s Star Turn | Questions about AI
September 9, 2024
|
|
|
|
|
Together with
|
|
|
|
“Sorry Brian Fantana… ‘60% of the time, it works every time’ isn’t good enough for AI results.”
|
|
Chen Yeshai Goodman, on recent news questioning the value of AI for radiologists.
|
|
|
|
The World Conference on Lung Cancer (WCLC) is underway in San Diego this week, and CT lung cancer screening has had a starring role at the meeting. The sessions come as lung screening continues to build momentum through 2024.
Low-dose CT lung screening got the green light from the USPSTF over a decade ago, but screening rates are still mired in the single digits in many regions.
- The evidence backing LDCT’s life-saving value has been building, however, and around the world countries are launching national screening programs to counter the smoking epidemic, the leading cause of preventable cancer death worldwide.
Sessions at WCLC 2024 have highlighted this progress, with many speakers focusing on ways to boost screening compliance or use tools like AI to detect more lung cancers.
Presentations on early lung cancer detection have included the following findings…
- Three years of lung screening starting in 2021 in Quebec produced a lung cancer detection rate of 1.6% in the first screening round, with 85% of cancers stage I or II.
- Advanced practitioner nurses are being trained in Australia to assess pulmonary nodules to alleviate workforce challenges when the country’s national lung screening program starts in July 2025.
- Using Coreline Soft’s AVIEW algorithm to read baseline LDCT exams helped BioMILD researchers move to a triennial screening interval without missing cancers.
- The QUILS system for lung cancer quality assurance helped assess quality across multiple LDCT screening sites in Kentucky.
- Over 10 years in which 2.3k patients were scanned, researchers found a 3.7% lung cancer detection rate and 100% survival for early-stage cancer.
- Among 4.2k patients, those who got screened had more stage I-II disease (72% vs. 37%) and higher rates of surgery-only treatment (56% vs. 25%) at three years.
- Using PanCan criteria to manage suspicious lung nodules worked better than Lung-RADS in 4.5k people screened, with fewer workup referrals (2.8% vs. 7.4%) and better PPV for high-risk malignancy (48% vs. 18%).
The Takeaway
This is just a selection of the exciting research being presented at WCLC 2024. It seems evident that CT lung screening’s future as a mainstream cancer test is closer than ever.
|
|
|
Memorial MRI’s Choice for Patient Comfort
Texas has one of the highest obesity rates in the US. So to best serve its patients, Memorial MRI & Diagnostic in Houston turned to United Imaging and its 3.0T uMR OMEGA MRI scanner with 75cm ultra-wide-bore. Learn more about their story.
|
|
Best in KLAS for Image Exchange
With over 15k connected facilities and a dedicated outreach team building new connections every day, it’s no wonder Nuance PowerShare Image Sharing was named #1 Best in KLAS 2024 for image exchange. Learn more about this award-winning solution.
|
|
Set a New Standard of Care
Give your patients clarity and peace of mind with Prenuvo’s advanced whole-body MRI scans. Prenuvo’s team works closely with you to help pinpoint potential problems, diagnose faster, and deepen your patient relationships. Learn more today.
|
|
- AI for ICH Falls Short: One use case for AI is detection of intracranial hemorrhage from non-contrast CT scans. But a study in AJR found Aidoc’s AI triage algorithm fell short in 7.4k patients, as radiologists working without AI had the same sensitivity as those using AI (98.6% vs. 98.9%) and slightly higher specificity (99.8% vs. 99.3%). Report turnaround times were largely the same without AI versus with (147 vs. 150 minutes). Aidoc executives noted the study used an older version of the algorithm, but the study has unleashed a storm of questions about the value of AI for radiology.
- AI of LDCT for Osteoporosis: The beauty of low-dose CT lung cancer screening is that the exams can be used for opportunistic screening of other conditions like osteoporosis. In a study in European Radiology, researchers from China used their homegrown AI algorithm to assess bone mineral density in 4.4k scans acquired on systems from three different vendors, finding that compared to gold-standard quantitative CT it had high sensitivity and specificity for osteoporosis (87% and 99%) and osteopenia (91% and 94%).
- DeepHealth Partners with HOPPR: In another sign of the rising prominence of AI foundation models, RadNet’s DeepHealth AI subsidiary has signed a data and development partnership with HOPPR. The partners will use HOPPR’s medical-grade generalized foundation models to develop new fine-tuned AI algorithms for breast, prostate, and lung cancer detection more quickly and efficiently. HOPPR emerged from stealth mode in 2023, while DeepHealth likewise launched at the end of last year to spearhead RadNet’s AI operations.
- AI-Directed Breast Screening Is Cost-Effective: Numerous research studies in the past year have supported AI for breast screening, but how cost-effective is it? U.K. researchers assessed that question in a study in JAMA Network Open, analyzing the monetary benefit of setting screening intervals based on MIT’s open-source Mirai algorithm, which predicts five-year breast cancer risk from negative mammograms. Screening every six years for low-risk, every 2-3 years for below- and above-average risk, and annually for high-risk women would save up to $109.2M – and detect more cancers.
- Can AI Reduce Double-Reading Workload? In a related study in Radiology: Artificial Intelligence, Danish researchers tested different workflow scenarios with Lunit’s Insight MMG algorithm to reduce the need for double-reading breast screening exams. Recall and arbitration rates as well as PPV and NPV varied based on whether AI was used to replace the first or second reader, or was used to triage high- and low-risk cases. Compared to double reading, AI reduced screen reading volume by about 50%, but its impact on accuracy and workload varied based on the scenario.
- HAP Signs PA Practice: Revenue cycle management firm Healthcare Administrative Partners continues to gain clients, signing a new customer in Pittsburgh, Pennsylvania. South Hills Radiology Associates signed on for HAP’s core revenue cycle services, billing, coding, carrier credentialing, business intelligence, and MIPS Measure Assurance Services, as well as comprehensive practice management services.
- RogueRad on Whole-Body MRI Screening: The always-quotable Saurabh Jha, MD, (AKA RogueRad) has a few things to say about whole-body MRI screening in a new editorial in JACR. Jha focuses on the possibility of overdiagnosis – or the detection of indolent disease that may not be a health threat – seeing it not as a failure to diagnose but rather a failure to accurately predict which diseases will be life-threatening. He also questions the economics of whole-body MRI, suggesting that consumers might be better off complying with guidelines for existing screening tests.
- Fewer Cancers Diagnosed During COVID: Nearly 150k cancers may have gone undiagnosed in 2020 and 2021 due to the COVID-19 pandemic. Writing in JAMA Network Open, researchers said cancer incidence rates were 9.4% lower than expected in 2020 and 2.7% lower in 2021, adding up to 149.6k potentially undiagnosed cases. Among screening-detected cancers, only female breast cancer rebounded in 2021, exceeding expected rates by 2.5%. Reductions remained lower that year for lung (-9.1%) and cervical cancer (-4.5%). Early-stage diagnoses in particular were lower.
- RP vs. UnitedHealthcare: The long-running dispute between Radiology Partners and UnitedHealthcare over radiology billing could be coming to an end. The dispute began in 2022 over billing claims involving RP’s Singleton Associates practice in Houston; Texas arbitrators in 2023 awarded Singleton $134.4M, but on review last month an arbitration panel threw out the award, saying Singleton breached elements of the contract. UnitedHealthcare’s claims against RP were also rejected, such that no one at either party will get paid, except maybe their lawyers.
- New Viz Partnerships: Viz.ai signed two partnerships last week, one addressing heart disease and the other lung cancer. The company partnered with Cleerly to bring that company’s AI algorithms for CCTA analysis onto the Viz.ai platform, expanding its functionality to coronary artery disease for the first time. For lung cancer, Viz is partnering with the Addario Lung Cancer Medical Institute (ALCMI) consortium to develop better tools for detecting and working up lung nodules that could indicate lung cancer.
- Imidex Launches Lung Alliance: Chest X-ray AI developer Imidex formed an alliance with diagnostic testing company Orbit Genomics designed to make chest X-ray more effective for lung cancer screening. Patients will get chest X-ray screening that’s analyzed by Imidex’s Visirad XR algorithm, which highlights suspicious nodules and masses; those with potential findings can be referred to Orbit’s OrbiSeq-L DNA-based blood test to confirm the presence of cancer. The firms see the alliance as a lower-cost, more-accessible alternative to CT-based lung screening.
- Nanox Teases Handheld X-Ray: Israeli digital X-ray developer Nanox appears to be developing a handheld digital X-ray device. Details are scarce beyond a snappy artist’s rendition on LinkedIn, but the company said the unit will be called Nanox.CONNECT and will be an ultra-portable solution for providing diagnostic images in settings like disaster zones and military applications. Nanox has been rolling out its cold-cathode multisource X-ray technology since getting FDA clearance in April 2023; it also recently got clearance for its HealthCCSng V2.0 coronary artery calcium software.
- AZmed AI Extended to Kids: AI developer AZmed received 510(k) clearance to market its Rayvolve algorithm for fracture detection on pediatric X-rays. Rayvolve was first cleared for adults in June 2022; to support the new clearance, AZmed worked with imaging center chain SimonMed Imaging to test the algorithm in 3k pediatric radiographs, achieving 96% sensitivity and 86% specificity.
- Update on Digital Health Funding: Silicon Valley Bank’s mid-year healthcare investment report echoed the sentiment of other recent updates: digital health funding is a tale of two startup classes. While Series A digital health startups collected $1.7B across 109 rounds in the first half – with median valuations swelling 16% to $44 million – Series B and C companies closed much smaller sums in extension rounds designed to hold them over until next year when they might be able to raise at higher valuations. Over 28% of first-half healthcare investments were down or flat rounds, the highest total since 2019.
|
|
A Cloud-Native Foundation for Your Imaging Needs
Merge by Merative’s Merge Imaging Suite is built on a cloud-native foundation to address your imaging needs today, providing an intuitive, feature-rich array of solutions to help you enjoy a seamless and frictionless workflow experience. Learn more about it today.
|
|
4 Reasons Why Subscription Is Better
The KLAS-leading Intuition advanced visualization solution from TeraRecon includes all the clinical features you need, plus a growing list of exclusive subscription-only content. Learn more about what’s included.
|
|
Which Imaging Vendors Are Ready for the Cloud?
Which medical imaging vendors are perceived as ready for the cloud? In a new survey from KLAS Research, AGFA HealthCare’s approach to the cloud won high marks from customers. Learn more about the report today.
|
|
- MRI of the Pelvis and Lower Extremities: Join Donald Resnick, MD, and guest faculty in this Virtual Conference series on MRI of the pelvis and lower extremities, hosted by Medality from September 30-October 4. Register your team and save on your group enrollment.
- Are You Getting the Most Out of Your Image Exchange Solution? Ask yourself these 5 questions from PocketHealth to find out whether your enterprise image exchange solution is maximizing patient satisfaction, efficiency, and driving revenue.
- AI for Lung Cancer Diagnosis and Screening: Check out this comprehensive new eBook from Calantic by Bayer on the role of AI in lung cancer diagnosis and screening. It explores AI’s potential role in improving lung cancer screening strategies, identifying high-risk individuals, and enhancing diagnostic accuracy. Download it today.
- The Clinical Value of Soft-Tissue Chest X-Ray: Soft-tissue techniques can improve the visibility and accuracy of chest X-ray. Learn about two important soft-tissue methods – bone suppression and dual-energy subtraction – in this white paper from Riverain Technologies.
- Looking to Get Started with AI? Learn how Milton Keynes University Hospital adopted a successful AI strategy in this on-demand webinar from Blackford. You’ll discover how to evaluate and select the best AI solutions for your facility, tips for getting started with AI, and more.
- Breast Density Classification: From Conception to Routine Use: Knowing an individual’s mammographic breast density is key to understanding their risk for cancer. An article from Visage Imaging explains how an AI-based solution can improve breast density assessment.
- AI for Detecting Post-Traumatic Bone Fractures: Check out this new research paper to learn how Gleamer’s BoneView AI solution enhanced radiologists’ ability to detect post-traumatic bone fractures by identifying cases without abnormalities, translating into improved patient care.
- Seamless, Connected Healthcare: Clearpath is committed to facilitating seamless, continued healthcare by optimizing automation and workflows between patients and providers. Learn about the company’s solutions and how they contribute to a healthcare environment where every step of the patient journey is connected and efficient.
- Preparing for the Future of Enterprise Imaging: Check out this white paper from Optum to learn what you need to know when moving your enterprise imaging to the cloud. Learn how to assess the various approaches and develop a strategy that works for you.
- The UK’s Lung Cancer Screening Rollout: The UK has launched a targeted lung cancer screening program to improve lung cancer outcomes through earlier detection. Learn about DeepHealth’s Saige Lung and how it’s the preferred AI solution in this case study.
- Fully Automated AI Echo vs. 3D and Human Readers: While 3D echo is becoming more accurate, 2D still dominates clinical care. A new study evaluates agreement in measures of LV volume and function between human readers, echo AI from Us2.ai, and the 3D Heart Model.
- Building a Mobile Lung Cancer CT Screening Program: The number of patients eligible for low-dose CT lung cancer screening has expanded, and so has the need to reach at-risk patients closer to where they live. That’s why Siemens Healthineers’ Mobile Lung Screening Solution combines the quality, ease of use, and flexibility needed to create a program that meets the real-life needs of your community.
- MRI Access and the Cost of Inpatient Stays: Longer inpatient stays due to delayed MRI access are a long-standing and costly issue for hospital systems. Find out how STAGE from SpinTech MRI can reduce your MRI backlog and inpatient stays by shortening brain scan times by 30%.
|
|
|
|
|
