|
AI Detects Interval Cancer | Density Deadline
August 29, 2024
|
|
|
|
|
Together with
|
|
|
|
“Fancy-sounding techniques do not matter when you don’t incorporate clinical domain expertise in your medical AI research and publish results based on inadequate evaluation metrics.”
|
|
Woojin Kim, MD, on a recent arXiv paper on LLMs and radiology report generation.
|
|
|
|
In yet another demonstration of AI’s potential to improve mammography screening, a new study in Radiology shows that Lunit’s Insight MMG algorithm detected nearly a quarter of interval cancers missed by radiologists on regular breast screening exams.
Breast screening is one of healthcare’s most challenging cancer screening exams, and for decades has been under attack by skeptics who question its life-saving benefit relative to “harms” like false-positive biopsies.
- But AI has the potential to change the cost-benefit equation by detecting a higher percentage of early-stage cancers and improving breast cancer survival rates.
Indeed, 2024 has been a watershed year for mammography AI.
U.K. researchers used Insight MMG (also used in the BreastScreen Norway trial) to analyze 2.1k screening mammograms, of which 25% were interval cancers (cancers occurring between screening rounds) and the rest normal.
- The AI algorithm generates risk scores from 0-100, with higher scores indicating likelihood of malignancy, and this study was set at a 96% specificity threshold, equivalent to the average 4% recall rate in the U.K. national breast screening program.
In analyzing the results, researchers found …
- AI flagged 24% of the interval cancers and correctly localized 77%.
- AI localized a higher proportion of node-positive than node-negative cancers (24% vs. 16%).
- Invasive tumors had higher median risk scores than noninvasive (62 vs. 33), with median scores of 26 for normal mammograms.
Researchers also tested AI at a lower specificity threshold of 90%.
- AI detected more interval cancers at this level, but in real-world practice this would bump up recall rates.
It’s also worth noting that Insight MMG is designed for the analysis of 2D digital mammography, which is more common in Europe than DBT.
- For the U.S., Lunit is emphasizing its recently cleared Insight DBT algorithm, which may perform differently.
The Takeaway
As with the MASAI and BreastScreen Norway results, the new study points to an exciting role for AI in making mammography screening more accurate with less drain on radiologist resources. But as with those studies, the new results must be interpreted against Europe’s double-reading paradigm, which differs from the single-reading protocol used in the U.S.
|
|
|
Go Virtual with Ease
Visage Imaging’s new Visage Ease VP solution makes it easy for you to take advantage of the unique clinical capabilities of the new Apple Vision Pro spatial computing environment. Learn how it can change your practice today.
|
|
Build Patient Engagement into Your Enterprise Image Exchange
87% of healthcare imaging leaders identify rising operating costs as a major challenge. This white paper from PocketHealth outlines opportunities to generate cost savings, retain patients, and increase referral volumes. Learn more here.
|
|
- Density Reporting Deadline: The deadline for mammography facilities to start reporting patient breast density is September 10, and the FDA has published guidelines to help mammography facilities comply. Mammography centers must give patients reports that assess their breast tissue density, as well as any recommendations made to healthcare providers if applicable. Reports must be given to patients within seven calendar days of final interpretation for exams assessed as “suspicious” or “highly suggestive of malignancy.” Some great info and compliance tips are available at DenseBreast-Info.org.
- Adding MRI to Mammography: If a woman does have dense breast tissue, what imaging tools should be used? Researchers writing in a modeling study in JAMA Internal Medicine examined 60 different breast screening strategies, concluding that biennial DBT screening starting from 50-74 with supplemental MRI for women with dense breast tissue could offer the best mix of boosting cancer detection while reducing harms from false positives. Biennial DBT screening averted 7.4 breast cancer deaths per 1k women with 151 false-positive recalls; adding MRI averted 8.0 deaths with 343 recalls.
- Study Supports Annual Mammography: Meanwhile, a study in Journal of Clinical Oncology found that annual breast screening saved the most lives over less-frequent intervals. Researchers tracked 8.1k women, finding that those who were screened annually had the lowest rate of late-stage cancers (9%) compared to those screened biennially (14%) and intermittently (19%). Death rates relative to annual screening were also worse for screening biennially (HR=1.42) and intermittently (HR=2.69). Researchers said the findings support annual screening for women 40 and older.
- Does AI Lack Proof? Do AI approvals lack data backing their safety and effectiveness? A new study in Nature Medicine questions the FDA approval process for AI algorithms, claiming 43% of over 521 AI-based regulatory authorizations they reviewed lacked clinical validation data. Researchers criticized the FDA’s practice of letting vendors use retrospective rather than prospective data for regulatory submissions, and said their study raised questions about AI safety. But other investigations have found AI products actually have fewer post-market event reports than general medical devices.
- Prenuvo Sets Up at Canadian Border: Whole-body scanning provider Prenuvo has set up a mobile MRI coach at the U.S.-Canada border in Buffalo, New York. The company said the move is in response to a waitlist of over 10k Canadians who want access to Prenuvo scans; the company had plans to expand to Toronto but has faced “significant barriers” due to Ontario’s restrictions on private imaging services. Like much of Canada, Ontario has struggled to meet growing demand for advanced imaging scans.
- Fetal Ultrasound Politics: In a sign of today’s politically charged times, a coalition of physician groups including RSNA, ACR, AIUM, and others updated the clinical terminology to be used for fetal ultrasound reporting. Stating the old language “could be used by third parties to negatively affect the physician-patient relationship,” the new terminology retires terms like “heart motion” or “heartbeat” in favor of “cardiac activity,” and eliminates terms like “live,” “living,” or “viable.” State laws banning abortion have had a chilling effect on many women’s health physicians.
- Will DEI Blowback Undo Radiology Diversity? Will a raft of laws seeking to roll back diversity, equity, and inclusion (DEI) programs undo efforts to make radiology more diverse? In a new study in JACR, researchers tracked 85 anti-DEI bills introduced in U.S. state legislatures from December 2022 to May 2024, finding that while only 16% have been signed into law, they could affect radiologist employment. Academic radiology programs are most at risk, as some anti-DEI laws enable state funding cuts and even employee termination for those in violation.
- Racial Differences in Diagnostic Testing: Meanwhile, a new study in JAMA Network Open found racial disparities in the rates at which patients received acute-care diagnostic testing for common conditions like chest and abdominal pain. Among 3.7M patient ED encounters, Black patients with a “nonspecific diagnosis of interest” were less likely to have diagnostic testing (including imaging exams) than White patients (OR=0.74). Researchers said the findings suggested test overuse among White patients, but acknowledged Black patients have historically experienced underuse as well.
- Siemens to Buy Novartis Radiopharma Biz: In a move to expand its European radiopharmacy operations, Siemens Healthineers plans to buy the diagnostic side of Novartis’ Advanced Accelerator Applications business. Novartis bought AAA for $3.9B in January 2018; it offers both diagnostic PET radiotracers like fluorine-18 FDG and neuroendocrine therapy agent Lutathera. A Financial Times story valued the deal at $224M and said AAA runs Europe’s second-largest cyclotron network with 14 sites; Siemens claims its PETNET Solutions business is the world’s largest radiopharmacy cyclotron network. Lutathera would remain with Novartis.
- PET Agent on FDA’s Fast Track: Radiopharmaceutical developer Life Molecular Imaging has received FDA fast track designation for a PET imaging tracer targeting tau neurofibrillary tangles connected to Alzheimer’s disease. The company’s fluorine-18 PI-2620 tracer is in phase 3 clinical studies for detecting Alzheimer’s as well as progressive supranuclear palsy and corticobasal degeneration, and is being developed through a collaboration between Life Molecular and AC Immune.
- Telix Submits NDA for Brain PET: Another PET tracer recently granted fast-track review status is moving ahead. Telix Pharmaceuticals said it has submitted a new drug application to the FDA for Pixclara (TLX101-CDx), its fluorine-18 floretyrosine (FET) agent for imaging glioma (brain cancer). Patients with glioma have low survival rates, making early diagnosis and accurate treatment monitoring important. Telix hopes to make Pixclara commercially available in 2025.
- Concerns about PET Policy: Once all these PET agents get to market, they’ll need to contend with a changing reimbursement environment. CMS recently proposed changing its policy to unbundle PET tracer payments over $630 per day from imaging exams, and SNMMI recently offered comments, including the suggestion of a lower $550 threshold. The group also offered comments on a proposed FDA policy that would require stability testing of PET drugs at each facility annually – SNMMI thinks it’s a bad idea as it could raise costs and affect tracer supplies.
- GE Touts Cardiac Imaging CE Marks: With the European Society of Cardiology 2024 about to begin, GE HealthCare announced European regulatory authorizations for several cardiac imaging products. CE Marks have been awarded to the Vscan Air SL wireless ultrasound scanner with Caption AI, as well as ECG-less cardiac CT scanning on the Revolution Apex platform. ECG-less CT enables acquisition of cardiac images without the patient’s ECG trace, which supports faster scanning in acute situations.
|
|
How to Standardize CT Images
The quality and appearance of CT scans can vary considerably. In this white paper from Riverain Technologies, find out how image normalization can standardize CT images, making them easier to analyze and interpret.
|
|
Women’s Health and AI
With new breast density reporting requirements coming in the U.S., there’s a need for objective density assessment and reporting. Check out this article from Blackford Analysis on the AI solutions that can help.
|
|
- How McFarland Clinic Reduced Downtime: McFarland Clinic Health Ventures recently experienced a downtime occurrence due to an HL7 interface problem. But thanks to a suite of medical imaging solutions from Merge by Merative, McFarland’s downtime lasted all of 15 minutes. Find out how they did it in this case study.
- Serving the Needs of Image-Driven Clinicians: TeraRecon serves the needs of image-driven clinicians and users across the entire healthcare enterprise. Schedule a demo today to explore the latest updates to industry-leading advanced visualization solutions and learn about new AI capabilities.
- Treating Local Patients Like Family: Holzer Health System in the rural community of Jackson, Ohio, treats local patients like family, offering the same level of top-quality care that can be found anywhere. Find out how United Imaging’s Software Upgrades for Life program helps them stay current.
- Presenting Unboxing AI: Check out CARPL’s video series, Unboxing AI, featuring experts discussing AI and its future in radiology. The next episode on August 30 features Avi Sharma, MD, of Einstein Healthcare Network – reserve your seat today.
- 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.
- 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.
- Get to Know DeepHealth: What’s the latest from DeepHealth? In this episode of the Imaging Wire Show, we talked to COO/CTO Sham Sokka about the company’s recent launch and their take on the value AI provides to radiology, especially for screening.
- Fully Automated AI for Mitral Regurgitation Grading: Echo AI can help address challenges with mitral regurgitation assessment quality and efficiency. Find out how researchers used Us2.ai’s fully automated echo workflow for grading MR severity and predicting one-year mortality.
- Building a Theranostics Program: Learn how 210 PET Imaging of Cary, North Carolina became one of the first non-academic outpatient imaging centers in the North Carolina Research Triangle to build a theranostics program in this case study from Siemens Healthineers.
- Next-Generation Enterprise Imaging: US healthcare organizations are making a definitive pivot to cloud technology. In this white paper from AGFA HealthCare and HIMSS, learn about the benefits of cloud-based enterprise imaging. Download the white paper today.
- A Personalized Reading Experience: The ability to personalize the digital imaging workspace can help radiologists make quicker decisions and deliver faster diagnoses to the patients who need it most. Learn how Intelerad’s InteleViewer allows you to tailor your reading environment to your preferences.
- Start at the Source to Improve MRI: Looking for ways to improve MRI speed and image quality while addressing broader concerns in healthcare? The answer may lie in proven MRI physics in your existing scanner – learn how to unlock it with STAGE from SpinTech MRI.
- Building a Brand in Radiology: Check out the next episode of Medality’s Radiology Report Podcast featuring Toronto Radiology’s Anish Kirpalani, MD, who talks to host Daniel Arnold about challenges and strategies in radiology, such as proactive recruitment and the importance of building a strong brand.
- Radiology Academy – Medical Education on Demand: Visit Calantic Radiology Academy by Bayer, where you’ll find the latest keynotes and symposium sessions on the use of artificial intelligence in radiology, ranging from challenges facing AI to bias in machine learning.
- Automated MSK Measurements: BoneMetrics from Gleamer is an automated AI solution that can help diagnose scoliosis by automating Cobb angle calculations on images acquired with X-ray and EOS systems. Learn more about how it works today.
- Image Sharing to Improve Patient Outcomes: In this case study, learn how AdventHealth improved treatment planning and delivery because of Nuance PowerShare’s immediate connection to other sites in the network for medical image sharing.
|
|
|
|
|
