“The enemy is closer than you think.”
One radiologist’s warning to his peers after a mammography AI solution detected breast cancer 1 to 2 years before five human radiologists.
Imaging Wire Sponsors
- Arterys – Reinventing imaging so you can practice better and faster.
- Bayer Radiology – Providing a portfolio of radiology products, solutions, and services that enable radiologists to get the clear answers they need.
- GE Healthcare – Enabling clinicians to make faster, more informed decisions through intelligent devices, data analytics, applications and services.
- Healthcare Administrative Partners – Empowering radiology groups through expert revenue cycle management, clinical analytics, practice support, and specialized coding.
- Hitachi Healthcare Americas – Delivering best in class medical imaging technologies and value-based reporting.
- Nuance – AI and cloud-powered technology solutions to help radiologists stay focused, move quickly, and work smarter.
- Riverain Technologies – Offering artificial intelligence tools dedicated to the early, efficient detection of lung disease.
- Siemens Healthineers – Shaping the digital transformation of imaging to improve patient care.
- Zebra Medical Vision – Transforming patient care with the power of AI.
The Imaging Wire
A University of Saskatchewan team developed a hybrid lung cancer screening method that uses AI risk scores to refine Lung-RADS classifications and could help reduce both false positives and follow-up costs. Here are some details:
- The Method – The new method uses AI-based malignancy risk scores to confirm or adjust radiologist-generated Lung-RADS classifications. With this strategy, patients with initial Lung-RADS categories of 1 or 2 and high AI risk scores are upgraded to Lung-RADS category 3 (requiring follow-up), while patients with initial Lung-RADS categories of 3 or 4 and low AI risk scores are downgraded to Lung-RADS category 2 (no follow-up required).
- The Study – The team applied the AI + Lung-RADS method to 192 baseline low-dose CT studies. They then weighted their results to represent a 3,197-patient cohort, comparing accuracy, follow-ups, and related costs.
- The Results – The AI-refined Lung-RADS scores achieved the same sensitivity as Lung-RADS-only scores from 6 radiologists (both 91%), while posting far high specificity (96% vs. 66%). As a result, the AI + Lung-RADS method prompted far more Lung-RADS downgrades (30% of 3-4 scores downgraded to 2) than upgrades (0.2% of 1-2 scores upgraded to 3), eliminating $72 to $242 in follow-up costs for every patient.
- The Takeaway – This seems like an AI approach that addresses a real world problem (LDCT screening’s high false positives and downstream costs) and similar AI-refinement methods could likely be applied to other radiology scoring systems.
- COVID’s Case Complexity Impact: A new study out of Mount Sinai Hospital revealed that COVID-related imaging volume declines had the greatest impact on complex / high-wRVU exams. A review of the healthcare system’s radiology studies during the first 33 weeks of 2020 (weeks 1-9 = pre-surge, 10-19 = surge, 20-33 = post-surge) revealed that wRVUs fell at a much faster rate than overall imaging volumes during COVID’s “surge” period (-69% vs. -57%). These complexity declines were greatest with radiography (-14.7%), cardiothoracic imaging (-16.2%), and at community hospitals (-15.9%), while complexity increased for breast imaging (+6.5%), interventional procedures (+5.5%), and outpatient imaging (+12.1%).
- Seno’s OA/UA FDA: Seno Medical announced the FDA approval of its Imagio Breast Imaging System, which combines ultrasound and opto-acoustic technology to help physicians differentiate between benign and malignant breast lesions. The Imagio system’s opto-acoustic images show a blood map in and around breast masses, while its ultrasound provides a traditional anatomical image, allowing physicians to inspect masses for angiogenesis and deoxygenation (both cancer indicators).
- AAC Imaging: A new JAHA study revealed that patients with abdominal aortic calcifications (AAC, captured with imaging) have higher cardiovascular risks, revealing another way imaging can support preventative cardiac care. The researchers reviewed 52 studies (36,092 participants), finding that patients with AAC or both AAC and kidney disease have much higher risks of cardiovascular events (1.83x & 3.47x higher risk) and all‐cause mortality (1.98x & 2.4x higher risk) than their peers.
- CMS Speeds Medicare Coverage: CMS rolled out its new Medicare Coverage of Innovative Technology rule (MCIT), creating an accelerated Medicare coverage pathway for products that the FDA defines as “breakthrough.” These products can gain Medicare nationwide coverage at the same time as their FDA approval (vs. lengthy delays and sometimes regional coverage). The MCIT approvals are valid for up to four years, followed by a formal CMS review to determine more permanent coverage.
- Pre-CT Triage Cuts Unnecessary Imaging: When Boston Medical Center adopted a clinical triage algorithm to determine whether patients with blunt head / neck trauma require CT imaging, BMC’s head / neck trauma CT volumes fell significantly. That’s from a review of 9k ≥15yr-old patients who were admitted at BMC with head / neck trauma between 2007 and 2013 (n= 4,030 before adopting triage algorithm, 4,969 after). The study revealed major declines in the percent of these patients who received head CTs (64.2% vs. 55.8%), cervical spine CTs (60.5% vs. 49.4%), but not head/neck CT angiography (9.8% vs. 9.7%).
- Solis’ D.C. Acquisition: Major independent breast imaging group, Solis Mammography (85 national imaging centers), expanded its Washington D.C. area presence with its acquisition of Progressive Radiology (D.C.’s largest rad-owned practice). Progressive Radiology will become part of Solis Mammography’s Washington Radiology business, while expanding the group’s focus beyond breast imaging (Progressive also specializes in MSK/sports imaging, neuroradiology, and pediatric imaging).
- Lung Incidentals & Low-Value Diagnostics: A JAMA study shared new evidence supporting less-intensive diagnostic evaluation methods for incidentally-detected lung nodules (e.g. fewer exams, less-invasive exams). The researchers found that patients who received less-intensive evaluations had the same stage III or IV lung cancer diagnosis rates as patients who were examined with guideline-concordant or more-intensive methods. However, the less-intensive and more-intensive evaluation methods had far different radiation exposure levels (−9.5 vs. +6.8 milliSieverts), median costs (−$10,916 vs. +$20,132), and procedural complication risks (−5.9% vs. +8.1%) compared to guideline-concordant evaluations.
- Promaxo’s $4M: Point-of-care MRI company Promaxo closed a $4.17M Huami-led investment round that it will use to accelerate the AI part of its product strategy. Promaxo’s forthcoming MRI platform will support guided interventions in different locations (e.g. physician offices) and at lower costs than we typically associate with MRI systems. The strategic investment also brings Huami deeper into the healthcare technology arena, following its partnership with Aspen Imaging last year.
- Visage Signs Intermountain: Visage Imaging added Intermountain Healthcare to its growing list of major enterprise imaging clients, announcing a 7-year, $40M contract that will extend its Visage 7 Viewer and Open Archive products across Intermountain’s hospital locations and departments. The deal adds to Visage’s impressive list of recent major hospital system clients (MedStar, NYU, Northwestern, OSU, Duke Health, Partners, Mayo Clinic, Mercy Health, and Yale-New Haven). It should also be noted that Visage’s Intermountain and MedStar Health implementations (its two most recent deals) will both be deployed on Google Cloud, potentially making them the largest PACS deals to be fully deployed in the public cloud.
- ImageNet Reality Check: Even though deep learning-based chest X-ray models are typically pre-trained on ImageNet models, a new Stanford AI Lab study just debunked the assumption that using ImageNet to improve model architecture actually results in better medical imaging task performance. There’s a lot to this study and it has some buzz in the AI community, so check it out if you haven’t already.
- COVID’s CT Impact: A new survey from IMV Medical revealed that the COVID-19 pandemic drove a 20% reduction in U.S. CT exam volumes in 2020 (to ~73m exams) largely due to a drop in electives and outpatient no-shows / cancelations. Interestingly, 13% of all CT exams were performed to assess / manage COVID-19 (17% of inpatient and emergency CTs).
- RSIP Vision’s Segmentation & Measurement Module: RSIP Vision launched a new multi-modality segmentation and measurement tool used to automate the detection objects of interest and their boundaries. The new AI-based module is available to medical device manufacturers who would integrate it into their medical device software.
- RadClip: A Case Western Reserve-led research team developed a nomogram that can predict prostate cancer recurrence before surgery by analyzing patients’ multiparametric MRI radiomics features and clinical data. The team developed the RadClip nomogram with data from 71 patients who underwent MRI exams before radical prostatectomy procedures and then tested it against a separate 127-patient dataset, finding that RadClip predicted biochemical recurrence-free survival and adverse pathology more accurately than three standard prostate cancer risk prediction models. With these results, the researchers suggested that RadClip could support prostatectomy planning and guide post-prostatectomy therapy.
The Resource Wire
– This is sponsored content.
- Watch this recorded webinar from Healthcare Administrative Partners where they examine how the 2021 Medicare Physician Fee Schedule (MPFS) and Quality Payment Program (QPP) final rules will impact radiologists.
- In this Nuance Q&A, Inference Analytics CEO Farrukh Khan discusses how its NLP analytics AI solution improves radiologists’ reporting productivity through the PowerScribe One platform.
- In this Hitachi blog, VidiStar users share how they’ve benefitted from the cardiovascular information system’s flexible SaaS-based pricing model and leveraged its productivity advantages to increase reimbursements.
- Check out Riverain Technologies’ on-demand webinar demonstrating how its AI solutions integrated into LucidHealth’s radiology workflow and sharing best practices on how to combine AI with radiologist expertise.
- Catch Arterys CEO, John Axerio-Cilies, on the Spill the Beans podcast where he discusses integrating AI models into the radiologist cockpit.
- Check out GE Healthcare’s upcoming webinar where a panel of breast health leaders discuss how AI is changing the game in breast imaging across modalities.
- Siemens Healthineers detailed the six steps that radiology practices can follow to build resiliency, providing a roadmap detailing the workflow, leadership, and financial changes required to successfully respond to the pandemic and come out stronger because of it.
- Learn how Windsong Radiology Group used Bayer’s MEDRAD Stellant FLEX CT injection system to drive cost efficiencies and standardization across its imaging centers.
- Learn how Zebra-Med’s AI Coronary Calcium solution can serve as a coronary artery disease ‘early-warning’ system using existing CT scans.