COVID-19 Imaging | Body CT’s AI Predictions

“There’s a difference between AI projects and products, and there’s a difference between AI products and companies.”

CureMetrix president, Kevin Harris, on how imaging AI is a lot more diverse than it’s often portrayed.


Imaging Wire Sponsors

  • Focused Ultrasound Foundation – Accelerating the development and adoption of focused ultrasound.
  • GE Healthcare – Providing point of care ultrasound systems, from pocket-sized to portable consoles, designed to support your clinical needs and grow along with your practice.
  • Healthcare Administrative Partners – Empowering radiology groups through expert revenue cycle management, clinical analytics, practice support, and specialized coding.
  • Nuance – AI and cloud-powered technology solutions to help radiologists stay focused, move quickly, and work smarter.
  • Qure.ai – Making healthcare more accessible by applying deep learning to radiology imaging .
  • Riverain Technologies – Offering artificial intelligence tools dedicated to the early, efficient detection of lung disease.

The Imaging Wire



COVID-19 Imaging

The threat of coronavirus/COVID-19 to the world’s health and economy continued to escalate last week, and although this story is WAY bigger than radiology, imaging once again played a key role in the latest batch of COVID-19 news.

  • Low Misdiagnosis – Another report out of China found that CT-based COVID-19 evaluations have low misdiagnosis rates, with just 2 out of 51 COVID-19 cases misdiagnosed as common infection.
  • SARS & MERS OverlapAn AJR study found that COVID-19’s imaging features “significantly overlap” with SARS and MERS scans (often with subpleural and peripheral areas of ground-glass opacity and consolidation). However, 75% of COVID-19 scans show bilateral lung involvement, while SARS & MERS imaging abnormalities are usually unilateral.
  • SARS Modeling – A Singapore hospital’s experience responding to the 2003 SARS outbreak gave its radiology department an operational head-start for its response to COVID-19. The hospital is now paying it forward by sharing a list of 43 operational considerations that other imaging departments can follow as they prepare for (or care for) a flow of COVID-19 patients. The overall goals of these guidelines are to ensure proper patient treatment (e.g. quickly scan patients & interpret scans), operational efficiency (e.g. integrate radiology dept with outbreak response teams), and to keep the virus from spreading within the hospital (e.g. use & isolate dedicated scanners).
  • The COVID-19 Business – Although it’s clear that coronavirus is bad for the economy, it continues to create opportunities for imaging manufacturers. The latest examples come from Philips (“dozens” of extra CT scanner orders in China, increased demand for other diagnostic and respiratory devices), Micro-X ($1m in additional orders for its Carestream mobile X-ray), and Neusoft (delivered 30 scanners in 15 days).



Body CT AI Predictions

AI may add predictive value to the millions of body CT scans that are performed each year. That’s from a University of Wisconsin study that found CT-based body composition biomarkers could be used to predict major cardiovascular events or even survival.

  • The Study – The retrospective study used AI to analyze body CT scans from 9,223 healthy adults and quantify a range of biomarkers (aortic calcification, muscle density, visceral to subcutaneous fat ratio, liver fat, and bone mineral density). The researchers then compared the CT analysis to standard predictive measures (FRS and BMI).
  • Results – The CT + AI technique more accurately predicted future serious cardiovascular events and overall survival. The five CT biomarkers predicted the patients’ 5-year survival with AUCs ranging from 0.619 to 0.743 (vs. 0.499 for BMI, 0.688 for FRS), while a combination of select biomarkers achieved even greater accuracy (aortic calcification, muscle density, liver density, and visceral to subcutaneous fat ratio = 0.789 5yr AUC).
  • Takeaway – The study authors position this approach as a new way to identify and treat high-risk patients before symptoms emerge.

The Wire

  • Breast Radiologists Are Burned Out: Burnout is particularly severe among breast imaging radiologists. That’s from a new study in The Journal of Breast Imaging (n = survey of 370 breast imaging radiologists) that revealed 78.4% of the specialists (290) were “highly burned out” within at least one burnout metric (54% w/ two metrics, 7.5% w/ all three metrics). Younger breast radiologists experienced the worst burnout rates (83.2% of 36-45yr-olds were highly burned out for at least one metric).
  • Sectra and UMC Utrecht’s AI Marketplace: Sectra and the University Medical Center Utrecht (UMC Utrecht) announced a collaboration intended to develop a global AI application delivery platform. Although they didn’t actually call it a “marketplace,” the platform would allow healthcare providers to integrate AI apps from “collaborating vendors” into existing clinical workflows, leveraging technical infrastructure developed by UMC Utrecht and Sectra’s enterprise imaging presence.
  • The Great UK Horse Imaging Scam: A pair of UK scammers were busted for selling a thermal imaging scanner intended for horses as a human breast cancer detection system that they claimed could spot breast abnormalities “eight to 20 years before mammograms.” Although the device was called the “Vet 2000” and openly positioned as a veterinary device that also happened to work with humans, the pair was still able to sell the system to at least three aspiring imaging providers for over $25k a pop.
  • Combined Biopsies More Effective: Performing combined prostate cancer biopsies (both MRI-targeted & US-guided systematic biopsies) is more effective than either of the biopsy methods on their own. This is from a NEJM study that performed combined biopsies on 2,103 men, leading to cancer diagnosis in 1,312 men (vs. 1,104 w/ systemic & 1,084 w/ MRI). The combined method also led to far fewer post-surgery upgrades (3.5%) than MRI-targeted biopsy (8.7%) and US-guided systematic biopsy (16.8%) on their own.
  • Philips Digitalizes: Philips announced the launch of its HealthSuite System of Engagement, positioning the new set of tools as a way for providers to access/integrate their healthcare data from disparate systems and use this data to digitalize their workflows. The HealthSuite System of Engagement brings a number of new tools including new cloud storage and interoperability features for DICOM data and an automated de-identification services to support the use of patient data in research.
  • Lawsuit Reinstated: A Chicago area appeals court reinstated a wrongful death lawsuit against a local radiologist and hospital due to a missed brain bleed. Here’s a recap of this pretty unusual case: 1. The woman was admitted with a severe headache; 2. The sued telerad found her CT to be normal and the hospital discharged her; 3. An attending rad later reviewed the scan, found her hemorrhage, and asked the woman to return; 4. The woman refused and went to different hospital; 5. The second hospital also missed the brain bleed, but the radiology report provided by the first hospital didn’t include her images; 6. The woman unfortunately died three days later; 7. A circuit court dismissed the case; and 8. Now the appeals court is overturning that ruling.
  • PET Cardiac Transplant Predictions: Physicians can predict the success of cardiac transplants by using PET imaging to quantify myocardial blood flow (MBF) and myocardial flow reserve (MFR). That’s from a recent study in The Journal of Nuclear Medicine (n = 99 cardiac transplant patients, 26 who died) that found that stress MBF, uncorrected MFR, and corrected MFR all helped identify patients with significant cardiac allograft vasculopathy (AUCs: 0.639, 0.724, 0.748), while uncorrected MFR most accurately predicted mortality.
  • Another Step Away from Shielding: The “patient shielding is unnecessary” movement gained more momentum, as a joint report from a number of influential UK radiology groups declared that the longstanding practice isn’t necessary and called for its discontinuation. The UK report is the latest example of the industry’s apparent step away from shielding, following a major NYT story in January and a series of studies arguing against the practice in recent years.
  • More Marrow Biopsies: A new JACR study revealed that radiologists significantly increased their share of bone marrow biopsy procedures performed in the U.S. between 2005 and 2016 (from 4.1% to 16.2%, or a 12.43% CAGR) due in part to the increased popularity of image-guided procedures. Although internal medicine physicians still perform the most BMBs, their procedure share plummeted from 60.2% to 36.6% during the same period.
  • Siemens’ Prostate Cancer Pathway: Siemens Healthineers announced the European approval of its AI-Pathway Companion Prostate Cancer system, a clinical decision support “companion” that aggregates, correlates, and visualizes patient-specific information at all stages along the prostate cancer pathway. Siemens’ AI-Pathway Companion and AI-Rad Companion platforms play key roles in the company’s AI strategy, making this long-awaited CE Mark approval quite notable.
  • Signify on TeraRecon: Signify Research shared its take on SymphonyAI Group’s acquisition of TeraRecon, suggesting that even though TeraRecon says it “will continue business as usual,” the company will benefit from increased financial backing that could allow it to have a much greater impact on the imaging market. More specifically, the acquisition could: 1. Allow TeraRecon and other SymphonyAI companies (e.g. Concerto HealthAI) to take advantage of new synergies; 2. Fund the development of new TeraRecon products, such as ‘AI Suite’ solutions; 3. Help TeraRecon protect its IP; and 4. Strengthen TeraRecon’s core AV business.
  • Surprise Billing Lobbying: The ACR and other major physician groups continued their efforts to influence Congress’ ongoing surprise billing legislation, encouraging the lawmakers to adopt arbitration as the primary way to settle surprise billing disputes rather using median in network rates. The letter referenced the impact of California’s surprise billing law, suggesting that the state’s adoption of median rates led to practice consolidation, contributed to healthcare costs doubling, and made anesthesiologists leave to other states.

The Resource Wire

– This is sponsored content.

  • This Nature Research study found that Qure.ai’s qXR tool can identify TB-associated abnormalities in chest radiographs with over 95% accuracy, potentially eliminating 2 out of 3 Xpert MTB/RIF tests.
  • In its latest Q&A, Nuance sat down with HealthLytix to understand the company’s journey and reveal its vision for how AI/ML-based solutions will improve the speed/accuracy of MR and CT imaging and reporting.
  • Healthcare Administrative Partners understands the issues facing your radiology practice. Visit HAP Booth #406 at RBMA PaRADigm to pick up a flash drive with educational videos discussing 2020 MPFS updates, surprise billing, and physician recruitment.
  • In this clip from The Agenda, University of Toronto brain surgeon Andres Lozano shares how deep-brain stimulation and high-intensity focused ultrasound are progressing as potential Alzheimer’s, Parkinson’s, and Huntington’s disease treatments.
  • This GE Healthcare white paper details how its suite of point of care ultrasound AI tools simplify complex patient assessments, enable faster clinical decisions, and calculate precise results.

Leave a Reply

Basic HTML is allowed. Your email address will not be published.

Subscribe to this comment feed via RSS

Get every issue of The Imaging Wire

Delivered right to your inbox. We will never sell or share your information.