DBT’s Marketing Juggernaut | Two-for-One PET | DL for TB

“It’s incredibly troubling. Everyone has a different stake in all this, and it all seems to be tied to financial gain.”

National Breast Cancer Coalition president, Fran Visco, on how the breast imaging industry drove the rapid rise of DBT screening.



Imaging Wire Sponsors

  • Carestream – Focused on delivering innovation that is life changing – for patients, customers, employees, communities and other stakeholders
  • Focused Ultrasound Foundation – Accelerating the development and adoption of focused ultrasound
  • Medmo – Helping underinsured Americans save on medical scans by connecting them to imaging providers with unfilled schedule time
  • Nuance – AI and cloud-powered technology solutions to help radiologists stay focused, move quickly, and work smarter
  • Pocus Systems – A new Point of Care Ultrasound startup, combining a team of POCUS veterans with next-generation genuine AI technology to disrupt the industry
  • Qure.ai – Making healthcare more accessible by applying deep learning to radiology imaging

The Imaging Wire



DBT’s Marketing Juggernaut

Kaiser Health News detailed the multi-million dollar marketing effort that drove the rapid adoption of 3D mammography in the U.S., even though DBT scans still “haven’t been shown to be more effective than traditional mammograms.” Here are the details:

Follow the Money – KHN found that DBT manufacturers have influenced “policy, public opinion and patient care” with the help of millions in marketing. This influence followed the familiar steps of marketing to doctors/hospitals by funding research and paying for speaking / consulting / dining ($9.2m in last 6yrs), direct-to-consumer marketing ($14m in last 4yrs, not including social media), and state and payor lobbying through advocacy group funding (an unspecified amount). The article provided very specific examples of how some of DBT’s biggest advocates have financial ties to DBT manufacturers, while positioning Hologic as a driving force behind much of these efforts.

Rapid Adoption – As a result of this funding (not to mention some decent evidence supporting DBT), 95% of insured women in the U.S. are now covered for 3D mammography and DBT systems are now in-use at 63% of breast imaging facilities. This is quite an accomplishment given that DBT has only been approved since 2011.

The Costs and Payoffs – Adding $50 to a typical mammogram, DBT has undoubtedly increased screening costs, including a $230 million total increase in Medicare screening costs between 2015 and 2017. Meanwhile, well over half of all breast imaging facilities in the U.S. have upgraded to DBT (usually a >$300k purchase) over the last nine years.

The Takeaway – Many may disagree with KHN’s argument that DBT is unproven, but it’s harder to debate its evidence that industry money helped 3D mammography quickly become mainstream, and it’s quite possible that DBT adoption actually did outpace the evidence that it would be worth its costs.



Two-for-One PET

Stanford researchers developed a “two-for-one” PET tracer that can identify multiple types of cancer (pancreatic, cervical, and lung), as well as the deadly and hard-to-detect lung disease idiopathic pulmonary fibrosis (IPF). Here’s how:

An IPF Tip – The tracer was initially meant to identify pancreatic cancer by attaching to alpha-v beta-6 proteins, but a suggestion from colleagues in the pharmaceutical industry led to a discovery that the protein was also elevated in patients with cervical/lung cancers and lung diseases like IPF.

Trials – Early trials successfully showed that the tracer can detect the cancers and IPF, while confirming that it is safe to use. The study even spotted signs of IPF in a “normal” control group patient, supporting its case as a method for early IPF detection. However, the team still needs more testing (and more funding) in order to tell how early the new tracer can detect these diseases.



DL for TB

A study from a team of international researchers found that deep learning algorithms can help identify TB-associated abnormalities in chest radiographs, recommending these solutions for TB programs with limited resources. Here are the details:

The Study – In a retrospective study using chest X-rays from 1,196 people in Nepal and Cameroon, the researchers found that Delft’s CAD4TB, Lunit’s INSIGHT, and Qure.ai’s qXR DL solutions identified signs of TB with similar AUCs (0.92, 0.94, 0.94). The solutions all achieved higher specificity than the study’s human radiologists in each region and maintained a sensitivity of at least 95%.

Real AI Science – The study also earned praise for its scientific quality, as it used independent evaluators, compared different models, leveraged data from multiple sites with different disease prevalence, followed STARD criteria, and blinded the developers to the disease data. That’s a lot more than can be said from the usual “AI beats radiologists” studies that we cover each week.

The Takeaway – The study found that each of the DL tools would be suitable for TB-stricken regions, improving productivity and reducing Xpert MTB/RIF test volume by 66%. Considering the prevalence of TB in some regions and the lack of peer-reviewed studies evaluating TB diagnostic algorithms, this is an important study and strong evidence that AI can help in the fight against TB.


The Wire

  • MediView XR raised $4.5 million to support the clinical trials (now in its second round) and ongoing development of its AR-based surgical system, leading up to a planned FDA clearance in 2021. MediView XR’s AR Surgical Navigation Platform leverages Microsoft HoloLens glasses, patient “maps” created from CT or MRI scans, and real-time ultrasound guidance to allow surgeons to navigate patients’ bodies and remove cancerous tumors with less radiation and potentially greater effectiveness than current fluoroscopy-based procedures.
  • South Korean AI developer, Vuno, launched a three-year collaboration with several European neuroimaging companies/institutes (NordicNeuroLab, Academic Medical Center in Amsterdam, Oslo University Hospital) to develop AI software used to predict long-term antidepressant efficacy for patients with major depression disorder (MDD). Vuno’s DEPREDICT software analyzes MDD patients’ MRI scans after they begin antidepressant treatment, making predictions that could shorten the antidepressant screening process by 75%. Given that MDD treatment evaluations often take over a year of trial and error, Vuno believes DEPREDICT could “fundamentally change” clinical decision making for MDD treatment.
  • A new paper published in Academic Radiology suggested that informaticists’ position as the link between radiology, data science, and IT makes them essential to the “development, evaluation and deployment of AI in the clinical environment.” Key parts of informaticists’ AI-related role include preparing data (curation, processing, labeling, and deidentification), serving as domain experts (helping AI developers understand imaging informatics and clinical context), serving as AI evaluators (workflow and clinical evaluations), and finally guiding AI workflow integration.
  • Butterfly Network announced the launch of Butterfly Enterprise, a new suite of tools intended to help hospital systems expand their POCUS programs. Butterfly Enterprise is intended to allow healthcare providers to scan / document / upload / review patient studies from their phones at the point-of-care and send this data to the EMR / PACS, simplify their POCUS quality assurance and credentialing processes, and provide enhanced security tools for Butterfly and other brand POCUS devices.
  • A new paper in JACR addressed the question of whether radiologists should wear white coats, detailing the symbolic reasons that radiologists wear them. Although radiologists’ decision whether to wear a white coat remains a more complex topic than some may expect, the paper concluded that this attire choice is up to individual radiologists and is less important than the quality of their work and communication skills.
  • A new ASRT survey (n=405) found that the U.S. radiographer vacancy rate (number of open and actively recruited positions) increased from 4.2% in 2017 to 8.5% in 2019. Nearly all disciplines posted higher vacancy rates in 2019 and many reached their highest rates in over a decade, led by CT (4.5% to 10.1%), MRI (3.9% to 8.7%), sonography (4.3% to 9.0%), mammography (2.7% to 5.6%), nuclear medicine (2.3% to 5.2%), and bone densitometry (1.7% to 3.7%). Only cardiovascular interventional technology saw vacancies drop (8.7% to 7.3%).
  • Nuance continued to expand its AI Marketplace solutions portfolio, adding Densitas’ densityai breast density assessment software and MaxQ AI’s Accipio Ix and Accipio Ax intracranial hemorrhage (ICH) detection software software. The additions come just a few weeks after revealing plans to support Qure.ai’s solutions and several months after adding VIDA Diagnostics’ pulmonary imaging analytics solution.
  • A new post from IHS Markit detailed the ultrasound industry’s trend towards chip-based transducers due the technology’s cost and size advantages versus piezoelectric crystal transducers, as well as their advantages with 3D imaging and full-body scanning. In addition to Butterfly and a number of well-funded startups, IHS reveals that major imaging and semiconductor manufacturers “are racing to produce chip-based ultrasound systems” in the next five years, suggesting that these future launches will have a significant impact on the imaging market and perhaps deliver on POCUS’ promise of expanding imaging to new areas of healthcare.
  • UC San Francisco and UC Berkeley scientists developed an AI algorithm that can identify and locate tiny brain hemorrhages in CT scans in 1 second, trace outlines of abnormalities in scans, and classify subtypes of each abnormality with greater accuracy (0.991 AUC) than two out of four radiologists. The team trained their “PatchFCN” algorithm on 4,396 CT exams, compensating for this small sample size by manually delineated each image at the pixel level and by only training the algorithm on one “patch” of an image at a time (vs. entire images or batches of images).
  • Dicom Systems was awarded a new U.S. patent (10,437,877 B2) for a “user-configurable radiological data transformation, integration, routing and archiving engine.” The patent covers technology in the Dicom Systems Unifier Enterprise Imaging platform for integrating radiological data (patient studies, orders, and reports) across disparate radiology systems (e.g., RIS, HIS, EMR, and/or other radiological systems) to provide a universal worklist and/or standardized data.

The Resource Wire

  • Did you know that imaging patients are most likely to no-show for their procedures on Mondays and Saturdays? By partnering with Medmo, imaging centers can keep their schedules full, despite the inevitable Monday no-shows.
  • This Carestream Special Report details how providers can get the greatest ROI from their X-ray technology as radiography demands increase and budgets head the other direction.

2 Responses to “DBT’s Marketing Juggernaut | Two-for-One PET | DL for TB”

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.

Join thousands of imaging professionals.