Paint Point AI | SR’s Strategic Shift

“And no radiologist in the world will be upset if they never have to do these studies again.”

Dr. Luke Oakden-Rayner with a warm welcome for an AI algorithm that can measure radiographic lower leg length discrepancies 96-times faster than specialists.


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



Paint Point AI

Usually when a new AI study comes out claiming significantly greater accuracy or speed than trained experts, radiologist eye-rolls are soon to follow. However, the results of a new AI study published in the Radiology journal received a much warmer reception because it showed how AI could streamline the “time consuming yet cognitively simple” task of measuring pediatric leg length discrepancy (LLD).

  • The Study – The researchers developed a deep learning model using X-rays from 179 children (70 training, 32 validation, 77 testing), which calculated leg length discrepancies in 1 second per image (vs. radiologists’ 96 seconds), and maintained a “high” correlation with radiologist measurements.
  • Rad Reactions – We’ve heard this plenty of times from the radiology community: the next big imaging AI success story is more likely to owe its success to saving time, rather than saving lives. That seems to be the case for this algorithm, which happens to automate a task that “no radiologist in the world will be upset if they never have to do” again.



SR’s Strategic Shift

Strategic Radiology (SR) revealed a major change in its structure and strategy, as it begins transitioning its affiliate practices “toward one class of membership–ownership.”

SR’s Big Change – This is a big change for SR, which historically positioned affiliate membership as a way for practices to achieve scale while staying independent.

New Ownership – The first SR affiliate practices to transition to “SR member owners” are Casper Medical Imaging (Casper, WY, 9 rads), Radiologic Medical Services (Coralville, IA, 8 rads), and Naugatuck Valley Radiological Associates (Waterbury, CT, 13 rads), joining SR’s original founding “member owners.”

Next Owners – Only 10 of SR’s 27 practices are currently “member owners” but we will see more announcements like this as the affiliate program is phased out in 2020 and SR affiliate members choose to retain membership by becoming owners..


The Wire

  • Radio-Genomics Glioma Milestone: A UT Southwestern Medical Center team developed a radio-genomics algorithm that might help patients with glioma tumors avoid pre-treatment surgery by assessing Isocitrate dehydrogenase (IDH) mutation status in T2 MRI brain scans (IDH assessment currently requires invasive surgery). The researchers used MRI data and corresponding genomic info from 214 patients (94 IDH-mutated, 120 IDH wild-type) to develop a pair of deep learning networks, finding that the algorithms predicted IDH mutation status with 97% accuracy.
  • EU CV 19 Workflow: A consortium of European healthcare institutions are assembling a CV19 image analysis workflow intended to speed up CV19 diagnosis and reduce radiation exposure. The workflow involves ultra low-dose CT scanning, AlgoMedica’s PixelShine AI tool (to improve ultra low-dose CT quality), and InferVision’s InferRead CT Pneumonia solution (to analyze the “improved” CT images) before the images are sent to radiologist teams in Italy, the Netherlands, and Germany.
  • The ABR Gives In: Amid rising backlash against its user agreement, the American Board of Radiology finally gave in a bit, removing language that required physicians to waive their right to sue the ABR. Although this is a step in the right direction, the ABR isn’t up for any popularity awards and waiving this clause doesn’t help the fact that it’s already facing a lawsuit for allegedly tying its initial certification and MOC products together.
  • Alphatec Backs Out: EOS imaging joined the growing list of companies (and reps) who thought they landed a deal back before COVID-19 became a pandemic only to have it fall through during March and April. In this case, Alphatec terminated its $122M acquisition of EOS imaging, specifically citing CV19’s impact on the orthopedic imaging company’s financial situation.
  • CT + Clinical CV19 AI: Case Western Reserve researchers are developing a computational tool that uses radiomic textural patterns in CT scans along with clinical / demographic inputs (symptoms, duration, comorbidities, blood cells, proteins) to identify COVID-19 patients that will require hospitalization and ventilation. The team built one model based on neural networks and another using radiomics, which predicted ventilation with 68% to 75% accuracy. That’s not very impressive at first glance, but the researchers believe accuracy will improve as they incorporate more clinical and comorbidity factors.
  • MITA Asks for CDC Flexibility: MITA called on the CDC to incorporate the new Fleischner Consensus into its guidance and encouraged the federal agency to create an ongoing partnership with MITA and other organizations to help broadcast COVID-19 imaging research across healthcare. Reading between the lines, MITA is asking the CDC, which does not recommend CT or X-ray alone for CV19 diagnosis, to recognize the areas where the Fleischner Consensus says imaging can be used for CV19 (i.e. patients with worsening respiratory status, patients at risk of disease progression, triage in resource-constrained environments).
  • HeartVista Adds $8.65M: AI-assisted MRI solution company HeartVista wrapped up an $8.65m Series A round (increasing its total to $16.35m) that it will use to develop new musculoskeletal and neural products (HeartVista was previously cardiac focused), expand internationally, and deepen its alliances with major cardiology centers. In an interesting twist, the new round was led by Vinod Khosla (the guy known to proclaim that AI will put radiologists out of business), while HeartVista’s previous investment came from the radiologist-led / radiology-focused Bold Brain Ventures fund.
  • Google AI, Better on Paper: A new MIT Tech Review story revealed how Google Health’s AI tool intended to diagnose diabetic retinopathy using eye scans achieved solid results in “the lab” (>90% accuracy, results in 10 minutes vs. up to 10 weeks) but didn’t work out in the real world. After deploying the tool in Thailand, Google Health found that the tool did improve result times when it worked, but sometimes failed to analyze scans because the model wasn’t trained for real world image quality (creating new inefficiencies). This isn’t our type of imaging AI, but it is yet another reminder that “even the most accurate AIs can actually make things worse if not tailored to the clinical environments in which they will work.”
  • OHIF Expands to CV19: Mass General Hospital announced that its Open Health Imaging Foundation (OHIF) web-based viewer, which was originally developed for cancer imaging research and clinical trials, can now be used to view / compare / analyze medical images for “any disease” including COVID-19. The CV19-focused announcement revealed that the free and “already popular” web viewer and/or its underlying Cornerstone libraries are now supporting a number of CV19 projects including the DetectED-X’s CovED education platform, VUNO’s CV19 algorithms, and the Nextcloud DICOM Viewer.
  • KM Scales Down with Rede PACS: Konica Minolta Healthcare announced the U.S. launch of its Rede PACS and Rede Mini PACS platforms intended for specialty clinics (orthopedic, urgent care, and family practice). The Exa-based and web-based Rede PACS platform provides quick image access from a PC and is scaled-down for specialty practices (<20k studies /yr, simple installation, lower tier-based pricing), while Rede Mini PACS offers a smaller hardware footprint and is bundled with Konica Minolta’s X-ray and ultrasound systems (<2.5k studies /yr).
  • Pulmonary Surgery AI: RSIP Vision launched a suite of pulmonary imaging AI modules developed for interventional pulmonologists. The suite’s primary module uses CT scans to segment airways and map the lungs for surgery planning, while suite’s other modules segment lung lobes / fissures and identify lung lesions. The new RSIP Vision suite is positioned as an alternative to traditional pulmonary surgery imaging tools including endoscopes, chest X-ray, and endobronchial ultrasound.
  • Nanox & Hadassah: Nanox launched a partnership with Hadassah Hospital to develop early detection screening protocols that will be used in the Nanox.ARC platform. The Israeli hospital’s radiologists will utilize the Nanox.ARC X-ray system to develop the new protocols, which will apparently be combined with a suite of third party AI diagnostic tools available through the Nanox.ARC platform (Nanox already signed Qure.ai and CureMetrix).

The Resource Wire

– This is sponsored content.

  • Join Healthcare Administrative Partners’ webinar, “Reentering the Post-COVID-19 Radiology Market,” on Wednesday, May 13 where they will discuss several factors radiology practices should take into consideration to ensure safe and successful reentry into the market.
  • Join Riverain Technologies’ CSO, Jason Knapp, at the Aunt Minnie Virtual Conference today at 4pm EDT as he discusses the foundational considerations for AI application design and validation, and how these principles apply to thoracic imaging.
  • Watch an informational video from Jeff Hersh, M.D., Chief Medical Officer of GE Healthcare on using ultrasound in a health crisis.
  • Ready to future proof your organization? Catch Nuance Diagnostic Solutions GM & SVP, Karen Holzberger, and VA radiologist Woojin Kim, MD at the Aunt Minnie Virtual Conference (Friday 5/1, 12pm EDT) as they discuss how the industry can regain momentum and begin its evolution using the power of structured data, AI and ambient technology.
  • This Qure.ai blog post details how it repurposed its qXR chest X-ray AI tool to detect signs of COVID-19, creating a CV19 detection tool that is now in use at 28 global sites.

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.