Intermountain’s Imaging Centers

Intermountain Healthcare expanded into outpatient imaging with the launch of its new imaging center subsidiary, Tellica Imaging. Plenty of hospital systems have outpatient imaging centers, but how and why Intermountain created Tellica brings some important takeaways.

About Tellica – Tellica Imaging plans to open a fleet of outpatient MRI and CT centers, starting with three Utah locations by late 2021, five locations in 2022, and more locations in “subsequent years.” The Tellica centers will prioritize patient convenience and value, targeting easy-to-access locations and adopting a novel flat-rate pricing model that’s well below typical in-hospital rates.

The Value-Based Angle – Given Intermountain’s role as one of the country’s flagship value-based care systems and its unique payor-provider structure, launching a series of imaging centers that are lower cost and more convenient makes a lot of sense. It’s also a step away from the hospital-based/owned procedure trend that’s helped hospitals from a reimbursement perspective, but brought a long list of unintended consequences (higher patient/payor costs, provider consolidation, imaging overuse, etc.).

The Payor Angle – Even though many patients use Intermountain’s in-house insurer (SelectHealth), Intermountain also works with a long list of commercial and government payors, nearly all of which have been incentivizing (or forcing) health systems to move more imaging procedures to outpatient centers. SelectHealth likely has the same preferences.

The Offsite Trend – In addition to the above payor pressures, there are some major trends underway that favor offsite imaging, including the rapid adoption of at-home/remote patient care, new COVID-related offsite policies, and the federal government’s efforts to make healthcare procedures more “shoppable.”

The Takeaway – Hospital-owned outpatient imaging centers aren’t all that unique, but Intermountain’s structure definitely is (payor-provider, value-based, non-profit) and so is its decision to launch these centers with such a patient-friendly value proposition. Even if most hospitals aren’t yet ready to offer flat-rate scans, the factors that drove Intermountain to create Tellica are likely forcing plenty of other systems to rethink their own approach to offsite imaging.

GE Acquires BK Medical

GE Healthcare’s ultrasound portfolio became a lot more diverse last week with its acquisition of surgical ultrasound company BK Medical. Here’s some details and perspectives:

The Acquisition – GE Healthcare will acquire BK Medical from Altaris Capital Partners for $1.45b, separating BK Medical from Analogic. That’s a pretty big investment considering that GE’s ultrasound unit brings in $3b a year.

GE’s Surgical Expansion – With BK Medical, GE’s ultrasound unit expands from diagnostics to intraoperative imaging and surgical navigation, which is reportedly a fast-growing and high-margin business for BK Medical. 

The BK Portfolio – BK Medical got its start in urology ultrasound, and more recently expanded to ultrasound systems used to guide minimally invasive and robotic surgeries and to visualize deep tissue during neuro and abdominal surgeries. That adds up to five unique ultrasound systems.

GE Impact – GE sees a lot of value in BK Medical. BK gives GE an ultrasound portfolio that the other OEMs can’t match (diagnostic, surgical, post-operative), “accelerates” GE’s precision health strategy, and will reportedly deliver “high-single-digit” ROI within five years.

GE Acquisition Trend – While GE Healthcare spent 2018 and 2019 selling major non-imaging businesses (value-based care to Veritas Capital, life sciences to Danaher), GE’s 2020 and 2021 acquisitions have focused on expanding its capabilities within imaging (Zionexa for radiopharmaceuticals, Prismatic Sensors for CT detectors, and now BK Medical for ultrasound). That says a lot about GE Healthcare’s imaging focus, and is quite different from Philips and Siemens, which have increasingly targeted M&A outside of imaging.

Silent Atherosclerosis

A new study in Circulation used coronary CTA scans and CAC scoring to reveal a surprisingly high prevalence of “silent” coronary artery atherosclerosis in the general population, suggesting that this could “lay the foundation” for future CT-based cardiac screening programs.

The Study – The researchers analyzed CCTA and CAC exams from 25k randomly recruited Swedish participants (50-64yrs, none w/ known coronary heart disease) finding that:

  • 42% had CCTA-detected atherosclerosis
  • 8.3% had noncalcified plaques
  • 5.2% had significant stenosis
  • 1.9% had serious coronary artery diseases
  • All participants with >400 CAC scores had atherosclerosis (yes, 100%), and 45.7% had significant stenosis
  • Some participants with 0 CAC scores had atherosclerosis (5.5%) and significant stenosis (0.4%)
  • So, CAC-based screening might still miss some at-risk patients

The Takeaway – 2021 brought a notable surge in academic and business efforts focused on CT-based cardiac screening, and this study’s revelation about “silent” atherosclerosis in the general population suggests that cardiac screening’s momentum will continue.

Volpara’s Lung Cancer Push

Breast imaging AI leader Volpara Health took a big step into the lung cancer AI segment last week, launching partnerships with Riverain Technologies and RevealDx. Here are some details.

Volpara & Riverain – Volpara and Riverain announced plans to integrate Riverain ClearRead CT (AI-based lung nodule detection) and the Volpara Lung platform (lung cancer screening reporting, tracking, and risk assessment), giving Volpara a market-leading detection partner and extending the clinical value of both tools.

Volpara & RevealDx – Within days, Volpara announced a $250k strategic investment in AI-based lung nodule diagnosis startup RevealDx, that will allow Volpara to sell RevealDx’s RevealAI-Lung tool (CE-marked, FDA pending) in the US and make Volpara its exclusive distributor in Australia / New Zealand. 

Not That Surprising – Volpara’s lung cancer screening expansion isn’t as surprising as some might think. Volpara first entered the lung cancer screening segment through its 2019 acquisition of MRS Systems, which likely targeted MRS’ breast cancer screening management software but also included its lung cancer screening platform (used w/ 8% of US LC screenings). Volpara also built its business around supporting population-scale cancer screening workflows and it has a long history of complementary partnerships within its breast imaging business.

The Takeaway – Lung cancer screening volumes are about to significantly increase in the US (and potentially globally), creating new bandwidth and workflow constraints, and driving demand for comprehensive solutions that support the entire screening and patient management pathway. With these alliances, Volpara, Riverain, and RevealDx are far better positioned to support that pathway.

Aidoc and Subtle Medical’s End-to-End Alliance

Aidoc and Subtle Medical launched an interesting new partnership that will make Subtle’s image acquisition / enhancement software available on the Aidoc AI platform.

End-to-End Partnership – The addition of SubtlePET and SubtleMR to the Aidoc AI platform will create what Aidoc called an “end-to-end” solution and “the first joint offering of AI for both image acquisition and triage.” Some folks might mistake that to mean that they will create new combined image acquisition+triage solutions, but they won’t be specifically linked (Aidoc doesn’t have MRI or PET tools yet anyway).

Aidoc, a Platform Company – Aidoc seems to be increasingly positioning itself as an AI platform company, which is an understandable strategy given users’ need for comprehensive / consistent AI workflows. Aidoc’s initial partnerships also allow the triage-focused vendor to offer a far more comprehensive value proposition (Subtle for acquisition, icometrix for stroke analysis/assessment).

Subtle Upsides – The alliance introduces Subtle Medical to Aidoc’s sizable list of clients (used at >500 medical centers, a high profile partnership w/ RP), and adds to Subtle’s current alliances with AI marketplace vendors (e.g. Blackford, Nuance, Incepto) and complementary solutions companies (e.g. Cortechs.ai).

The Takeaway – Although AI platform alliance stories don’t usually earn a spot at the top of The Imaging Wire, this alliance is pretty notable given what it suggests about Aidoc’s AI platform strategy and about the growing trend towards complementary AI alliances. It’s also a nice way for Subtle Medical to expand its reach.

Bad AI Goes Viral

A recent mammography AI study review quickly evolved from a “study” to a “story” after a single tweet from Eric Topol (to his 521k followers), calling mammography AI’s accuracy “very disappointing” and prompting a new flow of online conversations about how far imaging AI is from achieving its promise. However, the bigger “story” here might actually be how much AI research needs to evolve.

The Study Review: A team of UK-based researchers reviewed 12 digital mammography screening AI studies (n = 131,822 women). The studies analyzed DM screening AI’s performance when used as a standalone system (5 studies), as a reader aid (3 studies), or for triage (4 studies).

The AI Assessment: The biggest public takeaway was that 34 of the 36 AI systems (94%) evaluated in three of the studies were less accurate than a single radiologist, and all were less accurate than the consensus of two or more radiologists. They also found that AI modestly improved radiologist accuracy when used as a reader aid and eliminated around half of negative screenings when used for triage (but also missed some cancers).

The AI Research Assessment: Each of the reviewed studies were “of poor methodological quality,” all were retrospective, and most studies had high risks of bias and high applicability concerns. Unsurprisingly, these methodology-focused assessments didn’t get much public attention.

The Two Takeaways: The authors correctly concluded that these 12 poor-quality studies found DM screening AI to be inaccurate, and called for better quality research so we can properly judge DM screening AI’s actual accuracy and most effective use cases (and then improve it). However, the takeaway for many folks was that mammography screening AI is worse than radiologists and shouldn’t replace them, which might be true, but isn’t very scientifically helpful.

Hospitals’ Outside Rebound

A new McKinsey survey found that healthcare system leaders expect patient volumes to surpass 2019’s levels by next year, while revealing an interesting shift in how/where many of these patients will be getting their care.

The Rebound – The leaders from 100 large private US hospitals reported that their ED and inpatient/outpatient volumes returned to 2019 levels in July 2021 (so, somewhat pre-Delta) and forecast that 2022 volumes will be 5%-8% above 2019.

The Outpatient Shift – Outpatient procedures are expected to drive much of this future patient growth (+8% in 2022, +9% in 2023), with the biggest outpatient increases in orthopedics, psychiatry, and cardiology.

The Virtual Shift – Although it would take another lockdown to return to 2020’s virtual care numbers, the leaders expect virtual visits to represent around 15% of their outpatient volume in 2022/2023, 300% higher than in 2019.

It’s Not Just McKinsey – You probably don’t need a prestigious consulting firm to tell you that more procedures are moving to outpatient settings and more patient visits are being held virtually. The outpatient shift has been going on for some time, and the recent evolution of telehealth tech and home care delivery has brought some major home care commitments from the biggest systems in the country. We even launched an excellent new newsletter to help providers keep up with healthcare’s virtual shift.

The Radiology Impact – Technically the McKinsey forecast didn’t mention imaging once, but patients’ continued shift to beyond hospital walls will definitely have an imaging impact, including more virtual radiologist consultations, more outpatient image-guided procedures, and more at-home and near-home imaging. It could also mean less in-hospital imaging.

Unsupervised COVID AI

MGH’s new pix2surv AI system can accurately predict COVID outcomes from chest CTs, and it uses an unsupervised design that appears to solve some major COVID AI training and performance challenges.

Background – COVID AI hasn’t exactly earned the best reputation (short history + high annotation labor > leading to bad data > creating generalization issues), limiting most real world COVID analysis to logistic regression.

Designing pix2surv – pix2surv’s weakly unsupervised design and use of a generative adversarial network avoids these COVID AI pitfalls. It was directly trained with CTs from MGH’s COVID workflow (no labeling, no supervised training) and accurately estimates patient outcomes directly from their chest CTs.

pix2surv Performance – pix2surv accurately predicted the time of each patient’s ICU admission or death and applied the same analysis to stratify patients into high and low-risk groups. More notably, it “significantly outperformed” current laboratory tests and image-based methods with both predictions.

Applications – The MGH researchers believe pix2surv can be expanded to other COVID use cases (e.g. predicting Long COVID), as well as “other diseases” that are commonly diagnosed in medical images and might be hindered by annotation labor.

The Takeaway – pix2surv will require a lot more testing, and its chance of maintaining this type of performance across other sites and diseases might be a longshot (at least right away). However, pix2surv’s streamlined training and initial results are notable, and it would be very significant if a network like this was able to bring pattern-based unsupervised AI into clinical use.

Veye Validation

A team of Dutch radiologists analyzed Aidence’s Veye Chest lung nodule detection tool, finding that it works “very well,” while outlining some areas for improvement.

The Study – After using Veye Chest for 1.5 years, the researchers analyzed 145 chest CTs with the AI tool and compared its performance against three radiologists’ consensus reads, finding that:

  • Veye Chest detected 130 nodules (80 true positive, 11 false negative, 39 false positives)
  • That’s 88% sensitivity, a 1.04 mean FP per-scan rate, and 95% negative predictive value
  • The radiologists and Veye Chest had different size measurements for 23 nodules
  • Veye Chest tended to overestimate nodule size (bigger than rads w/ 19 of the 23)
  • Veye Chest and the rads’ nodule composition measurements had a 95% agreement rate

The Verdict – The researchers found that Veye Chest “performs very well” and matched Aidence’s specifications. They also noted that the tool is “not good enough to replace the radiologist” and its nodule size overestimations could lead to unnecessary follow-up exams.

The Takeaway – This is a pretty positive study, considering how poorly many recent commercial AI studies have gone and understanding that no AI vendor would dare propose that their AI tools “replace the radiologist.” Plus, it provides the feedback that Aidence and other AI developers need to keep getting better. Given the lack of AI clinical evidence, let’s hope we see a lot more studies like this.

Get every issue of The Imaging Wire, delivered right to your inbox.

You're signed up!

It's great to have you as a reader. Check your inbox for a welcome email.

-- The Imaging Wire team