pix2surv Unsupervised | DBT Impact

“We’re through the eye of the needle on this one.”

Nisonic AS’ Tormod Selbekk on developing the company’s new intracranial pressure ultrasound.

Imaging Wire Sponsors

Arterys | Bayer Radiology | Blackford Analysis
Canon Medical Systems | Fujifilm Healthcare Americas
GE Healthcare | Novarad | Nuance
Riverain Technologies | Siemens Healthineers
United Imaging | Zebra Medical Vision

The Imaging Wire

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.

ClearRead CT’s Impact at Einstein Medical

This Riverain Technologies case study details how Einstein Medical Center adopted ClearRead CT enterprise-wide (all 13 CT scanners) and how the solution allowed Einstein radiologists to identify small nodules faster and more reliably.

– Sponsored.

Ramapo Radiology’s Case for Novarad CryptoChart

See how New Jersey’s Ramapo Radiology Associates overcame their CD burning problems and improved their physician and patient experiences with Novarad CryptoChart.

– Sponsored.

The Wire

  • ICP Ultrasound: Norway’s Nisonic AS is earning headlines for its P-100 ultrasound system, which combines optic nerve sheath ultrasound scans and AI to noninvasively measure intracranial pressure (ICP). Current ICP measurement approaches are definitely invasive (surgically implanted brain sensors or drains), suggesting that the Nisonic P-100 might lead to the first ICP diagnosis and treatment processes that never involve invasive procedures (e.g. just Nisonic scans and diuretics).
  • TB AI Works: A new Lancet Digital Health study found that the leading tuberculosis AI tools “significantly outperform” experienced readers and would cut Xpert TB tests in half. In the study, five current TB AI tools (Delft’s CAD4TB, Infervision’s InferRead DR, Lunit’s INSIGHT CXR, JF Healthcare’s CXR-1, Qure.ai’s qXR) detected TB in 23,954 CXRs better than three radiologists (AI AUCs: 84.9% – 90.8%). qXR and CAD4TB were the top performing tools (AUCs: 90.8% & 90.3%) and the only two that met the WHO’s triage accuracy guidelines (≥90% sensitivity and ≥70% specificity).
  • DBT’s Cost Impact: A new JAMA study provided more evidence that DBT’s adoption boom significantly increased breast screening costs. The review of 15.6m screening claims from 2013 to 2019 (8.4m women), found that DBT’s rapid adoption (13% in 2015 to 70% 2019), higher exam costs ($315 DBT vs. $238 DM), and higher downstream testing costs ($99 per DBT vs. $75 per DM) increased private payors’ screening expenditures by 33% ($3.9b in 2013 to $5.2b in 2019).
  • MRIguidance Funding: MRIguidance landed $1.25m in VC funding (total now ~$3m) that it will use to support the US launch of its BoneMRI software, which transforms MRI images into CT-quality 3D bone scans. MRIguidance became available in Europe earlier this year and plans to raise more capital as it scales-up its commercial efforts.
  • Native Language MRI Prep: Showing native language instructional videos to English as a Second Language patients (ESL) helps them follow abdominal MRI exam instructions and improves image quality. NYU researchers showed Spanish or Mandarin instructional videos to 29 ESL patients and compared their exams to patients who didn’t watch the videos (50 ESL, 81 English-speaking). As you might expect, images from the ESL patients who didn’t watch the videos had lower image quality and respiratory motion ratings than the other patients.
  • DRX-Evolution Plus Evolves: Carestream updated its DRX-Evolution Plus X-ray room system, adding a number of workflow (larger panel, extended tube column, new LEDs showing exam progress, Smart Room option) and patient comfort (updated tabletop) features. The DRX-Evolution Plus has been Carestream’s flagship room X-ray system since way before The Imaging Wire existed, and has relied on regular updates like this to remain competitive.
  • Screening Out TNBC Disparities: Mammography screening helps catch early stage Triple Negative Breast Cancer (TNBC, higher mortality than non-NTBC), thus mitigating breast cancer disparities among Black women who are 2x more likely to develop TNBC. Weill Cornell researchers reviewed 756 TNBC cases (301 screening-detected), finding that screening-detected TNBC cases were more likely to have T1 lesions (73.1% vs. 32.9%) and be node-negative (51.9% vs. 40.4%) than non-screening TNBC cases, while having far higher 5-year survival rates (92.8% vs. 81.5%).
  • Simplifying Appropriate Use: A new Health Affairs editorial urged CMS and Congress to simplify the PAMA advanced imaging Appropriate Use Criteria (AUC) program, arguing that AUC’s administrative burdens now outweigh its actual impact on over-imaging. The authors proposed folding the AUC program into CMS’ existing QPP programs, or as a “less effective” plan-B they suggest simplifying the PAMA AUC program’s claims process.
  • Quality Over Quantity: A new survey of 1k US healthcare consumers found that 62% had never heard of value‐based care, yet 59% said they would support a healthcare system that compensates providers based on the quality of their work (vs. quantity). This disconnect apparently goes beyond terminology, as many patients were unaware that value-based care exists.
  • RBknee Cleared: Danish AI startup Radiobotics announced the FDA 510(k) approval of its RBknee product, which detects knee osteoarthritis findings in X-rays (osteophytes, subchondral sclerosis, joint space narrowing), and measures joint space width in both compartments of the knee. Although we’ve seen a rise in academic osteoarthritis AI projects, RBknee is one of the few OA AI tools to gain FDA approval (joining IB Lab’s KOALA… maybe others).
  • The Case for Short Stroke MRI: A new European Radiology study found that performing additional short-protocol brain MRIs on emergency patients with mild stroke symptoms and negative head CT results improves detection of minor strokes. By allowing earlier detection and treatment, the added MRIs reduced these patients’ overall healthcare costs (CT + short MRI: $26,304, CT-only: $27,109) and increased their quality-of-life years (14.31 vs. 14.25 QALYs).

Breaking Barriers in MRI

Siemens Healthineers’ MAGNETOM Free.Max is the world’s first MRI with an 80 cm bore that is improving the patient experience. See how it is breaking barriers in MRI.

– Sponsored.

The Resource Wire

  • Room for more efficiency in your breast imaging operations? Check out this GE Healthcare post detailing how new technologies are improving patient experiences and making breast imaging teams more efficient.
  • Why United Imaging’s MI (uMI)? Every United Imaging molecular imaging system features its “uEXPLORER Inside” technology platform, which is designed for total-body scanning, is scalable for clinical systems, and excellent in an MR environment – you’ll see a big difference and your patients can benefit from their focus on coverage, clarity, and sensitivity.
  • The flow of new AI applications makes it hard for radiology groups to determine which tools would help them and how IT teams can handle increased AI adoption. In this Blackford Analysis white paper, radiology and IT leaders from NYU and Canopy Partners share how a platform approach alongside a curated marketplace can help solve these challenges.
  • See how Arterys LungAI matched (and actually exceeded) radiologists’ accuracy measuring lung nodule volumes in CT scans.

You might also like

You might also like..

Select All

You're signed up!

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

-- The Imaging Wire team

You're all set!