Pediatric CMR time drops to 29 minutes

Good morning, there. Free breathing DL cardiac MRI cut pediatric scan time from 43.0 to 29.2 minutes.

I see this as a practical AI win because it changes sedation risk, not just image quality. Shorter scans and far less intubation could expand pediatric CMR access while preserving quantitative confidence.

Where would shorter AI reconstructed MRI most change your pediatric workflow?

Here's what you need to know about Radiology AI last week:

• Faster pediatric CMR without breath holds

• FDA clears Transpara 2.1.0

• Mammography AI needs device level tuning

• Novice operators screen aortic stenosis with AI

• Plus: 2 newly released datasets, 6 FDA approved devices & 4 new papers.

BUILT BY US

Handle multiple studies faster with Batch mode

🫀 Faster pediatric CMR without breath holds

RadAI Slice: This study stands out because the workflow gains are clinically tangible.

The details:

• Compared 150 free breathing DL CMR exams with 150 conventional exams

• CMR time fell to 29.2 vs 43.0 minutes

• Intubation with muscle relaxation dropped to 14% vs 90%

• Anesthesia time fell to 70.8 vs 98.3 minutes

• RV and LV cardiac index correlated with flow at R 0.92 and 0.84

Key takeaway: For pediatric radiology, this moves AI reconstruction from image polish to a sedation sparing workflow change.

🧾 FDA clears Transpara 2.1.0

RadAI Slice: This clearance adds another regulated breast AI option for screening workflows.

The details:

• FDA 510k K260898 covers Transpara 2.1.0

• ScreenPoint Medical is the listed manufacturer

• The tool supports mammography lesion detection for possible cancer

• The payload lists no public performance metrics for this clearance

Key takeaway: For breast imaging teams, the clearance reinforces the need to evaluate AI in local screening populations and devices.

🎛️ Mammography AI needs device level tuning

RadAI Slice: This finding feels directly relevant to real world breast AI deployment.

The details:

• Study included 22,673 mammograms across 6 screening sites

• Data came from 3 mammography devices and women aged 50 to 69

• Double reading CDR was 5.07 per 1000 screens

• AI CDR was 4.81 with a general threshold and 4.90 with device thresholds

• Consensus workload fell by nearly one third with device thresholds

Key takeaway: Local thresholding may be essential, since one device still showed poor AI performance despite device specific tuning.

🩺 Novice operators screen aortic stenosis with AI

RadAI Slice: This study shows how AI may shift acquisition and interpretation beyond expert echo labs.

The details:

• Model development used 6753 Mayo Clinic patients

• Prospective novice cohort included 1302 focused ultrasound exams

• Automated analysis was possible in 1258 of 1302 exams

• Sensitivity was 93% and specificity was 96%

• Expert review raised PPV from 49.4% to 91.1%

Key takeaway: The radiology lesson is that acquisition guidance plus expert review can make imaging AI safer than standalone automation.

MedReaMM (2026-08-23)

Modality: Multimodal imaging: WSI/pathology, CT | Focus: Multi-organ, rare diseases | Task: Differential diagnosis, multimodal reasoning

• Size: 625 patient cases; about 1,744 medical images, 2.79 images per case.

• Annotations: 1,042 diagnoses with ICD-11 codes. Primary and secondary labels, filtered captions, and optional reasoning paths.

• Institutions: Huazhong University of Science and Technology, Fudan University, et al.

• Availability:

  Public: GitHub

• Highlight: Expert-validated case-report benchmark for open-ended multimodal diagnostic synthesis.

AS UBM Measurement Dataset (August 24, 2026)

Modality: UBM | Focus: Eye; anterior segment | Task: Segmentation; automated measurement

• Size: UBM images from 185 eyes of 138 patients. Ages 3 weeks to 26 years. Includes 1086 measurements.

• Annotations: 27 quantitative anterior segment parameters. Includes demographics and diagnosis metadata. No pixel masks reported.

• Institutions: University of Maryland School of Medicine; Children’s National Hospital; et al.

• Availability:

  Public: Zenodo

• Highlight: Open pediatric UBM dataset with standardized anterior segment measurements across healthy and diseased eyes.

🏛️ FDA Clearances

• K253808 - GE Medical Systems received 510k clearance for AIR Recon DL, a deep learning MRI reconstruction tool.

• K262452 - Radformation received 510k clearance for AutoContour RADAC V6 to support automated radiotherapy contouring.

• K261916 - Cortechs Labs received 510k clearance for NeuroQuant PET to quantify brain PET images for clinical review.

• K253794 - Lucida Medical received 510k clearance for Prostate Core, an MRI tool for prostate image assessment.

• K260086 - BeamWorks received 510k clearance for CadAI B Dx, an AI tool for lesion detection in medical images.

• K262449 - ADVAHEALTH SOLUTIONS received 510k clearance for AdvaView, a radiological image processing system.

• Explore last week's 7 radiology AI FDA approvals.

📄 Fresh Papers

• doi:10.1016/j.acra.2026.08.034 - FDA approved mammography AI detected 81% of retrospectively visible interval breast cancers on false negative screens.

• doi:10.3174/ajnr.A9609 - DMGtracker improved pediatric diffuse midline glioma segmentation on 88 external trial MRI studies.

• doi:10.1016/j.ebiom.2026.106457 - Across 600 head CT and MRI studies, invasive defacing degraded downstream AI while mri_reface stayed within 3% to 5%.

• doi:10.1148/ryct.250464 - A Radiology Cardiothoracic Imaging study found 466 stable CT radiomic features for pulmonary nodules across scanners.

• Browse 318 new radiology AI studies from last week.

📰 Everything else in Radiology AI last week

• A real world Academic Radiology study found four commercial chest radiograph AI tools offered workflow benefits but limited accuracy gains.

• AI analysis of more than 1000 CT scans linked higher thymus radiation dose with worse lung cancer outcomes.

• Tokyo researchers tested facial video AI in 215 participants, reporting over 90% hypertension accuracy in 5 seconds.

• Read 5 fresh radiology AI news articles for last week.

That's it for today!

👋 Quick favor: drag this into your Primary tab so you don’t miss next week. Or just hit Reply with one thought.

See you next week.

P.S. We keep building free tools to accelerate your radiology work. What's the most time-consuming pain point in your day that we should help speed up? Reply and share your take so we keep building around you.