A YOLOv8 model achieved precision 0.983, recall 0.988, F1 0.987 for classifying TMJ disc displacement with vs without reduction on open-mouth MRI (n=237 patients).
A multimodal MRI ensemble predicted meningioma WHO grade with external AUC 0.742, but did not significantly outperform nnU-Net (AUC 0.732; ΔAUC +0.011, p=0.737).
CT-derived GFR showed the strongest correlation with measured iohexol clearance (R²=0.77) and lower mean absolute error (12.8 mL/min) vs. Cockcroft-Gault (15.6), cystatin C-based (14.2), and Lund-Malmö (13.0) estimates. Routine renal CT can precisely estimate kidney function.
A benchmarking framework with synthetic breast datasets and task-based image quality measures revealed that deep learning reconstruction methods for photoacoustic computed tomography often fail to accurately recover lesions, despite performing well on traditional metrics.
Cross-method analysis of Segment Anything Model (SAM) adaptations shows no strategy is universally optimal; effectiveness depends on capability-task gap, data scale, domain discrepancy, imaging dimensionality, resources, and clinical constraints. Priorities for clinical use: sta…
Reflections on the 2026 Intersociety Summer Conference
Weill Cornell New York City - Musculoskeletal Radiologist Job Opening in New York, New York
An AI system achieved AUCs of 0.86–0.89 for detecting clinically significant prostate cancer on MRI across three international datasets, meeting criteria for diagnostic interchangeability with radiologists.
Interventional rads treat half of Medicare liver cancer patients
Lu-177 PNT2002 before SBRT extends disease control in prostate cancer
Survey of 231 imaging staff worldwide: 60% of institutions have sustainability policies for iodinated contrast, but 26% of respondents never do waste management. Hospital size and younger age (25–44) linked to better implementation.
AI on par with radiologists for finding prostate cancer on MRI
CT Colonography Gains Ground in Colorectal Cancer Screening
Butterfly Network, VCOM form handheld ultrasound education alliance
Philips Unveils Latest Imaging Innovations at EuroPCR 2026
ACR Announces 2027 Gold Medalists and Honorary Fellows
HPI: Evidence contradicts broad claims about AI, radiology workforce
Recent Jobs - Teleradiology - American College of Radiology Career Center
VIDEO INTERVIEW | Dr. Božinovska: What does BI-RADS mean and when is mammography necessary?
Radiologists are shifting to teleradiology for better pay and less pressure as traditional partnership paths lose appeal due to declining reimbursements and private equity, per Diagnostic Imaging.
AI applications closest to neuro-oncology clinical use include imaging segmentation and quantification, AI-assisted image reconstruction, DNA methylation-based classification, and workflow optimization; radiogenomics and other promising tools still need prospective validation.
A 2.5D multi-instance learning MRI model differentiated small-cell from non-small-cell lung cancer brain metastases with an external validation AUC of 0.826, comparable to a clinical model (AUC 0.826, p=1.000).
Adding contrast-enhanced ultrasound features and nodule location to American College of Radiology TI-RADS raised the area under the curve from 0.75 to 0.90 and sensitivity from 60.3% to 88.5% for TR4 or higher thyroid nodules, cutting unnecessary biopsy rate from 50.6% to 32.9%
Beyond the Headlines: What the Evidence Actually Says About Radiology Today and Into the Future.
In 2446 asymptomatic UK Biobank participants, 14.6% had moderate/severe vertebral fractures on DXA. More common in men (18.8% vs 10.8% women); adjusted OR for women 0.51 (0.40–0.66). BMD and conventional risk factors poorly predicted fractures.
No CT or PET/CT restaging method consistently predicted both R0 resection and survival in localized pancreatic cancer after systemic therapy across 20 studies; RECIST 1.1 is limited for partial response and stable disease. Integrating CA19-9 may improve accuracy.
Integrated AI-CAD reduced exam time (P<.001) and had higher sensitivity (87.6% vs 76.9%) but lower specificity (74.2% vs 90.0%) than sequential AI-CAD in 242 patients. AUC was similar.
Diagnostic performance of an open-source artificial intelligence model for detecting chest infections on after-hours emergency department chest radiographs in Singapore.
