The 2026 Chinese lung cancer guideline adds AI and opportunistic screening in low-dose CT, expands molecular testing, and updates treatment regimens for NSCLC, including perioperative, consolidation immunotherapy, and targeted therapy for EGFR, ALK, KRAS, HER2, and MET alteratio…
On 76 test cases, the AI framework achieved Cobb angle MAE 2.4°, Lenke curve type accuracy 0.895, and reduced review time from 8.2 to 1.5 min/case.
PSAD remained below 0.11 ng/mL² across all ages in 1037 healthy German men, contrasting with continuously rising PSA. 95th percentile reference: PV 51.39 mL, PSA 2.79 ng/mL, PSAD 0.071 ng/mL².
A multiphase CT angiography spot sign identified high-risk intracerebral hemorrhage: hematoma expansion occurred in 25/51 (49%) phase-1, 13/28 (46%) phase-2, and 4/9 (44%) phase-3 spots versus 12/71 (17%) with no spot sign.
Closing the Gap in Femoral-Neck Fracture Detection: AI Assistance with Pelvic and Hip Radiographs.
MRI-guided amyloid PET reconstruction improved interreader agreement (Fleiss κ 0.91 vs 0.79) and specificity (99.2% vs 91.6%), reducing false positives from 8.1% to 0.8% vs OSEM in a retrospective study of 120 patients.
Deep learning model OccuNet detected femoral-neck fractures on hip/pelvic radiographs with 97.5% sensitivity, 98.8% specificity, AUC 0.99. It raised reader sensitivity and cut reading time: radiologists 93.7% to 97.2%, emergency physicians 84.3% to 95.6%, time -14.9% and -18.9%.
Super-resolution contrast-enhanced US (SRCEUS) differentiated moderate/severe from remission/mild Crohn disease with AUC 0.92, sensitivity 94.1%, and specificity 95.0% in a prospective study of 54 patients. Performance exceeded conventional CEUS (AUC 0.78, P=.03).
Black-White Racial Differences at Breast MRI.
A Progressive Shift in Procedural Sedation: Is It Time to Embrace Ketamine in Interventional Radiology?
Assuring Transparency of AI Software after Deployment
Comprehensive multimodal model achieved AUC 0.86 for baseline cerebral microhemorrhage detection in ADNI; minimal clinical-only model still reached AUC 0.72, supporting resource-limited screening.
AI tools for postoperative radiographs after reverse total shoulder arthroplasty are most clinically ready for implant identification and automated glenosphere orientation measurement. Detection of scapular notching, loosening, and stress fractures remains exploratory. The key b…
AdaptSeg, using cross-fraction priors, boosted abdominal OAR segmentation mean DSC by 4.1% (HD95 −43%, ASSD −39%) vs. best single-fraction method, with inference <1.6s—fast enough for online MR-guided adaptive radiotherapy.
Erratum to “Radiomics features of FDG PET images predict ApoE4” [Eur. J. Radiol. Artif. Intell. 2 (2025), 100009]
A 3D generative AI model synthesized FDG-PET from T1-weighted MRI and arterial spin labeling in Alzheimer's disease, achieving >93% structural similarity to real PET and generalizing to an independent external cohort.
Test-set composition and the interpretation of biopsy reduction in AI-guided breast MRI
In 206 participants followed up to 10 years with 791 MRI scans, higher cumulative depression severity predicted steeper hippocampal (ηp2=0.261) and dlPFC (ηp2=0.216) gray matter loss, while diagnosis alone showed no differential change.
FDA 510(k) clearance for Siemens’ MAGNETOM Free.Max and Free.Star MRI systems, classified under radiology product code LNH.
Eizo's RadiForce MX247W diagnostic monitor received FDA 510(k) clearance for radiology image interpretation. It is classified under product code PGY for diagnostic displays.
A diffusion model (D3Recovery) for inter-crystal scatter recovery in Prism-PET achieved 70.1% crystal identification accuracy for ICS events, reduced localization error from 2.18 mm to 0.99 mm, and improved DOI resolution to 3.6 mm FWHM. Image quality and uniformity also improve…
Patients in the most disadvantaged neighborhoods had brain age gaps 2.72–3.02 years higher, lower total brain volume, and more white matter hyperintensities on brain MRI, in a study of 2826 patients.
Brain Health and Neighborhood: Is Disadvantage Hard Wired?
Helical 3D DSA enabled single-injection whole-head-and-neck angiography (aortic arch to vertex) in a first human case, with 419.4 mm coverage in 7.6 sec, 11 lp/cm resolution, and 2.57% stenosis measurement error vs conventional DSA.
Generative Pretrained Transformer (GPT)-5-Thinking matched subspecialist top-1 accuracy in diagnosing orbital (76.7% vs 78.3%, P=.64) and head-and-neck tumors (67.5% vs 68.0%, P=.92). Generalist accuracy rose from 61.4% to 70.3% for orbital and 47.0% to 61.0% for head-and-neck t…
Bone Changes in Desmoplastic Fibroma of Bone.
From Opportunistic Chest CT to Predictive Radiology in MESA: An Early Career Perspective.
Muscle and Bone Health: One Step Closer to Understanding the Link between Risk Identification and Prevention.
A scoping review of 29 studies on AI in radiology education found 86% reported positive or improved outcomes. Skill development was most studied (55.2%), then performance (27.6%) and engagement (17.2%). Generative AI showed promise but raised reliability, bias, and ethics concer…
Review of agentic AI in radiology: GPT-4 catches 82.7% of report errors, matching humans; intelligent triage cut turnaround time ≤43.7%. But evidence is single-center, retrospective — prospective validation lacking.
A CT deep-learning model combining U-Net segmentation and clinical features distinguished tuberculous from malignant pleural effusion with test-set area under the receiver operating characteristic curve (AUC) 0.934 (95% CI 0.8733-0.9955), sensitivity 0.875, specificity 0.900.
In 1,072 adults having health-exam chest CT, higher atherogenic index of plasma (OR 1.38, 95% CI 1.16-1.64) and systemic inflammation response index (OR 1.37, 1.14-1.64) were independently associated with pulmonary nodules; adding indices did not significantly improve AUC (0.77…
Tools and Architectures for Agentic Systems in Clinical AI and Radiology
A self-supervised graph contrast model for functional brain networks improved major depressive disorder identification accuracy by 3% and 5% at two sites versus prior state of the art. Adapted Grad-CAM flagged key frontal, limbic, striatal, thalamic, and insular regions.