Luminal stenosis is not the sole determinant of coronary risk. Integrating coronary CT angiography, plaque characterization, fractional flow reserve, and AI with angiography enables personalized risk stratification and treatment.
A two-stage deep learning model detected endotracheal tubes on pediatric chest X-rays with a 0.994 area under the curve (95% CI 0.986-1.000) and localized tube tips with 6.59 mm mean error; inter-reader error was 2.01 mm; single-center pilot.
Liquid biopsy and molecular panels (Afirma, ThyroSeq), integrated with strain elastography, ACR TI-RADS, and 99mTc-MIBI scintigraphy, may reduce unnecessary thyroidectomies for indeterminate nodules, per a comprehensive review.
LLMs (large language models) failed to reproduce a radiology residency rank list from applications alone (median τ 0.15–0.36), but adding human interview scores raised agreement to τ 0.84–0.93. Application-only LLM rankings placed female, international, and non-MD applicants low…
A physics-informed deep learning model trained entirely on simulated wrist 4DCT data effectively corrected motion artifacts, improving LIRS in cadaveric bone regions from 0.58±0.19 to 0.77±0.14 (p<0.01) and SSIM in simulated scans from 0.87 to 0.95.
Strategic Multidimensional Benchmark Deployment - Operationalizing National Standards on Prostate MRI and PI-RADS Adoption.
A Meaningful First Step Toward Large Language Model Quality Control of Breast Ultrasound Reports.
Over 950 AI/ML devices cleared by FDA, ~75% in radiology; yet an Indian review finds a persistent knowledge gap, with fewer than 30% of radiologists familiar with AI. Algorithms validated superior for lung nodules, breast screening, TB triage, stroke, glioma.
Real-world AI in radiology faces a deployment gap where performance drops. This review discusses continuous monitoring on clinical platforms to maintain utility.
Screen-detected breast cancers (vs non-screen-detected) were earlier stage, lower grade, and had 43% lower all-cause mortality (5-year survival 92.9% vs 73.6%) in a cohort of 1,023 women. Benefit persisted in those ≥75 years.
AI can automate image interpretation, enhance dosimetry, and enable personalized radiopharmaceutical therapy, but challenges in data quality, interpretability, and validation currently limit clinical adoption.
In 261 nondysplastic knee MRIs from ages 8–22, deep learning-based 3D modeling showed proximal physis migration and trochlear deepening between 13–19 years, with males having greater lateral trochlear depth after age 16. Findings may guide trochleoplasty timing.
Structured longitudinal ultrasound education integrating theory with supervised clinical practice prepares medical students for responsible use. A review covers physics, knobology, artifacts, safety (ALARA), and debates on simulation, handheld devices, and AI integration.
FDA 510(k) clearance for Boston Scientific’s SpaceOAR Flex System (product code OVB, radiology panel).
FDA clears MICSI-PET, a positron emission tomography system from Microstructure Imaging, under the 510(k) pathway (product code KPS).
FDA cleared BoneVue (Curvebeam AI/Curvebeam, LLC) via the 510(k) premarket pathway.
AI algorithms for detecting lines, drains, and airways on chest radiographs can approach radiologist-level accuracy, but most still fall short of clinical deployment. Multicenter validation, pediatric datasets, and anatomically informed models are needed for integration into cli…
White matter hyperintensities (WMH) disrupt brain structural networks differently in cerebral small vessel disease vs dementia, with stronger correlation to connectivity changes in at-risk individuals (r=-0.32) than dementia (r=-0.03).
A dual-curriculum deep learning model outperforms state-of-the-art approaches in segmenting retinal OCT and fundus images across unseen imaging devices, potentially reducing annotation costs.
A multimodal AI education tool reached 38.83% top-1 retrieval recall across 65,419 images and 12.5x faster CPU synthesis; blinded radiologists saw only a small quality drop, and its LLM auditor correlated 0.805 with an expert radiologist.
A ResNet-50 on 421 AP knee radiographs with expert labels reached 0.345 balanced accuracy for six-class Schatzker classification, only 0.168 above a non-anatomical baseline (95% CI 0.106–0.230, p=0.002). Tibia-only crops performed better than whole radiographs, and the mask alon…
Radiomics extracts high-dimensional quantitative features from spinal CT and MRI, enabling early detection of osteoporosis, fragility fracture prediction, differentiation of benign vs malignant lesions, and assessment of intervertebral disc degeneration.
Accelerated sagittal cervical spine magnetic resonance imaging (MRI) using VORTEX (Variable-rate selective excitation [VERSE] specific absorption rate [SAR] reduction, Optimised radiofrequency [RF] waveform, Refocus flip angle modulation, Tuned hyper echoes, External volume suppr
More than 1,200 physicians have launched the Cardiac PET Alliance, representing 101 cardiovascular practices in 24 states that perform over 400,000 cardiac positron emission tomography (PET) studies annually, to advocate for expanded access.
Recruiting the Radiologists of Tomorrow
Radical radiotherapy use rising for treating lung cancer
<![CDATA[Video: Stamatia Destounis, MD, FACR, Discusses Key Changes for the Updated BI-RADS System]]>
NYU Langone Health's AI Tool Enhances Personalized Breast Cancer Screening
New Imaging Data Supports Taladegib's Efficacy in Treating Idiopathic Pulmonary Fibrosis
Radiologist Workforce Growth Trails Behind Rising Healthcare Demand
