Prognostic Value of Coronary CT Angiography–Derived Quantitative Flow Ratio in Suspected Coronary Artery Disease Open Access 

Radiology, published 12/10/24 | Estimated Read Time: 25 minutes

New research reveals CT angiography-derived quantitative flow ratio (CT-QFR) independently predicts major adverse cardiovascular events with a hazard ratio of 1.9 (P=.04) and demonstrates comparable 5-year prognostic value to invasive coronary angiography/SPECT (AUC: 0.64 vs 0.67, P=.19), while delivering substantially lower radiation exposure (3.2 mSv vs 21.7 mSv). However, CT-QFR's predictive power significantly declined in patients with prior percutaneous coronary intervention (AUC: 0.44 vs 0.70, P<.001), suggesting potential limitations in this specific patient population. 

Nerve cross‐sectional area in vincristine‐induced polyneuropathy: A nerve ultrasound pilot study Open Access

Journal of Neuroimaging, published 12/9/24 | Estimated Read Time: 15 minutes

Research on vincristine-induced polyneuropathy (VIPN) reveals that high-resolution nerve ultrasound detected significant increases in total nerve entrapment site cross-sectional area at 8 weeks (p=.007), coinciding with peak clinical symptoms, though individual nerve sites showed no significant changes. While this 12-patient pilot study was limited by its small sample size, it demonstrated that clinical examination outperforms ultrasound for early VIPN detection, with nerve conduction studies confirming significant deterioration in sensory nerve action potentials after 6 months across multiple nerves (sural p=.001, radial p=.004, ulnar and median p<.001).

Surveillance, Epidemiology, and End Results Data Show Increasing Rates of Distant-Stage Breast Cancer at Presentation in U.S. Women Open Access

Radiology, published 12/10/24 | Estimated Read Time: 25 minutes

New research analyzing SEER data from 2004-2021 (71-80 million women annually) reveals distant-stage breast cancer rates increased significantly across all age groups, with the sharpest rise among women aged 20-39 years (2.91% annual increase, P<.001) and Black women experiencing 55% higher incidence rates than White women. The COVID-19 pandemic in 2020 disrupted diagnosis patterns, particularly affecting older women and racial/ethnic minorities, with the study showing increased rates of distant-stage disease for Asian women (2.90%, P<.001), Black women (0.86%, P=.008), Hispanic women (1.56%, P<.001), and Native American women (3.86%, P=.04) over the study period.

How Prostate Cancer Growth Patterns Impact Detection and Interreader Agreement on Multiparametric MRI Open Access

Academic Radiology, published 12/5/24 | Estimated Read Time: 15 minutes

Study of 52 prostatectomy cases demonstrates that infiltrative prostate cancer growth patterns are more frequently missed on T2-weighted imaging (29% vs 4%) and show higher ADC values (0.97×10⁻³ mm²/s vs 0.83×10⁻³ mm²/s) compared to expansive patterns, with excellent inter-reader agreement (κ = 0.88-0.90). The research reveals expansive growth patterns significantly correlate with higher PI-RADS scores (OR 2.1-2.2), though the study's limitations include its small sample size and potential bias from readers' awareness of prostatectomy status.

The role of deep learning in diagnostic imaging of spondyloarthropathies: a systematic review Open Access

European Radiology, published 12/10/24 | Estimated Read Time: 20 minutes

Systematic review analyzing 21 studies demonstrates that deep learning models are achieving remarkable accuracy in diagnosing spondyloarthropathies across imaging modalities, with some models reaching AUCs up to 0.98 and matching expert radiologist performance. The most impressive performance came from an ensemble model diagnosing ankylosing spondylitis with precision of 0.91 and recall of 0.90, though the review notes that most studies lack external validation and sufficient sample sizes for immediate clinical implementation.

A Laser Atherectomy-Thrombectomy System for Primary Management of Acute Limb Ischemia Open Access 

Journal of Vascular and Interventional Radiology, published 12/9/24 | Estimated Read Time: 25 minutes

Study (n=24) demonstrates 92% technical success using the Auryon laser atherectomy-thrombectomy system as primary treatment for acute limb ischemia, with all patients achieving immediate post-procedure clinical success and 92% maintaining success at 30 days. While the results suggest potential for reducing reliance on traditional treatments, with no major device-related adverse events and only 12.5% requiring adjuvant thrombolysis, the study's limitations include its single-center retrospective design and a cohort that was predominantly Rutherford class 1 (75%), warranting larger prospective studies with more severe cases.

National Utilization Trends of Inpatient Procedures for Symptomatic Uterine Fibroids and Adenomyosis: A 10-year Analysis

Journal of the American College of Radiology, published 12/9/24 | Estimated Read Time: 25 minutes

Landmark study analyzing 247,476 inpatient procedures (2011-2020) demonstrates that hysterectomy remains the predominant treatment for uterine fibroids and adenomyosis (80.1%), while uterine artery embolization (UAE) continues to be underutilized at just 2.6% despite comparable outcomes. Though myomectomy rates for fibroids increased significantly from 15.4% to 24.0%, UAE utilization remained stagnant, highlighting potential gaps in treatment awareness and access - however, the study's focus on inpatient procedures only (which declined 68.3% over the decade) may not fully capture the shift toward outpatient care.

Uterine Artery Embolization for Mixed Adenomyosis and Fibroids: Outcomes According to Initial Particle Sizes

Journal of Vascular and Interventional Radiology, published 12/8/24 | Estimated Read Time: 20 minutes

Researchers found that initial particle size in uterine artery embolization (UAE) for mixed adenomyosis/fibroids (small: 300-500μm vs. large: 500-700μm) did not significantly affect clinical outcomes, with both groups showing similar high improvement rates in menorrhagia (89.1%), pelvic pain (92.3%), and bulk symptoms (97.4%). While the large-particle group demonstrated greater reduction in fibroid enhancement (84% vs 55%, p=0.013), this difference did not translate to meaningful clinical outcomes in the 76-patient study, though the retrospective design and sample size limited detection of effects below 30-40%.

Three-dimensional volumetric CT image fusion and trans-abdominal US: Adjunct guidance to portal vein cannulation for TIPS

European Journal of Radiology, published 12/4/24 | Estimated Read Time: 20 minutes

Research comparing guidance techniques for TIPS procedures (n=86) demonstrates that both CT-fluoroscopy image fusion and trans-abdominal ultrasound guidance significantly reduced procedural time (~130-140 min vs 159 min) and contrast usage (~120-145 ml vs 195 ml) compared to conventional methods (p < 0.001), while maintaining comparable technical success rates of 92-96%. The CT-fluoroscopy fusion approach showed particular promise in reducing fluoroscopic time (43.19 vs 63.05 min, p = 0.012).

Outcomes of Intranodal and Modified Intranodal Lymphangiography for Treatment of Traumatic Chylous Leaks in the Thorax and Neck Open Access 

Cardiovascular and Interventional Radiology, published 12/10/24 | Estimated Read Time: 15 minutes

Study of 239 patients with traumatic chylous leaks, modified intranodal lymphangiography using pneumatic compression achieved superior technical success (94.2% vs 76.5%, p=0.002) and significantly reduced procedure times (83.4 vs 119.2 minutes, p<0.0001) compared to traditional intranodal lymphangiography. While both techniques demonstrated comparable clinical success rates of approximately 95%, this research suggests mIL should be the new standard, though findings from this single-center study with experienced operators warrant validation across multiple institutions. 

The Value of Machine Learning-based Radiomics Model Characterized by PET Imaging with 68Ga-FAPI in Assessing Microvascular Invasion of Hepatocellular Carcinoma Open Access 

Academic Radiology, published 12/7/24 | Estimated Read Time: 25 minutes

Study demonstrated that 68Ga-FAPI PET radiomics models can predict microvascular invasion in hepatocellular carcinoma with high accuracy (AUC=0.896, accuracy=87.5%, specificity=100%), significantly outperforming traditional imaging methods like ultrasound (AUC=0.731) and CT (AUC=0.86). Using logistic regression models with a 50% SUVmax threshold, researchers analyzed 84 HCC patients (53 MVI-positive, 31 MVI-negative), though the study was limited by its small sample size and lack of clinical model comparison.

Comparing Large Language Model and Human Reader Accuracy with New England Journal of Medicine Image Challenge Case Image Inputs Open Access

Radiology, published 12/10/24 | Estimated Read Time: 25 minutes

OpenAI's GPT-4 Omni achieved 59.6% accuracy on New England Journal of Medicine Image Challenge cases, surpassing medical students (47.1%, p<.001) but falling significantly behind junior faculty radiologists (80.9%, p<.001) and residents (70.2%, p=.003). A crucial discovery was that LLMs' performance depended more on text length than actual image interpretation, with models showing 73.1%-79.1% accuracy for long text inputs versus just 34.8%-40.6% for short inputs (p<.001), regardless of whether images were provided.

Current status and future directions of explainable artificial intelligence in medical imaging

European Journal of Radiology, published 12/6/24 | Estimated Read Time: 20 minutes

In a timely review, researchers analyze how explainable AI could address the "black box" barrier in clinical AI adoption, examining parallel methods between radiologists' cognitive processes and XAI approaches (Grad-CAM, LIME, SHAP). While commercialized FDA-approved AI solutions like Aidoc and PathAI have successfully integrated explainability features through heatmaps, the review finds that state-of-the-art XAI methods remain insufficient for clinical implementation, with current techniques proving too complex for medical professionals to readily interpret and apply in practice, suggesting a critical need to better align AI explanations with clinical decision-making patterns.

Enhancing Oncological Surveillance Through Large Language Model-Assisted Analysis: A Comparative Study of GPT-4 and Gemini in Evaluating Oncological Issues From Serial Abdominal CT Scan Reports Open Access

Academic Radiology, published 12/9/24 | Estimated Read Time: 15 minutes

In a comparison of GPT-4 and Gemini analyzing serial CT scan reports from 205 cancer patients, GPT-4 significantly outperformed Gemini in matching corresponding findings (96.2% vs 91.7%, P < 0.01) and determining tumor status (87.6% vs 73%, P < 0.001), with both models completing analyses in under 30 seconds. While promising for clinical workflow integration, both models struggled with precision (GPT-4: 0.68, Gemini: 0.63, P = 0.006) due to false positives in classifying benign findings as oncological issues, suggesting the need for specialized medical training of these language models.

Comparing fully automated AI body composition biomarkers at differing virtual monoenergetic levels using dual-energy CT

Abdominal Radiology, published 12/7/24 | Estimated Read Time: 25 minutes

In a retrospective study of 88 contrast-enhanced DECT scans, researchers found that AI-derived body composition biomarkers demonstrated significant variability at lower virtual monoenergetic levels (40-55 keV), with median attenuation differences ranging from 39-358 HU compared to standard 70 keV imaging. The findings revealed that Agatston calcium scores were particularly susceptible to energy-level variations, with low keV measurements leading to 39.1% upstaging and 23.6% downstaging, while measurements at or above 70 keV showed minimal variability and strong correlation (R² ≥0.92) for most biomarkers except bone and calcium scoring.

Upcoming Conferences and Approaching Abstract Submission Deadlines

If you have any comments, questions, or feedback, please respond to this email!

Thank you,

Emily Barnard