Radiology

State-of-the-art breast ultrasound: beyond the simple mass workup

This review positions breast US as a core adjunct to mammography and MRI for screening, diagnosis, staging, and intervention guidance. Screening whole-breast US, handheld or automated, is emphasized for dense breasts as a supplement, not replacement, to mammography. The article highlights AI tools to enhance diagnostic accuracy and workflow, and emerging echotexture quantification as a risk biomarker. It also reviews evolving technologies—US tomography, optoacoustic imaging, and contrast-enhanced US—and stresses meticulous scanning technique for reliable diagnosis. 1

DeepENE improves CT detection of extranodal extension in laryngeal and hypopharyngeal cancer

In 289 patients with laryngeal or hypopharyngeal squamous cell cancer, DeepENE accurately detected pathologic extranodal extension on pretreatment CT. The model achieved AUCs of 0.93 internally and 0.87–0.96 across three external LHSCC test sets. DeepENE consistently outperformed five head and neck specialists, particularly for early ENE, with markedly higher sensitivities at matched specificities. These results support integrating automated ENE assessment into staging workflows to better guide surgery, radiotherapy fields, and systemic therapy decisions. 2

CT-based prediction of outcome and EVT benefit in anterior circulation LVO

Using individual data from seven randomized EVT trials (n=1391), this study built a CT-only model predicting functional outcome after anterior LVO. Baseline CT features related to stroke severity and brain frailty yielded substantial discrimination for good outcome (mRS 0–2; C≈0.70). Adding age and NIHSS improved performance, approaching that of MR PREDICTS for both outcome and EVT treatment benefit prediction. The work supports using routine baseline CT, combined with limited clinical data, to estimate individual EVT benefit when advanced imaging is unavailable. 3