Radiology

Abbreviated breast MRI approximates full protocols for post-NAC response assessment

In 478 women treated with neoadjuvant chemotherapy and surgery, abbreviated breast MRI was compared with full-protocol MRI for predicting pathologic complete response. Abbreviated MRI used T2-weighted and early pre- and postcontrast T1-weighted sequences, read independently by three radiologists. Sensitivity, NPV, PPV, and overall accuracy for pathologic complete response were similar between abbreviated and full protocols across readers and definitions. Full-protocol MRI had modestly higher specificity than abbreviated MRI for both pathologic complete response definitions. 1

Two-week FDG-PET prognosticates outcomes in newly diagnosed metastatic breast cancer

This multicenter IMPACT-MBC cohort enrolled 200 patients with nonrapidly progressive, newly diagnosed metastatic breast cancer starting first-line systemic therapy. Baseline biopsy, FDG-PET, and CT were followed by early FDG-PET at 2 weeks and CT response assessment at 8 weeks. Non–progressive disease on early FDG-PET had a 94.7% negative predictive value for non-progression on 8-week CT across metastatic subtypes and bone-only disease. Patients without early PET progression had substantially longer median progression-free and overall survival than those with early progression under standard care. 2

Review highlights imaging-based selection paradigms for acute stroke reperfusion

This review summarizes how acute ischemic stroke imaging now underpins diagnosis, prognostication, and eligibility for thrombolysis and endovascular therapy. It describes the evolution from rigid time windows toward CT- and MR-based tissue and collateral status assessment to refine benefit–risk. The article details ongoing and emerging trials where perfusion, core–penumbra mismatch, and collateral imaging drive reperfusion eligibility. Key future priorities include streamlined protocols for non-tertiary centers and improved imaging frameworks for wake-up and very-late-presenting strokes. 3

PACS-integrated AI tool standardizes longitudinal follow-up of brain metastases

In 40 patients with brain metastases and 158 MRIs after stereotactic radiotherapy, a PACS-integrated AI lesion tracking tool was evaluated against manual analysis. The AI system used a U-Net–based model for automatic 3D segmentation on contrast-enhanced T1 MRI and generated diameter measurements for neuroradiologist review. Segmentation quality was good, with a median Dice similarity coefficient of 0.771 compared with reference labels. Manual reads by two neuroradiologists showed high but imperfect agreement, with 74 lesions detected by only one reader, including some that later progressed. 4