Insights into Imaging

🫁 Can deep learning reconstruction improve CT radiomics?

🔬 In pulmonary nodules, deep learning reconstruction showed high radiomic feature reproducibility, including in low-dose CT. Nodule size and reconstruction strength influenced stability.

💡 Optimising reconstruction protocols is key to more reliable CT radiomics.

buff.ly/wX6uq7D (Huiyuan Zhu et al.)

Pancreatic cancer after neoadjuvant therapy: can MRI reveal treatment response?

🔬 Diffusion-weighted imaging, T1 mapping, extracellular volume imaging, and USPIO-enhanced MRI may help distinguish viable tumour from treatment-related fibrosis and improve restaging.

💡 Better response assessment could support more informed treatment decisions.

buff.ly/uCw43JO (Giulia A. Zamboni et al.)

🎗️ AI-assisted mammography: improving diagnostic performance.

🩻 In a multicenter study, AI assistance increased sensitivity from 54.8% to 83.9% and specificity from 83.6% to 91.1%.

💡 AI decision support may help radiologists detect breast cancer and reduce diagnostic uncertainty.

buff.ly/rHbsrq4 (Nasrin Ahmadinejad et al.)

🚨 Ultraportable ultrasound in disaster medicine: imaging when resources are limited.

🩻 Across 63 studies, FAST achieved 92.1% sensitivity for intra-abdominal bleeding, while ultrasound changed patient management in up to 44% of cases.

💡 Promising for rapid triage and bedside guidance, but stronger evidence on patient outcomes is still needed.

buff.ly/YC9i1ld (Félix Amiot et al.)

🎨 Colour in medical imaging: more than just visualisation.

🩻 Inconsistent colour maps across ultrasound, CT, MRI, and PET may lead to misinterpretation of quantitative data.

💡 Standardisation could improve lesion conspicuity, cross-platform comparability, and multicentre research.

buff.ly/EbcVpJg (Nico Sollmann et al.)