Angio-FILM addresses a practical tradeoff in coronary angiography: lower frame rates can reduce radiation exposure, but may compromise temporal detail. This study shows that latent flow matching can…

This study moves glioma recurrence assessment from lesion-level labels toward spatial maps of suspected recurrent tumor. Using physiologic MRI and machine learning, the authors linked predicted intralesional patterns…

A multicenter study in nasopharyngeal carcinoma shows that combining anatomy, spatial dose, DVH parameters, and clinical data can predict radiation-induced temporal lobe injury before treatment.

Passive smartphone keyboard data can capture suicide-related language among adolescents, but interpretation is clinically delicate.

Lingjian is a domain-specific vision-language model built for clinical laboratory microscopy, showing expert-level performance across diagnostic morphology tasks.

HRRT models renal replacement therapy as a sequence of linked ICU decisions, from whether to start treatment to modality, dose, and cessation.

A new annotation-free framework uses foundation model guidance to reconstruct and segment uterine fibroids from routine multi-planar MRI, narrowing the gap with fully supervised target-domain training.

Transcript INTRO Welcome to Medical AI Digest Podcast. Today, a paper from Circulation. MAIN Today we’re looking at a multinational, multimodal study titled “Contrastive Machine Learning to Quantify…

Transcript INTRO Welcome to Medical AI Digest Podcast. Today, a paper from NPJ digital medicine. MAIN The paper is titled “Advanced machine learning-based screening for primary aldosteronism with…

A multicenter MRI-based AI model, BCALN-Net, showed strong performance for preoperative assessment of axillary lymph node status in breast cancer.