Bi-temporal Image-driven Acute Stroke Evolution Analysis
The study developed a bi-temporal framework to characterize ischemic tissue by analyzing features from two time points: admission (T1) and post-treatment follow-up (T2). Using statistical descriptors, radiomics, and deep learning embeddings from mJ-Net and nnU-Net, the researchers found that healthy tissue that recovered showed feature similarity to preserved brain tissue, while infarct regions formed distinct clusters. This supports using acute CTP features for stroke tissue phenotyping.