Beyond Core and Penumbra: Bi-Temporal Image-Driven Stroke Evolution Analysis

Md Sazidur Rahman, Kjersti Engan, Kathinka Dæhli Kurz, Mahdieh Khanmohammadi

Keywords: Perfusion imaging, CTP, Diffusion MRI, DWI, Radiomic features, CNN embedding, Stroke progression analysis.

TL;DR: We do bi-temporal stroke evolution analysis using image-derived features.

Abstract:

Computed tomography perfusion (CTP) at admission is routinely used to estimate ischemic core and penumbra, whereas follow-up diffusion-weighted MRI (DWI) obtained after treatment provides the definitive infarct outcome. However, single time-point segmentations do not capture the biological heterogeneity of stroke and ignore its continuous temporal evolution. We propose a bi-temporal analysis framework that characterizes ischemic tissue using statistical descriptors, radiomic texture features, and deep feature embeddings from two architectures (mJ-Net and nnU-Net). Bi-temporal refers to the admission ($T_1$) and the first post-treatment follow-up ($T_2$). All features are extracted at $T_1$ from CTP, and follow-up DWI is aligned with CTP to ensure spatial correspondence. Manually delineated masks at $T_1$ and $T_2$ are intersected to construct six regions of interest encoding both initial tissue state and final outcome. Extracted features were aggregated per region and analyzed in feature space. Evaluation on 18 patients with successful reperfusion demonstrated meaningful clustering of region-level representations. Regions classified as penumbra or healthy at $T_1$ that ultimately recovered exhibited feature similarity to preserved brain tissue, whereas infarct-bound regions formed distinct groupings.

Deep feature spaces, particularly mJ-Net, showed strong separation between salvageable and non-salvageable tissue, while nnU-Net further revealed a gradient-like organization with core and healthy brain at opposite ends and penumbra occupying transitional positions.
These findings suggest that encoder-derived feature manifolds may reflect underlying tissue phenotypes and progression trajectories, providing an inshight toward imaging-based quantification of stroke evolution.

Primary Subject Area: Application: Radiology

Secondary Subject Area: Detection and Diagnosis

Registration Requirement: Yes

Visa & Travel: Yes

Read CFP & Author Instructions: Yes

Originality Policy: Yes

Single-blind & Not Under Review Elsewhere: Yes

LLM Policy: Yes

Submission Number: 275