Communication Dans Un Congrès Année : 2025

Lesion-centered vision transformer for stroke outcome prediction from image and clinical data

Résumé

Accurate prediction of stroke functional outcome, particularly the 3-month modified Rankin Scale (mRS), is crucial for personalized treatment. Vision Transformers excel in medical imaging and multimodal fusion but struggle with stroke MRI due to data scarcity and rigid tokenization, which may miss subtle anomalies. In response, we propose the Lesion-Centered Vision Transformer (LC-ViT), integrating lesion-focused MRI preprocessing, adaptive token merging, and multimodal fusion. LC-ViT extracts axial, coronal and sagittal views centered on ischemic lesions to optimize visibility and employs a pretrained TC-Former (token-clustering transformer) for adaptative token generation. A mutual cross-attention mechanism further integrates imaging and clinical data. Evaluated on a retrospective private cohort comprising DWI MRI and 62 clinical variables (e.g. demographics, neurological assessments.) of 119 stroke patients treated with thrombectomy (65% favorable outcome), LC-ViT achieves a new state-of-the-art performance (AUC:0.80± 0.03, Accuracy: 0.77 ± 0.02) significantly outperforming single modality based deep architectures. Our results highlight the potential of lesionfocused tokenization for stroke outcome prediction and interpretability and broader applications in lesion-localized multimodal analysis. Our code is available at https://github.com/mingtian12345/LC-VIT.

Fichier principal
Vignette du fichier
MICCAI_2025__Copy_.pdf (1.32 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05303811 , version 1 (08-10-2025)

Licence

Identifiants

Citer

Mingtian Liu, Nima Hatami, Laura Mechtouff, Tae-Hee Cho, Carole Lartizien, et al.. Lesion-centered vision transformer for stroke outcome prediction from image and clinical data. Medical Image Computing and Computer Assisted Intervention -- MICCAI 2025, Sep 2025, Daejeon, South Korea. ⟨10.1007/978-3-032-05182-0_29⟩. ⟨hal-05303811⟩
464 Consultations
223 Téléchargements

Altmetric

Partager

  • More