Rapid prediction of hemorrhagic transformation after endovascular thrombectomy: a multimodal model in patients with post-thrombectomy cerebral hyperdensities

1

Hemorrhagic transformation (HT) occurs in 70.5% of patients after endovascular thrombectomy (EVT) for acute ischemic stroke.
2

A 2.5D multimodal deep learning framework was developed to predict HT using post-thrombectomy cerebral hyperdensities.
3

The Fusion_Transformer model achieved an AUC of 0.803 in predicting HT, outperforming human neuroradiologists.
4

The study included 393 patients from three stroke centers, validating the model across training, validation, and external test cohorts.
5

The model integrates imaging features with clinical data to provide rapid and accurate risk assessments for HT post-EVT.

Frontiers In Neurology

by Ziwen Wang, Guolan Song, Ying Tang, Jiahong Xu, Junli Wang, Qingdian Cong, Jingjing Fu, Yue Wang, Jibo Hu, Leling Tu, Song Cheng, Jian Ding, Sheng Hu
July 7, 2026

1