kamruleee51/TAM

We propose a scalable Temporal Attention Module (TAM) to inject cardiac motion information into segmentation models. TAM captures dynamic changes across temporal frames using multi-headed, cross-temporal attention. It is adaptable across imaging modalities, scalable from 2D to 3D, and adds minimal computational overhead.

Size: 59.6 KB - Last synced at: 2 months ago - Pushed at: 2 months ago - Stars: 0 - Forks: 0