Motion tracking results of DeepMesh (the proposed method).
Results are shown by overlaping meshes and images. Red contours are predicted results while green contours are ground truth
SAX | 2CH| 4CH
--- | --- | ---
|
|
Comparison methods
FFD | dDemons | 3D-UNet | MulViMotion|MeshMotion|DeepMesh(Proposed)
--- | --- | --- | ---| --- | ---
|
|
|
|
|
The network architecture of the Deformation network (H_D(·))
The network architecture of the Motion network (H_M(·))
The network architecture for 2D feature learning in the Motion network
The network architecture for 3D feature learning in the Motion network