VolMoVis: Real-Time Volume Generation and Motion Visualization with Dynamic Tomographic Reconstruction.

Area of Science:

• Medical imaging
• Computational anatomy
• Computer-aided interventions

Background:

• Visualizing 3D anatomical structure motion is crucial for interventions.
• Conventional 4D reconstruction methods have low temporal resolution and complex segmentation needs.
• Real-time dynamic volume visualization remains a challenge.

Purpose of the Study:

• To develop a method for real-time dynamic tomographic reconstruction and volumetric motion visualization.
• To address limitations of conventional 4D reconstruction regarding temporal resolution and segmentation.
• To enable continuous visualization of anatomical motions for improved interventions.

Main Methods:

• VolMoVis employs a continuous implicit neural representation.
• Dynamic volumetric data is decomposed into a static reference volume and a continuous deformation field.
• An efficient deformation network facilitates real-time processing.

Main Results:

• VolMoVis achieves real-time volume generation and volumetric visualization of continuous anatomical motions.
• The method demonstrates effectiveness in accurate anatomical reconstruction and motion tracking on phantoms and patient data.
• Real-time simultaneous volume generation and tumor segmentation were achieved for 4D tumor tracking.

Conclusions:

• VolMoVis offers a novel approach for dynamic tomographic reconstruction with enhanced temporal resolution.
• The framework facilitates real-time visualization of continuous anatomical motions, aiding computer-aided interventions.
• VolMoVis shows significant potential for applications such as image-guided radiation therapy and 4D tumor tracking.