ART 3.5D: an algorithm to label arteries and veins from three-dimensional angiography

Area of Science:

• Neurosurgery
• Medical Imaging
• Image Reconstruction

Background:

• Preoperative 3-D visualization of brain vasculature is crucial in neurosurgery for identifying critical structures and guiding navigation.
• Standard 3-D reconstructions from computerized tomography (CT) angiography lack direct distinction between arteries and veins, requiring surgeon inference.
• Existing 4-D CT perfusion scans offer dynamic information but involve longer acquisition times, higher radiation doses, and reduced resolution.

Purpose of the Study:

• To present ART 3.5D, a novel method for dynamic information recovery from standard 3-D contrast-enhanced cone-beam CT (CE-CBCT) scans.
• To overcome the clinical limitations of 4-D CT by enabling artery and vein segmentation from 3-D CE-CBCT data.
• To improve preoperative visualization of brain vasculature for enhanced neurosurgical planning.

Main Methods:

• The ART 3.5D method reprocesses hidden contrast dynamics from raw 3-D CE-CBCT data to approximate dynamics within segmented voxels.
• A classification algorithm is employed to label the angiographic voxels as either artery or vein.
• Numerical simulations were conducted on a digital vascular phantom with simulated CE-CBCT projections for method testing.

Main Results:

• The ART 3.5D approach successfully recovered approximate contrast dynamics from standard 3-D CE-CBCT scans.
• A classification accuracy of up to 90% was achieved in simulations for distinguishing between arteries and veins.
• The study demonstrated the feasibility of dynamic information recovery for artery and vein segmentation using the proposed method.

Conclusions:

• ART 3.5D offers a promising solution for dynamic artery and vein segmentation from standard 3-D CE-CBCT scans in neurosurgery.
• This method overcomes limitations of 4-D CT, providing valuable preoperative insights without extended protocols or increased radiation.
• The technique has the potential to enhance surgical navigation and planning by providing clearer differentiation of vascular structures.