Improving 3D Thin Vessel Segmentation in Brain TOF-MRA via a Dual-Space Context-Aware Network

Area of Science:

• Medical Imaging
• Computer Vision
• Biomedical Engineering

Background:

• 3D cerebrovascular segmentation is complex, often challenging due to the intricate nature of blood vessels.
• Traditional segmentation methods struggle with fine vessel structures and background noise.
• Magnetic Resonance Angiography (MRA) provides detailed vascular data but requires sophisticated processing for accurate segmentation.

Purpose of the Study:

• To develop an accurate and efficient method for 3D cerebrovascular segmentation.
• To improve the segmentation of fine vessel structures in MRA volumes.
• To reduce the computational burden and enhance the accuracy of vessel segmentation.

Main Methods:

• Proposed a Dual-space Context-Aware Network (DCANet) for 3D vessel segmentation.
• Utilized Maximum Intensity Projection (MIP) to create a vessel-segmentation-friendly 2D space (Regional-MIP).
• Implemented a Regional-MIP Image Fusion Block (MIFB) for integrating dual-space features and a Dual-mask Spatial Guidance TransFormer (DSGFormer) decoder.

Main Results:

• DCANet demonstrated superior performance across four datasets (TubeTK, IXI-IOP, Xiehe, IXI-HH).
• Achieved significant improvements in Dice Similarity Coefficient (DSC) for thin vessel segmentation, with increases of at least 2.17%.
• The dual-space approach effectively captured fine vessel structures and reduced background interference.

Conclusions:

• DCANet offers a robust solution for 3D cerebrovascular segmentation, outperforming existing methods.
• The integration of MRA and Regional-MIP spaces enhances the capture of intricate vascular details.
• The proposed method shows promise for clinical applications requiring precise vessel segmentation.