A fully-automatic locally adaptive thresholding algorithm for blood vessel segmentation in 3D digital subtraction...

Area of Science:

• Neurosurgery
• Medical Imaging
• Computational Fluid Dynamics

Background:

• Subarachnoid hemorrhage from ruptured cerebral aneurysms remains a critical medical issue.
• Accurate cerebral vessel diameter assessment is vital for planning endovascular aneurysm therapies.
• Hemodynamic simulations require reliable vessel segmentation for effective treatment planning.

Purpose of the Study:

• To develop a fully-automatic, locally adaptive, gradient-based thresholding algorithm for cerebral vessel segmentation.
• To accurately assess vessel diameters for endovascular aneurysm therapy planning.
• To improve the reliability of hemodynamic simulations in neurovascular procedures.

Main Methods:

• Proposed a two-step algorithm: iterative parameter estimation for global thresholding followed by local adaptation.
• Evaluated the algorithm on 8 clinical 3D Digital Subtraction Angiography (DSA) datasets.
• Developed a method for reference segmentation selection using 2D DSA measurements.

Main Results:

• Achieved high sensitivity (97.4%), precision (98.7%), and Dice-coefficient (98.0%) for large vessels like the internal carotid artery.
• Demonstrated strong performance on smaller vessels (approx. 1mm diameter) with sensitivity (95.7%), precision (88.9%), and Dice-coefficient (90.7%).
• The automated segmentation method proved effective across various vessel sizes.

Conclusions:

• The proposed fully-automatic algorithm provides accurate cerebral vessel segmentation for endovascular aneurysm therapy.
• This method enhances the reliability of hemodynamic simulations by improving vessel diameter assessment.
• The algorithm offers a valuable tool for neurosurgeons and interventional radiologists in treating cerebral aneurysms.