Automatic recognition of subject-specific cerebrovascular trees

Area of Science:

• Medical Imaging
• Neuroscience
• Biomedical Engineering

Background:

• Current magnetic resonance imaging (MRI) techniques struggle to visualize small cerebral blood vessels due to contrast suppression and image distortions.
• Incomplete visualization of cerebrovascular angioarchitecture limits diagnostic information for physicians and hinders detailed hemodynamic analysis.

Purpose of the Study:

• To introduce a novel Hessian-based filter for enhanced contrast and accurate reconstruction of cerebrovascular trees from MRI data.
• To develop a method for blood vessel reconstruction that avoids dangling segments and improves the detection of small vessels.

Main Methods:

• A Hessian-based filter was developed for contrast enhancement in MR angiography and venography.
• Filter performance was quantified using receiver-operating-characteristic and dice-similarity-coefficient analyses.
• Validation involved calculating total extracted vascular length, number of segments, volume, surface-to-distance, and positional error.

Main Results:

• Reconstruction of cerebrovascular trees from six volunteers revealed more complete subject-specific vascular networks.
• Phantom model validation demonstrated the filter's ability to detect blood vessels across all length scales without failure at bifurcations or diameter distortion.
• The filter successfully rendered more complete representations of cerebrovascular networks.

Conclusions:

• The novel filter can improve the diagnosis of cerebrovascular diseases by providing detailed vascular metrics and anatomy.
• Automated analysis of large datasets is facilitated by computing operator-subjectivity-free biometrics.
• High-quality vascular reconstruction enables subject-specific hemodynamic simulations.