Automatic image processing pipeline for tracking longitudinal vessel changes in magnetic resonance angiography.

Area of Science:

• Neuroimaging
• Medical Engineering
• Quantitative Biology

Background:

• Longitudinal magnetic resonance angiography (MRA) enables objective quantification of cerebral vessel diameter changes.
• Automated analysis of MRA facilitates vessel change assessment and downstream modeling.

Purpose of the Study:

• To characterize longitudinal changes in intracranial vessel diameter using time-of-flight (TOF) MRA.
• To develop and validate an automated method for measuring cerebral vessel diameters from MRA data.

Main Methods:

• Retrospective analysis of 112 pediatric patients with craniopharyngioma undergoing annual TOF MRA (1.5T and 3T) from 2006-2011.
• Automated vessel segmentation and diameter measurement at 1-mm intervals.
• Quantile regression used for outlier detection, with interventional angiography as a reference standard.

Main Results:

• Successful segmentation of major Circle of Willis vessels, excluding posterior communicating arteries due to hypoplasia.
• Median computed vessel diameter was 1.25 mm, with registration distortion < 0.04 mm in 99% of segments.
• Outliers (<4.34%) were more frequent in smaller vessels and near bifurcations (P < 0.001).

Conclusions:

• The proposed method allows noninvasive, objective assessment of cerebral vessel diameter changes from routine longitudinal MRA.
• High-throughput analysis of vascular trees combined with clinical data enables rigorous modeling of vessel diameter alterations.