Intra-Scanner and Inter-Scanner Reproducibility of Automatic White Matter Hyperintensities Quantification

Area of Science:

• Neuroimaging
• Radiology
• Cerebrovascular Diseases

Background:

• White matter hyperintensities (WMH) are common in aging and cerebrovascular diseases.
• Accurate WMH quantification is essential for longitudinal and multicenter studies.
• Variability in WMH measurements can arise from various factors in magnetic resonance imaging (MRI).

Purpose of the Study:

• To evaluate the reproducibility of white matter hyperintensities (WMH) quantification using MRI.
• To examine the impact of scan-rescan procedures, scanner types, imaging protocols, software upgrades, and automated segmentation tools on WMH quantification.

Main Methods:

• Six post-stroke subjects underwent repeated MRI scans (3D T1-weighted, 2D and 3D T2-FLAIR) across four scanners.
• Two automated tools were used for WMH segmentation and volume measurement.
• Reproducibility was assessed using coefficient of variation (CV), Dice similarity coefficient (DSC), and intra-class correlation (ICC).

Main Results:

• 3D T2-FLAIR MRI provided the best intra-scanner reproducibility (CV: 2.69-2.97%).
• 2D T2-FLAIR MRI exhibited higher variability (CV: 15.62-29.33%) compared to 3D T2-FLAIR MRI, attributed to slice thickness.
• WMH segmentation tool choice significantly impacted quantification results, alongside image acquisition parameters.

Conclusions:

• WMH quantification reproducibility is a critical consideration for multicenter and longitudinal neuroimaging studies.
• 3D T2-FLAIR MRI is recommended for enhanced WMH measurement consistency.
• Standardized protocols and tool selection are vital for reliable WMH quantification in research settings.