BIANCA

BIANCA segments and quantifies white matter hyperintensities (WMHs) from MRI using a supervised k-nearest neighbour (k-NN) algorithm to support studies of neurological and vascular disorders.

Key Features:

• Automated Segmentation: Performs automated detection and segmentation of WMHs from MRI and provides quantitative WMH volume estimates.
• Supervised Framework: Operates as a supervised classifier that leverages labeled training data for WMH classification.
• k-Nearest Neighbour (k-NN): Implements the k-NN algorithm for voxel-wise classification of abnormal intensities.
• Flexible Spatial Weighting: Provides options for weighting spatial information when classifying voxels.
• Local Spatial Intensity Averaging: Uses local intensity averaging to account for local variations and improve detection precision.
• Adaptable Training Point Selection: Allows selection of different numbers and locations of training points to suit diverse datasets.

Scientific Applications:

• Neurodegenerative disease research: Quantifies WMHs in studies of Alzheimer's disease and other forms of dementia.
• Vascular disease assessment: Measures WMH burden in investigations of small vessel disease and related vascular conditions.
• Cohort and cross-protocol studies: Enables WMH quantification across different MRI protocols and large cross-sectional population cohorts.

Methodology:

BIANCA applies a supervised k-NN segmentation with selectable training points, optional spatial weighting and local spatial intensity averaging; parameters were optimized and validated on two distinct datasets by adjusting settings to maximize overlap and volumetric agreement with manually segmented WMH masks, and validation showed volume estimates correlated with manual segmentation volumes and visual ratings.