Labeling White Matter Using Region Growing
Based on the estimates of the white and grey matter peaks in the previous step, the Segment white matter (region growing) step performs a region growing process to label white matter voxels. The outer boundary of this labelling process results in an initial representation of the cortical sheet of both hemispheres at the white / grey matter border.
The region growing process needs a minimum and a maximum intensity value as well as a "seed" voxel to start the growing process. The minimum value is determined automatically based on the intensity histogram analysis described in the previous step, the maximum value is set to 225, which is the highest intensity value used (white). The detected white matter voxels are set to color value 240 (visualized as blue). The seed voxel is determined by a scanning process starting in the centre of the brain until a voxel is found with an intensity value larger than the determined WM-GM threshold. From the determined seed voxel, neighboring voxels (6-neighborhood) are considered for inclusion into the white matter segmentation. A neighboring voxel is only added to the WM segmentation, if its intensity is larger than the intensity threshold. All included voxels themselves check their neighbors for inclusion using the same threshold. This process continues until no more neighbor voxel is found passing the threshold. The snapshot below shows a typical result of the region growing step with blue voxels identifying the segmented white matter voxels.
After the region growing step, the grey matter voxels and remaining head tissue is still visible. These intensity values are set to zero resulting in a representation containing only the labelled blue white matter voxels (see snapshot in next step).
Copyright © 2017 Rainer Goebel. All rights reserved.