40 Chapter 3 WMH segmentation outside of the brain: n = 6; ventricle segmentation failed: n = 4); artefacts: n = 30; infarcts > 1.5 cm: n = 460; tumor: n = 12; technical error: n = 7; traumatic brain injury: n = 2). WMH oversegmentation outside of the brain was automatically corrected using brain masks. Another 1672 participants were excluded because of missing follow-up data (death: n = 505 (30%), disability or refused: n = 859 (51%), lost to follow-up: n = 104 (6%), claustrophobia: n = 86 (5%), MRI contradictions: n = 116 (7%), technical issues: n = 2 (0.1%)). A total of 2297 participants were included in the current study. Figure 3.1 Flowchart illustrating the exclusions of the study. Participants with infarcts bigger than 1.5 cm were excluded to avoid false positive WMH segmentation. WMH oversegmentation outside the brain was automatically corrected using brain masks. 3.3.2 MRI acquisition protocol The same MRI scanning protocol was used at baseline and follow-up on a 1.5 Tesla Signa Twinspeed system (General Electric Medical Systems, Waukesha, Wisconsin). Sequences included in the protocol: a 3D T1-weighted, spoiled-gradient echo (repetition time = 21 ms; time to echo = 8 ms; field of view = 240 mm; slice thickness = 1.5 mm; voxel size = 0.94 × 0.94 × 1.50 mm3); a fluid attenuated inversion recovery (FLAIR) (repetition time = 8000 ms; time to echo = 100 ms; field of view = 220 mm; voxel size = 0.86 × 0.86 × 3.00 mm3); a 2D T2*-weighted gradient-echo-type echo planar
RkJQdWJsaXNoZXIy MTk4NDMw