Jasmin Annica Kuhn-Keller

42 Chapter 3 Figure 3.2. Illustration of the WMH shape image processing pipeline. Lateral ventricles were segmented using T1-weighted MRI images. WMH segmentation was performed on the registered FLAIR images. Using two different inflated ventricle masks (3 mm and 10 mm), WMH were classified into three types (deep, periventricular and confluent). Based on the resulting WMH type, different WMH shape markers were calculated using the shown formulas. Occurrence of new subcortical brain infarcts, microbleeds, enlarged perivascular spaces, cerebellar infarcts, and cortical infarcts at follow-up were rated by trained radiographers by comparing the baseline and follow-up MRI scans.11 Follow-up MRI scans were examined and if a lesion was detected, the baseline scan was checked to see if the lesion was new. Microbleeds were scored if they were visible on the T2*-weighted scans.12 Infarcts were scored if they were visible on the FLAIR, T2-weighted and the proton-density scan.13 Cortical infarcts were labelled as such if they involved or were limited to the cerebral cortex and were surrounded by an area of high signal intensity on FLAIR images. Cerebellar infarcts had no size criteria. Subcortical infarcts were scored if they did not extend into the cortex and were surrounded by an area of high signal intensity on FLAIR images with a minimal diameter of 4 mm.13 Enlarged perivascular spaces were defined as round or tubular defects with a short axis >3 mm in the subcortical area, without a surrounding area with high signal intensity on FLAIR images.14 3.3.5 Baseline characteristics and cardiovascular risk factors Baseline information, such as age, sex, education level, and smoking status, were collected via questionnaires. The highest completed education level (primary school, secondary school, college, or university) was noted. Non-smokers were participants

RkJQdWJsaXNoZXIy MTk4NDMw