According to the latest figures from Alzheimer Switzerland, there are currently 48,242 women and 15,711 men aged 85 years and older living with dementia in Switzerland. So-called White Matter Hyperintensities (WMH), which can be measured in MRI images of the brains white matter, are distinct biomarkers of dementia diseases. The techniques currently used to quantitatively determine WMH allow statements such as "Mr. Meier has more WMH in the visual cortex than Mr. Miller". However, the quantitative change of WMH over time, which would allow important conclusions to be drawn about the course of the disease, is not possible with the current state of the art. Statements such as "The volume of WMH in Mr. Müller's visual cortex has increased by 12% in the last 6 months" still cannot be estimated quantitatively, but only qualitatively through visual inspections by radiologists. In this project, an intelligent algorithm for determining the longitudinal change of WMH is to be developed. The following difficulties shall be addressed:

• The coregistration of the data sets from different points in time must be normalized.
• The change in the volume of the WMH can be very small and requires an accuracy and precision not achieved so far. Simple differences are not sufficient, since the WMH change morphologically.
• The threshold for the dichotomous identification of the WMH is based on the individual intensity distribution of a data set.
• For longitudinal comparisons an overall threshold must be found, because even with the same acquisition sequence the average intensity varies.

The algorithm will then be used in various studies to investigate the relationship between morphological changes in the WHM and the course of dementia.