Image Analysis is a topic
becoming increasingly important in medicine, especially in the
neuroscience domain. Often, the task ranges from basic segmentation to
classification as the basis for medical decision making. Before the
image processing can start, often a pre-processing takes place, e.g., to
register images, i.e., all images follow the same coordinate system,
but also anonymization and pseudomization in a way the data can be used
without any privacy concern when it is processed.
The goal of
this thesis is to develop and evaluate a library implementing different
algorithms for anonymization and pseudomization techniques for
neurological MRI image data. Typically, this is subsumed by terms like
defacing and skull stripping. The intended algorithms should be
implemented in Python to be re-usable as part of a larger pre-processing
pipeline.
Contact person: Lars Hempel <lars.hempel@medizin.uni-leipzig.de>