This project can be adapted to a Bachelor or Master Thesis.
Language is English or German!
Start: As soon as possible (01.04.2019)
Digital measurements always contain errors due to the nature of digitalization. In camera-based recording systems the spatial information is discretized. This introduces a significant error in length measurement, which was shown on the institutes own in silico model creating a synthetic set of data with known ground truth (see figure). This error has a large impact especially in the clinical context, where accurate decisions need to be taken. To minimize the error hardware as well as software solutions are possible. Both options should be analyzed and first of all compared within their own class (HW & SW) and then combined. The result helps first of all to understand the dependencies of the accuracy of length measurements and their error sources and then it will help biomedical engineers in developing new solutions to assess a e.g. vessels geometry. This provides the clinical staff assistance in surgeries and can probably raise an interventions quality.
The project consists of a literature research on discretization errors in image processing, their dependencies on various parameters and interpolation methods. The institutes model needs to be extended to the requirements set up beforehand. Afterwards analysis of the results is needed. The implementation is done in MATLAB.
Extension of an existing model to create synthetic images. Analyze the effects of HW and SW changes on the length measurement. Contextualizing of the results.