Transfer learning for cell image reconstruction
Type: Bachelor EE/CS/HMI
If you are interested please contact :
Methods for the automated analysis of bio-medical images, and in particular microscope cell images, have to deal with unexpected corruptions and perturbations in the images to be analyzed, which deviate from the characteristics and properties of the training images.
This is, for instance, the case of image analysis systems trained on images taken with a specific microscope device (or by a certain human operator) and then deployed on devices that have different settings or managed by a different operator. Hence, it is important to study the robustness of existing algorithms when applied to images taken with different devices.
The goal of this project is to study the generalization abilities and cross-data-set performance of AutoEncoder models for cell image reconstruction, trained on different cell data sets .