DEVELOPMENT OF A SEMI-AUTOMATIC SOFTWARE FOR DETECTION AND CHARACTERIZATION OF NEURONAL CELL MORPHOLOGY (NEUROTRACKER)

Abstract

This thesis presents the optimization and improvement of the Neurotracker software developed by Dimitri Simon for the German Aerospace Center in 2017. While one of the main threats in spaceflight is cosmic radiation the Medical Institute of the German Aerospace Center in cologne studies the effects on the human body and the influence of cosmic radiation specifically on neuronal cells. To inspect morphological changes of these cells, microscopy images of stained fluorescent cells were acquired and evaluated. Neurotracker is a semi-automatic software with the purpose to speed up evaluation of these images. It makes use of the OpenCV code library for image processing and object recognition. The software required to be optimized in different aspects as the segmentation was not recognizing cells properly under some circumstances and filter and tracking methods needed a long computation time. The software was therefore extended and optimized. Tools for segmentation using the watershed segmentation algorithm were added and filters were optimized in performance by multithreading. Additional options for morphological operations and correction of recognized objects were implemented. A user interface to organize and save data about fluorescence staining was added and the dendrite tracking was optimized by adjusting the parameters of the algorithm. The software was evaluated with cells and by measuring speed of methods on different devices. The user interface was optimized in cooperation with users and in assistance of their Feedback.