Detecting and Tracking the Tips of Fluorescently Labeled Mitochondria in U2OS Cells

摘要

We present a method for automatically detecting the tips of fluorescently labeled mitochondria. The method is based on a Random Forest classifier, which is trained on small patches extracted from con-focal microscope images of U2OS human osteosarcoma cells. We then adopt a particle tracking framework for tracking the detected tips, and quantify the tracking accuracy on simulated data. Finally, from images of U2OS cells, we quantify changes in mitochondrial mobility in response to the disassembly of microtubules via treatment with Nocodazole. The results show that our approach provides efficient tracking of the tips of mitochondria, and that it enables the detection of disease-associated changes in mitochondrial motility.