In this paper, we develop a deep neural network based jointclassification-regression approach to identify microglia, a residentcentral nervous system macrophage, in the brain using fluorescencelifetime imaging microscopy (FLIM) data. Microglia are responsible forseveral key aspects of brain development and neurodegenerativediseases. Accurate detection of microglia is key to understandingtheir role and function in the CNS, and has been studied extensivelyin recent years. In this paper, we propose a jointclassification-regression scheme that can incorporate fluorescencelifetime data from two different autofluorescent metabolic co-enzymes,FAD and NADH, in the same model. This approach not only represents thelifetime data more accurately but also provides the classificationengine a more diverse data source. Furthermore, the two components ofmodel can be trained jointly which combines the strengths of theregression and classification methods. We demonstrate the efficacy ofour method using datasets generated using mouse brain tissue whichshow that our joint learning model outperforms results on thecoenzymes taken independently, providing an efficient way to classifymicroglia from other cells.
展开▼
