Living cells as phase objects require not only non-invasive measurement but also quantitative phase information during dynamic biopsy. Digital Holographic Microscopy (DHM), measuring three-dimensional morphology without changing the active condition of cells and in situ inspection, is becoming excellent tools for biology research. We have described a DHM method for quantitative, unlabeled observation of living cell subjected to fluid shear stress (FSS) in flowing fluid. The holographic recording system combined with the fluid shear system is improved. The numerical reconstruction technique firstly employed deep learning Convolutional Neural Network model filter, which achieved automatically processing large scale the spectrum of holograms immediately. Osteocytes as the experimental samples were observed and their morphological changes under the stimulation of FSS was successfully measured.