Histological analysis with white light microscopy of few micron-thick sections of tissue remains a gold standard for tissue biopsy. However, this approach requires a cumbersome multi-step preparation of the sample and relies heavily on the skills of a pathologist. In general, any tissue pathology (e.g. inflammation, fibrosis, cancer) affects the constituents of tissue and changes both optical properties of cells and extracellular matrix. We suggest using the images of transmittance, retardance and depolarization of unstained tissue samples, acquired with the Mueller microscopy, as an input data for automated diagnostic image segmentation. The potential of combining polarimetric data with non-linear microscopy data, such as Second Harmonic Generation (SHG) and Two Photon Emission Fluorescence (TPEF), is also explored for the diagnostic of tissue specific components (e. g. collagen, elastin). The unstained histological cuts of full thickness skin equivalents and cervical tissue of mouse were used for multi-modal microscopy imaging. First, the statistical approach was tested for a segmentation of the images of artificial skin using the images of retardance, depolarization, and transmitted unpolarized intensity. The algorithm of density-based spatial clustering of applications with noise (DBSCAN) clearly discriminates the regions of dermal and epidermal layers of skin cuts. The same approach was tested on the extended set of data including polarimetric, SHG and TPEF images of the histological cuts of mouse uterine cervix. The typical circumferential arrangement of collagen fibers around the cervical os as well as elastin lining of the internal cavities was observed on the segmented images. Merging polarimetric data with SHG data makes clear the appearance of fine structures in the cervical os region. The work is ongoing on testing different data post-processing algorithms for increasing both contrast and accuracy of tissue image segmentation, thus, paving the way for the digital histology for tissue diagnostic.