Evaluating Tissue Mechanical Properties Using Quantitative Mueller Matrix Polarimetry and Neural Network

Evaluation of the mechanical properties biological tissues has always been an important issue in field biomedicine. The traditional method for measurement is to perform vitro tissue deformation experiments. With fast development optical and image processing techniques, more non-invasive non-contact methods have applied analysis features. In this study, we use Mueller matrix polarimetry quantitatively obtain bovine tendon tissues. Firstly, study structural information changes characteristics under different stretching states, 3 × images samples are acquired by backscattering setups based on a polarized camera. Then, extract frequency distribution histograms (FDHs) elements reveal clearly during process. Last, calculate transformation (MMT) parameters, total anisotropy t1 direction α1 processes characterize their states. central moments MMT parameters can be used distinguish states tissue. For better discrimination design multilayer neural network that takes first-order as input After training, high-precision classification model finally obtained, accuracy achieves 98%. experimental results show potential tool evaluation.