|
Crohn's disease (CD) and gastrointestinal luminal tuberculosis (ITB) are two kinds of similar inflammatory bowel diseases, whose incidences are growing rapidly worldwide. Due to the lack of a general gold standard to distinguish between CD and ITB samples, misdiagnosis often occurs in clinical detections, leading to inappropriate treatments and side-effects. The characteristic features of both CD and ITB tissues include tuberculosis and surrounding fibrous structures, which can be quantitatively evaluated by polarimetric techniques. In this study, we apply the transmission Mueller matrix microscope developed in our previous study on the CD and ITB tissue samples to attain their 2D Mueller matrix images. We calculate the Mueller matrix polar decomposition and transformation parameters, which can provide information about the location, density and distribution behavior of the tuberculosis areas surrounded by fibrous structures. In order to evaluate the different distribution behaviors of the fibrous structures quantitatively, we analyzed the retardance related Mueller matrix derived parameters images, which show different features between the CD and ITB tissues, using the Tamura images processing method (TIPM). The preliminary results show that the TIPM analysis of the retardance related parameters can provide some quantitative parameters to describe the different textures of fibers in the CD and ITB tissues. Moreover, we use the machine learning method based on Mueller matrix derived parameters to distinguish between CD and ITB tissues. It is demonstrated that the Mueller matrix derived parameters combined with machine learning methods can be helpful for clinical diagnosis.
|
