MUELLER MATRIX MICROSCOPY AND POLARIZATION DIGITAL PATHOLOGY
In digital pathology, stained histopathology slides are digitized by scanners to generate digitized whole-slide images (WSI). Then image processing and classification based on either imageomics or artificial intelligence (AI) approaches are used for tasks such as object detection and segmentation, as well as predicting disease diagnosis and prognosis of treatment response on the basis of patterns in the images. It has been known that Mueller matrices encode rich information on the microstructural features of complex samples, such as histopathology features of tissues. WSI by a Mueller matrix microscope is expected to reveal more abundant information in both the polarization and image features for medical doctors making more objective decision in clinical diagnosis. In this talk, we present a brief summary on our continuous efforts to realized polarization digital pathology. We have developed Mueller matrix microscopes by adding polarization optics in the existing optical path of commercial upright transmission optical microscope, and used them for taking Mueller matrix WSI of both stained and unstained pathological slides from adjacent sections of the same tissues. Diagnosis by medical doctors based on high resolution color images of the stained slides are used as gold standard for supervised learning to extract the corresponding polarization features. Although the results are still preliminary, we have proved that taking into account both the polarization and image features results in significant improvements in the performance of digital pathology.
File with abstract