Towards histopathological analysis based on X-ray fluorescence elemental imaging supported by multivariate analysis - Case study of ovarian cancers

The goal of this study was to verify if the segmentation of elemental distribution images can differentiate various types of tissues in the interface between stromal and neoplastic area of the ovarian tumour tissue. Distribution of minor and trace elements ranging from carbon to zinc of seven human ovarian tissue samples was obtained by using micro X-ray fluorescence imaging technique. The K-means clustering and non-negative matrix factorization were applied for image segmentation of acquired elemental distribution maps into three sub-regions corresponding to stromal area, tumour area and unclassified area. The resulted images were confronted with the results of the histopathological examination.