A Fourier-optical approach for segmentation of ion microprobe images into tissue and non-tissue covered areas

Fourier optics provides a powerful and objective approach to handling Poissonian statistics in images. Fourier optics has been used to develop an approach to segment microbeam images into tissue and void (non-tissue) areas (e.g. villi, crypts, blood vessels and lymph canals etc.). Image segmentation is important in order to accurately measure the major element composition and thickness of biological sections. The method is based on comparing the spatial frequency of non-zero pixels in a reference void with tissue regions using a convolution method. The method automatically and conservatively segments tissue areas and uses no free-parameters. Since the method is based on spatial frequencies it can be used over a wide span of image counting statistics.