Adaptive thresholding of tomograms by projection distance minimization

Abstract

Segmentation is an important step to obtain quantitative information from tomographic data sets. However, it is usually not possible to obtain an accurate segmentation based on a single, global threshold. Instead, local thresholding schemes can be applied that use a varying threshold. Selecting the best local thresholds is not a straightforward task, as local image features often do not provide sufficient information for choosing a proper threshold. Recently, the concept of projection distance was proposed by the authors as a new criterion for evaluating the quality of a tomogram segmentation [K.J. Batenburg, J. Sijbers, Automatic threshold selection for tomogram segmentation by reprojection of the reconstructed image, in: Computer Analysis of Images and Patterns, in: Lecture Notes in Computer Science, vol. 4673, Springer, Berlin/Heidelberg, 2007, pp. 563–570.]. In this paper, we describe how projection distance minimization (PDM) can be used to select local thresholds, based on the available projection data from which the tomogram was initially computed. The results of several experiments are presented in which our local thresholding approach is compared with alternative thresholding methods. These results demonstrate that the local thresholding approach yields segmentations that are significantly more accurate compared to previously published methods, in particular when the initial reconstruction contains artifacts.