Automated segmentation of computed tomography images of fiber-reinforced composites by deep learning

September 08, 2020

Aly Badran (1), David Marshall(1), Zacharie Legault (2), Ruslana Makovetsky (3), Benjamin Provencher (3), Nicolas Piché (3), Mike Marsh (3)

Journal of Materials Science, 55, 08 September 2020:16273–16289. DOI: 10.1007/s10853-020-05148-7


A deep learning procedure has been examined for automatic segmentation of 3D tomography images from fiber-reinforced ceramic composites consisting of fibers and matrix of the same material (SiC), and thus identical image intensities. The analysis uses a neural network to distinguish phases from shape and edge information rather than intensity differences. It was used successfully to segment phases in a unidirectional composite that also had a coating with similar image intensity. It was also used to segment matrix cracks generated during in situ tensile loading of the composite and thereby demonstrate the influence of nonuniform fiber distribution on the nature of matrix cracking. By avoiding the need for manual segmentation of thousands of image slices, the procedure overcomes a major impediment to the extraction of quantitative information from such images. The analysis was performed using recently developed software that provides a general framework for executing both training and inference.

How Our Software Was Used

Dragonfly was used for image manipulation and analysis and to design the convolutional neural network model used in this work.

Author Affiliation

(1) Department of Aerospace Engineering and Sciences, University of Colorado, Boulder, 3775 Discovery Dr. Boulder, Boulder, CO 80303, USA
(2) Department of Computer and Software Engineering, Polytechnique Montreal, Montreal, Canada
(3) Object Research Systems, Montreal, Canada