Image Processing Workflow for Characterising X-Ray Computational Tomography of Polymeric Scaffolds with Cells

Abstract

Micro-computational tomography (µCT) is a useful technique for acquiring 3-D imaging of tissue-engineered scaffolds for morphology characterisation and analysis of the mechanical interactions between scaffold and cells. We used synchrotron light µCT at Diamond Light Source (UK) to image jet-sprayed nonwoven fibrous scaffolds, with and without human adipose-derived stem cells.

Large-volume imaging was achieved by stitching 2×2 tiled datasets and reconstructing them into 1 mm³ volumes at sub-micron resolution, enabling clear scaffold segmentation from the background. However, cells and fibres produce the same X-ray attenuation, this provides challenges in segmentation between fibres and cells. A deep learning algorithm with morphological recognition was employed. It enabled rapid selective segmentation, which allowed the analysis of cell distribution and morphology, revealing that cells preferentially adhered and proliferated along in-plane structures at full scaffold colonisation. We hypothesise that the cells minimise energy expenditure by expanding in directions of least resistance.

This process for analysing tissue-engineered scaffold opens new avenues for rapid, non-destructive, high-resolution, large-volume characterisation to elucidate cell and structural interaction.