Evaluation of bone formation in calcium phosphate scaffolds with μCT-method validation using SEM. (5th October 2017)

• Record Type: Journal Article
• Authors: Lewin, S; Barba, A; Persson, C; Franch, J; Ginebra, M-P; Öhman-Mägi, C
• Is Part Of: Biomedical materials. Volume 12:Number 6(2017:Dec.)
• Publication Date: 2017-10-05
• Subjects: calcium phosphate scaffolds -- micro-computed tomography -- bone formation -- image analysis -- segmentation; Biomedical materials -- Periodicals; 610.28
• DOI: 10.1088/1748-605X/aa801d
• Languages: English
• ISSNs: 1748-6041

Abstract

Abstract: There is a plethora of calcium phosphate (CaP) scaffolds used as synthetic substitutes to bone grafts. The scaffold performance is often evaluated from the quantity of bone formed within or in direct contact with the scaffold. Micro-computed tomography ( μ CT) allows three-dimensional evaluation of bone formation inside scaffolds. However, the almost identical x-ray attenuation of CaP and bone obtrude the separation of these phases in μ CT images. Commonly, segmentation of bone in μ CT images is based on gray scale intensity, with manually determined global thresholds. However, image analysis methods, and methods for manual thresholding in particular, lack standardization and may consequently suffer from subjectivity. The aim of the present study was to provide a methodological framework for addressing these issues. Bone formation in two types of CaP scaffold architectures (foamed and robocast), obtained from a larger animal study (a 12 week canine animal model) was evaluated by μ CT. In addition, cross-sectional scanning electron microscopy (SEM) images were acquired as references to determine thresholds and to validate the result. μ CT datasets were registered to the corresponding SEM reference. Global thresholds were then determined by quantitatively correlating the different area fractions in the μ CT image, towards the area fractions in the corresponding SEM image. For comparison, area fractions were also quantified using global thresholds determined manually by two different approaches. In the validation the manually determined thresholds resulted in large average errors in area fraction (up to 17%), whereas for the evaluation using SEM references, the errors were estimated to be less than 3%. Furthermore, it was found that basing the thresholds on one single SEM reference gave lower errors than determining them manually. This study provides an objective, robust and less error prone method to determine global thresholds for the evaluation of bone formation in CaP scaffolds.