Open access publications in the SLU publication database

Abstract

Computerized image analysis, the discipline of using computers to automatically extract information from digital images, is a powerful tool for automating time consuming analysis tasks. In this thesis, image analysis and visualization methods are developed to facilitate the evaluation of osseointegration, i.e., the biological integration of a load-carrying implant in living bone.

Adequate osseointegration is essential in patients who are in need of implant treatment. New implant types, with variations in bulk material and surface structural parameters, are continuously being developed. The main goal is to improve and speed up the osseointegration and thereby enhance patient well-being. The level of osseointegration can be evaluated by quantifying the bone tissue in proximity to the implant in e.g., light microscopy images of thin cross sections of bone implant samples extracted from humans or animals. This operator dependent quantitative analysis is cumbersome, time consuming and subjective. Furthermore, the thin sections represent only a small region of the whole sample.

In this thesis work, computerized image analysis methods are developed to automate the quantification step. An image segmentation method is proposed for classifying the pixels of the images as bone tissue, non-bone tissue or implant. Subsequently, bone area and bone implant contact length in regions of interest are quantified. To achieve an accurate classification, the segmentation is based on both intensity and spatial information of the pixels. The automated method speeds up and facilitates the evaluation of osseointegration in the research laboratories.

Another aim of this thesis is extending the 2D analysis to 3D and presenting methods for visualization of the 3D image volumes. To get a complete picture, information from the whole sample should be considered, rather than thin sections only. As a first step, 3D imaging of the implant samples is evaluated. 3D analysis methods, which follow the helix shaped implant thread and collects quantified features along the path, are presented. Additionally, methods for finding the position of the 2D section in the corresponding 3D image volume, i.e., image registration, are presented, enabling a direct comparison of the data from the two modalities. These novel and unique 3D quantification and visualization methods support the biomaterial researchers with improved tools for gaining a wider insight into the osseointegration process, with the ultimate goal of improved quality of life for the patients.