Articulated Statistical Atlas Construction for the Analysis of Bone Erosion in Models of Rheumatoid Arthritis

Project completed in 2014. 

Supervisors:
Professor Ela Claridge, School of Computer Science
Dr Amy Naylor, Institute of Inflammation and Ageing
Dr Andrew Filer, Institute of Inflammation and Ageing
Dr Iain Styles, School of Computer Science

Background: Rheumatoid arthritis is an autoimmune disease affecting approximately 1% of the population. The disease is characterised by chronic inflammation of the synovial joints resulting in localised bone erosion, generalised bone loss (osteopenia/osteoporosis) and abnormal bone formation (osteophytes). If inflammation persists these changes in bone turnover cause permanent joint deformity and ultimately loss of function. Various models of rheumatoid arthritis are available and their use in research has led to the development of revolutionary treatments for the disease, most notably anti-TNF therapy [original paper: Williams et al. 1992].

Aim: The aim of this project is to develop a method by which to quantify bone changes in inflammatory arthritis models available at the University of Birmingham.

Methodology: High resolution MicroCT will be used to generate high resolution X-ray images of joints. These types of image have been used by other groups to show bone destruction in models of arthritis [Zwerina et al. 2007; Seeuws et al. 2010] but the lack of effective quantification of bone erosion from them means that only very large differences can be identified and progression of disease over time is difficult to measure.

The primary parameter of interest to be measured is the total bone volume lost or gained due to arthritis. Initially multiple control samples will need to be combined using non-rigid (elastic) registration to create a foot 'atlas' [Adeshina & Cootes 2010]. Sample/diseased images can then be co-registered to the atlas and volumetric differences, representing bone erosions and osteophytes, can be quantified. In addition, bone density at specific sites can be measured to identify any osteopenia which may occur. Other parameters can be derived using existing software packages (e.g. CTAn, CTVol, 3D slicer, Fiji (Image J), and CTVox).

Scientific outcomes: The methods developed will be applied to all of our arthritis models in order to identify genes which exacerbate or ameliorate bone destruction. This technique will improve our understanding of the disease process and has the potential to identify genes/proteins that could be future targets for therapy. New algorithms for characterisation of the arthritis-affected joints will be developed and will be considered for clinical diagnostic use.

Link to ethesis: http://etheses.bham.ac.uk/6349/