An Investigation of Bone Image Texture Analysis for Predicting Fracture Risk
|Arnason, Neil (Computer Science) Leslie, William (Internal Medicine) Goertzen, Andrew (Physics & Astronomy)
|Walton, Desmond (Computer Science)
|Master of Science (M.Sc.)
|Osteoporosis is caused by loss of bone mineral content, which leads to bone fractures or structural deformations of bone. Osteoporosis usually occurs when people get older, after menopause in women, or it can be caused by a lack in the intake of a sufficient amount of calcium and vitamin D. Until recently, osteoporosis was considered to be an unavoidable part of aging, but today, approved and effective treatments can be used to deal with the consequences. At present, determination of risk of bone abnormalities is done by measuring the density of bone (largely determined by calcium content). Dual energy X-ray Absorptiometry (DXA) is the gold standard technique for measuring bone mineral density (BMD). Even though BMD is one of the principal determinants of bone strength, BMD measurements do not give information about variation of trabecular structure of bone. That's why DXA alone has limited ability to predict who will sustain an osteoporotic fracture. To predict fracture risk of patients, the texture analysis of the DXA images is of interest as a measure to predict fracture in addition to BMD. This thesis focuses on the application of texture analysis to digital images of bone scans of patients at risk of fracture and osteoporosis. Texture analysis was performed by analyzing the variation of grey level patterns of pixels of DXA images. Texture analysis of such images will give an idea of the variation of grey scale patterns of pixels between normal and osteoporotic DXA images of bone. Existing texture analysis measures such as contrast measures of co-occurrence matrices and mean slope value of fractal dimension based measure are used to analyze the texture of DXA images. An alternative partitioning technique is proposed as a measure of the texture analysis.
|An Investigation of Bone Image Texture Analysis for Predicting Fracture Risk