IMAGE RECONSTRUCTION AND GEOMETRICAL MODELING OF BONE CROSS SECTIONS FROM LIMITED ANGLE DUAL ENERGY PROJECTION DATA

Image reconstruction;Geometrical modeling;Dual energy x-ray;Bone;Electrical Engineering

Khaled, Alia ShawkyTsui, Benjamin M. W. ;

Johns Hopkins University

关键词: Image reconstruction;    Geometrical modeling;    Dual energy x-ray;    Bone;    Electrical Engineering;   

Others  :  https://jscholarship.library.jhu.edu/bitstream/handle/1774.2/39296/KHALED-DISSERTATION-2014.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: JOHNS HOPKINS DSpace Repository

【 摘 要 】

Osteoporosis, the most common bone disease is characterized by reduced mechanical strength leading to increased bone fragility and fracture. Because susceptible bones tend to be less dense, clinical diagnosis is currently based on bone density measured typically by Dual-Energy X-ray Absorptiometry (DXA) or Quantitative Computed Tomography (QCT). Unfortunately, density criteria identify less than half of people at risk of osteoporotic fracture and do not adequately characterize the mechanical basis of fracture susceptibility. Identification of fragility could be greatly improved by the use of advanced 3D mechanical engineering analysis based on CT data. CT scanners, however, are expensive and its radiation doses are too high for routine screening of asymptomatic populations. To address these issues, a low dose multi-projection DXA scanner system is proposed that can develop a valid bone model for mechanical analysis from limited angle dual energy projection data. The proposed system presents multiple technical challenges addressed by this dissertation. The technical approach is based on image reconstruction constrained to three DXA distinguishable tissues present in human extremities: bone, adipose and lean tissues. Because DXA principles can discriminate only two tissues at once, we developed an algorithm to estimate the missing tissues in regions where three tissues are present. Next, using a small number of DXA projections over a rotational angle of 90˚ an algorithm was developed to reconstruct high quality discrete cross-sections. We can then construct 3D models of an extremity from a stack of multiple 2D DXA images to perform a 3D mechanical analysis. A critical aspect of the methods developed in this dissertation is the application of engineering analyses to 2D and 3D bone data. To demonstrate feasibility, engineering analysis was applied to previously acquired DXA and CT data from several clinical research studies of the femoral neck. The methods developed in this work, together with the reported applications in bone studies, constitute a frame work wherein low-dose imaging is combined with geometrical modeling to provide a more rational alternative to existing bone-density techniques, both in patient safety and analysis accuracy. The proposed methods could have many applications in the study of osteoporosis.