The growing number of women receiving drug treatment for osteoporosis underscores the need for sensitive methods to monitor therapy. Because it permits selective assessment of the trabecular and cortical bone, which may respond differently to disease and therapy, quantitative computed tomography (QCT) is well-suited to this purpose. However, the ability of QCT to resolve therapy- or disease-induced changes in bone mineral density (BMD) is limited by variable precision errors. These precision errors mostly relate to the operator-dependence of current techniques, which involve manual slice selection and region of interest placement. In our Phase I Grant, we have addressed this problem by developing and demonstrating the feasibility of an image registration algorithm which aligns serial images and uses the resulting 3D transformation to map a baseline region into the same volume of the follow-up image. As we have documented in the Phase I Final Report, this fast and automated approach yields precision errors in vivo of 1.0 mg/cc and 0.9 mg/cc for spinal and proximal femoral trabecular BMD measurements respectively. In this application, we propose to develop and commercialize a prototype software package which will combine diagnostic 3D QCT BMD measurements with our new approach for highly reproducible longitidunal measurements.
We will provide commercial CT imaging centers with a powerful set of tools which they can use to compete in the bone mineral density measurement market. The diagnostic measurement component will allow CT scanners to measure bone mineral density in the spine and hip in bone regions comparable to those measured by DXA. Moreover, the high reproducibility of our technique will permit CT centers to offer a method for monitoring therapy effects which will be superior to DXA.
| Li, Wenjun; Kornak, John; Harris, Tamara B et al. (2009) Bone fracture risk estimation based on image similarity. Bone 45:560-7 |
| Li, Wenjun; Kornak, John; Harris, Tamara et al. (2009) Identify fracture-critical regions inside the proximal femur using statistical parametric mapping. Bone 44:596-602 |
| Li, Wenjun; Kezele, Irina; Collins, D Louis et al. (2007) Voxel-based modeling and quantification of the proximal femur using inter-subject registration of quantitative CT images. Bone 41:888-95 |
| Li, Wenjun; Sode, Miki; Saeed, Isra et al. (2006) Automated registration of hip and spine for longitudinal QCT studies: integration with 3D densitometric and structural analysis. Bone 38:273-9 |
