Statistical Shape and Density Modeling of Clinical Image Data, 18-R8039Printer Friendly Version
Inclusive Dates: 03/02/09 07/02/09
Background - The problem of increased risk of skeletal fractures caused by bone-mass loss in aging or disease is a major clinical problem. Although the mechanisms related to the increased risk of fracture are not well understood, animal studies have shown that it is likely that different combinations of bone shape and density distribution in humans can lead to similar bone strengths. In consideration of bones as structures, it is expected that combinations of geometry, material properties, and the distribution of material properties will lead to variations in the structural response to loading, bone strength and the potential of fracture in an individual bone.
Approach - The objective of this project was to apply SwRI-developed statistical shape and density modeling methods (developed under a previous project) to a subset (40 patient datasets: 20 suffered a femur fracture and 20 have not suffered a femur fracture within the follow-up period) of baseline clinical image data to identify trait combination variables (e.g., eigenvalues or principal components) that differentiate patients who have fractured from those who have not fractured.
Accomplishments - The results of this application of statistical shape and density modeling methods to a clinical data set suggest that combinations of high fidelity trait combination variables describe inherent combinations of bone traits that may lead to different mechanisms of fracture (and prevention of fracture). In three specific cases of comparison between a clinical participant with a femur fracture and a clinical participant without a femur fracture, trait combination variables formed using SSDM methods were able to distinguish between the bone that suffered a fracture and the bone that did not. In these comparisons, trait combination variables were able to predict the bone with increased likelihood of fracture when a bone mass density, strength, and load-to-strength did not perform as well.