PURPOSE: Numerous orthopedic disorders involve periaxial rotation of long bones. We have developed and evaluated a computer-assisted method that segments a bone from X-ray CT, graphically unwraps the bone around its long axis into a surface plot (signature landscape), and measures periaxial rotation as the translation shift of the landscape. Bones in known rotations and partially segmented surfaces were used, respectively, to test accuracy and problematic situations in bone segmentation. METHOD: CT images of three chicken femora at known rotations were analyzed to determine their relative periaxial rotations, which were compared with the known rotations. RESULTS: The regression slope between measured and expected periaxial rotations was 1.005 +/- 0.003, with a maximum discrepancy of 2 degrees for same-bone and 3 degrees for interbone comparisons. Rasterization artifacts and the use of partial surfaces (with < 40% surface omission) resulted in a < 3 and < 1 degree error, respectively. CONCLUSION: The current method provides accurate and objective periaxial rotation measurements of a long bone.
PURPOSE:Numerous orthopedic disorders involve periaxial rotation of long bones. We have developed and evaluated a computer-assisted method that segments a bone from X-ray CT, graphically unwraps the bone around its long axis into a surface plot (signature landscape), and measures periaxial rotation as the translation shift of the landscape. Bones in known rotations and partially segmented surfaces were used, respectively, to test accuracy and problematic situations in bone segmentation. METHOD: CT images of three chicken femora at known rotations were analyzed to determine their relative periaxial rotations, which were compared with the known rotations. RESULTS: The regression slope between measured and expected periaxial rotations was 1.005 +/- 0.003, with a maximum discrepancy of 2 degrees for same-bone and 3 degrees for interbone comparisons. Rasterization artifacts and the use of partial surfaces (with < 40% surface omission) resulted in a < 3 and < 1 degree error, respectively. CONCLUSION: The current method provides accurate and objective periaxial rotation measurements of a long bone.