Genu valgum is a cause of knee pain and early arthritis which requires therapeutic action with external braces or surgery at a young age. This paper describes an image processing workflow and validation study for automatic characterization of the valgus deformity (i.e. genu valgum) from tibio-femoral x-ray radiographs. We implement an image processing pipeline starting with basic filtering and bone segmentation, followed by application of a Hough transform to determine the centerline of the diaphyses of the femur and tibia based on which a TF subtended angle is measured for each leg. Feasibility of this workflow is demonstrated on 21 short TF radiographs. The automatically computed angles were highly correlated (r~2 = 0.85 and p<<0.001) to the ground truth with a mean absolute error as low as 1.97°.
展开▼