Research Of Shape Recognition On Femoral Shaft Fracture Images

The occurrence of fractures causes great harm to the human body, seriously affecting people's normal life and learning. Clinically, femoral shaft fractures are common and frequently occurred. Weather the location of femoral shaft fractures are determined accurately is directly impact on the fracture treatment programs. Since the artificial interpretation of the images affected by doctor's experience,level,responsibility, the misdiagnosis rate of the fracture images is high. Therefore, using the computer to achieve the interpretation of the location of femoral shaft fractures, assist doctors for treatment has important significance.This paper makes X-ray fracture images of femoral as study objects, to determine the fracture injury site of the fracture image by studying on shape recognition which is on the basis of image segmentation. In order to facilitate the shape recognition of the fracture images, femur fracture was divided into complete and incomplete fracture two types, design algorithms according to the different characteristics of the two types of fracture.For complete femur fracture, first, get the line features of every fracture areas through area filling, mathematical morphology thinning and region labeling, and then use Hough transform to determine the main direction of the fracture areas. Later, rotate every fracture areas and integral projection after determining the main direction, design reasonable criteria curve stitching and analysis the combined curve. At last, establish feature interpretation criterion for the curve, achieve the automatic interpretation of the fracture injury site.For incomplete femur fracture, get the skeleton of the fracture image, through removing the branch to keep the main skeleton. Take a different line fitting method after analysis of the main skeleton. Establish horizontal integral projection to the fracture image and according to the tilt angle of the skeleton obtain from the linear fitting to calculate the real width of the bone area. Revise feature interpretation criterion for the curve and achieve the automatic interpretation of the fracture injury site.Experimental results demonstrate the validity and accuracy of the algorithm.