Semantic segmentation method for CBCT knee joint image

Abstract

The invention discloses a semantic segmentation method for a CBCT knee joint image, and the method comprises the steps: (1) obtaining a CBCT image of a target part of a patient, and carrying out the segmentation of each section through employing a maximum entropy threshold value, and obtaining a foreground region and a background region in the CBCT image; (2) segmenting the foreground region in each slice through an image segmentation algorithm to obtain a binary image of a skeleton region in the foreground region, and further obtaining a binary segmentation image of the target part of the patient; (3) calculating a signed distance field of the binary image of the skeleton region as a feature map for marking the watershed; marking the femur, the tibia, the fibula and the patella in the binary segmentation image according to the characteristics of the bone area in the knee joint to obtain a marked graph; and respectively adding semantic tags to the femur, the tibia, the fibula and the patella of the knee joint in the CBCT image by using a watershed marking method. The problems that in the prior art, the segmentation speed of the thighbone and the tibia of the CBCT knee joint image is low, and the segmentation effect is unstable can be solved.

Images