Segmentation of Bone Tissue from CT Images

Segmentation of the bone structures in computed tomography (CT) is crucial for research as it plays a substantial role in surgical planning, disease diagnosis, identification of organs and tissues, and analysis of fractures and bone densities. Manual segmentation of bones could be tedious and not suggested as there could be human bias present. In this paper, we evaluate some existing approaches for bone segmentation and present a method for segmenting bone tissues from CT images. In this approach, the CT image is first enhanced to remove the artifacts surrounding the bone. Subsequently, the image is binarized and outliers are removed to get the bone regions. The proposed method has a Dice index of 0.9321, Jaccard index (IoU) of 0.8729, a precision of 0.9004, and a recall of 0.9662.