Automatic Segmentation of Bones in X-ray Images Based on Entropy Measure

Abstract

In this paper, we introduce an efficient method for segmenting the bone region of an X-ray image from its surrounding muscles and tissues. Automated segmentation of the bone part in a digital X-ray image is a challenging problem because of its low contrast with the surrounding flesh. The presence of noise and spurious edges further complicates the segmentation. Most of the existing methods either suffer from noisy contour detection or need training samples for manual tuning of certain thresholding parameters. We propose a fully automated segmentation technique, which utilizes a variant of entropy measure of the image. This scheme has been shown to be useful for fast and efficient analysis of a wide class of human X-ray images including skull, chest, pelvic region and ortho-dental zones. In order to quantify the quality of segmentation, we propose a new metric called average contour distortion index (ACDI) based on certain neighborhood properties of the contour pixels. Experiments on several X-ray images reveal encouraging results compared to other approaches as evident from the ACDI metric. We also re-validate the quality of several segmented bone images using segmentation entropy quantitative assessment (SEQA), and boundary-based precision–recall profile. All three metrics establish the superiority of the proposed technique to prior art.