Evaluation of tumor segmentation
Sonam Bansal, Ajay Mittal · 2013 Fourth International Conference on Computing, Communications and Networking Technologies (ICCCNT) · 2013
X-ray images are commonly used for diagnosis because of the property that different tissues absorb X-rays differently. Detection of lytic areas using X-ray images is a challenging problem due to irregular shapes, intensity in-homogeneity and diffused boundaries. The problem has been addressed by segmenting lytic area on the basis of different cues such as intensity, density, and texture. However, the approaches based on single cue have failed to produce the desired results. In this paper, a hybrid technique has been tested to segment lytic region using X-ray images. The hybrid method has been qualitatively evaluated by taking assistance from the domain experts. The results of evaluation indicate that the hybrid approach shows better result as compared to single cue techniques.