Segmentation of Prostate Sonoelastography Images Using Quantitative Elasticity Measures

Non-invasive ultrasound imaging is an emerging approach for detection of prostate cancer. However appropriate segmentation of ultrasound images of prostate tissue is challenging due to their inherent complex structure and heterogeneity. To overcome such limitation, sonoelastography is a promising method for cancer detection based on variation in tissue stiffness. In this study, a scaling based segmentation algorithm has been developed for sonoelastography (SE) images of prostate. Proposed algorithm converts the input color SE image in the form of a parametric image by representing pixels in terms of elasticity values. Further, segmentation of images was performed through Otsu's multilevel method. Segmentation results are verified by expert clinician justifying the validation of the method for probable use in segmentation of SE images.