Publications & Research

Automatic region growing segmentation for medical ultrasound images

Ultrasound images are difficult to segment due to presence of speckle noise and the boundaries of abnormal regions are also difficult to recognize due to similarity. It is important to segment the image for correct and effective diagnosis. Manual method of segmentation is good but not effective for segmentation of large data sets, due to this an automatic or computerized segmentation is motivated. An automatic region growing segmentation for ultrasound images is presented in this work. An automatic selection of seed is adopted because of time consumption, poor accuracy and need of human interaction in manual seed selection. In ultrasound images, identification of the boundaries of abnormal regions is impossible but automatic seed selection provides accurate location of abnormal regions. The proposed method outperforms the existing state-of-the-art techniques based on the texture features and visual results.

Click here to read full article