A complete automatic region growing method for segmentation of masses on ultrasound images

Ultrasonography is a technique which helps the physician to visualize the structures of internal organs of human body. In some cases it is difficult to identify the boundaries of abnormal regions in the image. Hence segmentation is necessary. The manual methods of segmentation needs high attention of sonographer and also suffer from poor accuracy and time consuming. The automatic segmentation of Ultrasound images can help the physician by providing accurate location of abnormal regions. This paper describes a complete automatic technique for segmentation of masses on ultrasound images by using region growing technique. This is a best method for segmentation of ultrasound images because it is not affected by speckle noise and also preserves spatial information. In this paper the seed point has been automatically selected from the abnormal region based on textural features such as co-occurrence features and run length features. Then the threshold to control the region growing process is automatically selected and the Gradient magnitude based region growing algorithm is adopted. This method has been tested on different abdominal masses such as cyst and liver tumors. The result shows that the proposed region growing algorithm is efficient in selecting seed point as well as segmenting the region of interest without manual intervention.