Abdominal image segmentation on Android based mobile devices

Abstract

Medical services have a great importance in international development. Hospitals have been forced to reinforce their health services and technological infrastructure developments because of the rapid development of information technology. Liver segmentation is a difficult task because of its variable shape it can be thought as a large footprint. In addition, the nearness of its color and texture to surrounding organs tissues makes its boundaries unclear. In this study, computed tomography images have been used to segment the liver that was the first part of the lesion liver detection. Obtained liver segmentation result achievements have been compared with the manual segmentation of the radiologists. Segmentation accuracies have been assessed by using Zijdenbos similarity index. The applied methodology achieved approximately 93% segmentation accuracy. After liver segmentation stage, this segmentation procedure is located on mobile environment that medical experts can access the software with their mobile devices. This is the first part of the ongoing decision support system study that can be used to define diagnostic lesions on the liver via android-based mobile devices. © 2014 IEEE.