A hybrid segmentation approach based on Neutrosophic sets and modified watershed: A case of abdominal CT Liver parenchyma

Liver cancer is one of the most common internal malignancies worldwide and also one of the most leading death causes disease. Early detection and accurate staging of liver cancer is considered an important issue in practical radiology. In this paper, a hybrid segmentation approach based on the modified Watershed algorithm and Neutrosophic logics is proposed for liver segmentation from abdominal CT images. The proposed approach consists of three fundamental phases: (1) preprocessing, (2) CT image transformation to Neutrosophic domain and (3) post-processing phase. At preprocessing phase, histogram equalization and median filter are applied to enhance the contrast and intensity values of the liver CT image as well as removing the noise. The enhanced CT liver image is transformed and represented in the Neutrosophic set domain via three membership sets. Finally, at post-processing phase, mathematical morphology and modified watershed algorithm are used to enhance the obtained truth image produced from the previous phase and to extract liver from CT image. Several measurements are used to evaluate the performance of the proposed segmentation approach. It obtains overall accuracy almost 95%. Moreover, it compared with other approaches and achieves better results.