Liver cancer is one of the most common cancers that causes many deaths every year. In recent years, the risk of men and women getting liver cancer has increased by 40% and 23%, respectively. In order to identify a tumor in the liver, segmentation is performed on CT images. The use of data fusion methods in data mining techniques is one of the most practical methods to improve accuracy, which also has many applications in the field of medical image processing. Correct and efficient diagnosis of liver abnormalities leads to a significant reduction in human error and a more accurate diagnosis by physicians. This requires the use of methods based on automatic and semi-automatic detection. Combining clustering methods and considering cluster uncertainty is an appropriate tool in solving clustering problems in medical image processing, especially cancer diagnosis. The proposed method, in addition to having high accuracy, has a high convergence speed.