Detection of hepatic tumours on Multi-phase CT images for surgical plan

Multi-phase CT images offer clinicians important signs about hepatic lesions, and accurate segmentation of liver components is very helpful to the surgical plan for liver donor, transplantation or cancer resection. Based on edge detection and subtraction processing methods, we propose an automatic method for segmentation of liver region, extraction of hepatic tumors, including the detection of Hepatocellular carcinoma (HCC). The sign of HCC is shown high intensity in arterial phase while low in equilibrium phase images by the effect of contrast material, while other tumors like the hepatic cyst, cavernous hemangioma or metastasis always have a different gray value to liver. Tumor candidates can be picked up by subtraction methods or referred to the difference between edge map and the liver region. Three frequency images are employed to eliminate false-positive (FP) candidates. Other FPs are further deleted by opening processing. Finally, an expanding process is applied to acquiring the 3D shape of tumors. 42 cases were used in this experiment, and the result demonstrates our edge-detection based method is effective in finding the cancer region by using the information from different phase images.