A Robust System for Segmentation of Primary Liver Tumor in CT Images

Abstract

The liver is a vital organ in human body, and Liver Tumor is considered to be a fatal disease. The tumors which can occur in Liver are cancerous or non-cancerous. For diagnosis of tumor, detection and demarcation of tumor is the initial step to be performed. After detection of the tumor, its type can be determined by using technique like biopsy, which is an invasive technique. To avoid such an invasive diagnosis technique, Non invasive techniques like diagnosis based on Medical Images using a CAD system can also be used. In such CAD systems, Detection and Segmentation of Tumor is performed automatically or semi-automatically. In this work, a system is developed to perform Segmentation of the Liver Tumor from abdominal CT image. This system segments the tumor in the two level operation. The first level of operation is segmentation of Liver from abdominal CT image and the second level is segmentation of Tumor from the result of first level. Segmentation of Liver is performed by using two methods namely Adaptive Thresholding with Morphological operations and Global Thresholding with morphological operations. Whereas segmentation of Tumor is performed by using three methods namely Adaptive Thresholding with Morphological operations, Fuzzy C Mean Clustering and Region Growing. This segmentation application generates and compares the outcomes of these implemented techniques. The system compares and selects the best of all the Tumor segmentation results and produces the final result. In this work, a robust system is proposed, by improving the accuracy of the segmentation for distinct quality and category of abdominal CT images, which contain liver tumors.

Segmentation of cancerous regions in liver using an edge-based and phase congruent region enhancement method

“…(14). By combining the non-feature image with the feature map, the boundaries of desired ROI are now enhanced, which otherwise are not enhanced (weak boundaries and low in contrast).…”

“…Linguraru et al [13] used shape correction methods in geodesic active contours for accurate segmentation of tumour and liver. Patil et al [14] designed semiautomatic segmentation method that employed multiple levels of thresholding for segmentation of tumour from liver. Smeets et al [15] used semiautomatic level set technique for segmentation of liver tumours in CT images in which the statistical pixel classification algorithm is used for the evolution of level sets.…”

CT liver tumor segmentation hybrid approach using neutrosophic sets, fast fuzzy c-means and adaptive watershed algorithm

“…This method requires manual selection of seed points. Patil et al in [27] proposed approach to segment tumors in the two-level operations. First level Segmentation of liver is performed by using two methods of adaptive threshold with morphological operations and global threshold with morphological operations.…”