Authors: Ankur Biswas; Paritosh Bhattacharya; Santi Prasad Maity
Addresses: Tripura Institute of Technology, Narsingarh, P.O. Agartala Aerodrome, Tripura (W), 799009, India ' National Institute of Technology, Agartala, Barjala, Jirania, Tripura (W), 799055, India ' Indian Institute of Engineering Science and Technology, Shibpur Howrah, 711103, West Bengal, India
Abstract: An efficient segmentation of liver tumour and volumetry facilitates medical experts to resolve the rate of tumour growth and follow up treatment. In this paper, a smart system for semi-automatic segmentation of liver tumour from medical images based on Geodesic active contours using level set method is proposed. The system passes through following stages. After the image is loaded, the Region of Interest (ROI) is selected and the Geodesic Active Contours with level set methods are initialised. Then segmentation result is updated to extract the tumour in three dimensions. The liver tumours detected by the system were compared with those delineated manually by experts, used as the ground truth results. The system was evaluated on Computed Tomography images of different datasets of tumours and compared with other methods. The result of proposed system obtained the Dice SC 0.948 and Jaccard SC 0.902 exhibits the steadfastness and effectiveness of the system.
Keywords: semi-automatic segmentation; medical images; geodesic active contour; level set; ROI; region of interest; computed tomography; dice SC; Jaccard SC.
International Journal of Product Development, 2019 Vol.23 No.2/3, pp.89 - 104
Received: 13 Jan 2018
Accepted: 01 Oct 2018
Published online: 02 Apr 2019 *