Morph-Rec: A Novel Computer-Aided Liver Segmentation Model based on Morphological Reconstruction Operation

Abstract

An abdominal Computed Tomography (CT) scan gives more information about diseases of the liver, gallbladder, and biliary tract. In the image processing approach, liver segmentation is an essential step to be done before liver lesion detection. Liver segmentation removes the unwanted regions from the CT image and makes the task of lesion detection easier. In this paper, a novel Morph-Rec model based on morphological reconstruction operation is proposed for liver segmentation from CT images. The proposed work is focused on segmenting the liver region from the CT slices irrespective of the size and shape of the liver region. The proposed model is validated on 2650 CT slices of 120 and 20 CT scans from the LITS and 3DIRCADb datasets, respectively. The proposed Morph-Rec method is evaluated using metrics such as dice score, accuracy, F1 score, Jaccard index and Matthew’s correlation coefficient. To justify the adaptability and efficiency of the proposed model, it is also validated on 50 CT slices of nine CT scans provided by a local scan centre. The proposed method has produced excellent results on all metrics and the obtained results are better than the state-of-the-art conventional methods for liver segmentation. Hence, the proposed technique is an automatic and dataset-generic model that can perform liver segmentation precisely on any CT acquisition of the liver.

KEYWORDS:

• 3DIRCADb
• Computed tomography
• Computer-aided diagnosis
• LITS
• Liver segmentation
• Morphological reconstruction operation

Acknowledgements

The authors thank the Management and Principal for providing the necessary infrastructure to carry out the research work. The authors would also like to thank the anonymous reviewers for their insightful comments that led to significant improvements in the manuscript. They also thank the Indian MRI Diagnostic and Research Ltd., Madurai, Tamil Nadu, India for providing the necessary CT scan images.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author.

Additional information

Notes on contributors

Emerson Nithiyaraj E

Emerson Nithiyaraj E received the BTech degree in electronics and communication engineering from Kalasalingam Academy of Research and Education, Krishnankoil in 2015 and the ME degree in VLSI design from Mepco Schlenk Engineering College, Anna University, Chennai in 2017. Currently, he is pursuing PhD at Anna University, Chennai as a full-time research scholar at the Department of Electronics and Communication Engineering at Mepco Schlenk Engineering College. His research interests include medical image analysis, machine learning and deep learning. Corresponding author. Email: [email protected]

Arivazhagan Selvaraj

Arivazhagan Selvaraj received the BE degree in electronics and communication engineering from Alagappa Chettiar College of Engineering and Technology, Karaikudi in 1986 and the ME degree in applied electronics from College of Engineering, Guindy, Anna University, Chennai in 1993. He acquired PhD degree from Manonmaniam Sundaranar University, Tirunelveli in 2005. He is currently a professor in the ECE Department and principal of Mepco Schlenk Engineering College, Sivakasi. He has 35 years of teaching and research experience. He has been awarded Young Scientist Fellowship by TNSCST, Chennai in 1999. He has published over 250 technical papers in international/national journals and conferences. He is currently the principal investigator of one Research and Development Project sponsored by DRDO, New Delhi. Also, he has completed 11 research and development projects, funded by ISRO, Trivandrum, DRDL, Hyderabad, ADE, Bengaluru, DRDO, New Delhi, NPOL, Kochi, DST, New Delhi and AICTE, New Delhi. His current research interests include machine learning, biometric system, digital image processing, steganography, steganalysis, and digital electronics. Email: [email protected]