Segmenting and Classifying MRI Multimodal Images Using Cuckoo Search Optimization and KNN Classifier

Abstract

Currently, the death rate of cancer diseases like brain cancer, lung cancer, skin cancer and breast cancer is increasing. To analyze the internal functionality of tumors, there are many modalities like Computer Tomography, Magnetic Resonance Imaging, Ultra Sonic, Mammography and Poisson Emission Tomography. They help radiologists ensure the presence of cancer and reduce the severity by early diagnosis. Our proposed approach extracts the tumor region of multimodal images of breast and brain using Contrast Limited Adaptive Histogram Equalization with Cuckoo Search Optimization. The extracted features are fed to K nearest neighbor classifier to codify the tumor as benign or malignant. Our approach produces 98.4% and 97.6% accuracy for segmentation and classification, respectively. The computational time for segmentation is also comparable with the existing approaches like SOMFCM and PSOFCM and classification with Form feed neural network and SVM. We utilize website like BRATS, Harvard brain dataset and RIDER for validation.

KEYWORDS:

• MRI imaging
• CLANE preprocessing
• Cuckoo search optimization
• K-nearest neighbor classifier
• Performance measures

Additional information

Notes on contributors

R. Sumathi

R Sumathi completed ME (Computer Science and Engineering) in Mepco Schlenk Engineering College, Sivakasi. Currently, she is working as assistant professor in the Department of Computer Science and Engineering, School of Computing, Kalasalingam Academy of Research and Education. Her areas of interest are machine learning, data mining and medical image analysis.

M. Venkatesulu

M Venkatesulu obtained his PhD in mathematics from the Indian Institute of Technology, Kanpur in the year 1979. He has more than 40 years of teaching and research experience. He worked as faculty in various institutions in Andhra Pradesh and Tamilnadu, India. He was a visiting Professor at the University of Missouri, Kansas City, USA during 2006–2007. He has published more than 70 publications in peer-reviewed journals and reviewer for two Elsevier Journals. He has executed four R&D projects funded by DST, Government of India. He has guided number PhDs in different areas of computer science and information technology. Email: [email protected]

Sridhar P. Arjunan

P A Sridhar received PhD degree in electronics and biomedical engineering from RMIT University, Australia. He worked as a research fellow with Biosignals Lab at RMIT University, Australia in various research projects. Currently, he has two externally funded projects from SERB and SPARC, MHRD. He is a recipient of Australia-India Research Fellowship, RMIT Research Scholarship, German State research exchange scholarship and CASS Australian Early Career Researcher grant. He has co-authored three books and published more than 100 research papers in journals and international conferences. Email: [email protected]