Title: Mammogram tumour classification using Q learning

Authors: S. Thamarai Selvi; R. Malmathanraj

Addresses: Department of Computer Technology, Madras Institute of Technology, Anna University, India. ' Department of ECE, National Institute of Technology, Tiruchirappalli, India

Abstract: In mammography, there have been quite a number of papers on image enhancement and denoising however segmenting the image is still in its developing stage. The mammograms, as normally viewed, display the texture property. In the Magnetic Resonance Imaging the mammograms with glandular tissues and malignant region have periodic repetition in the image and make the detection of small malignancies difficult. In this paper, it is intended to contribute to the medical community by implementing a novel approach in segmentation using Q learning algorithm for multilevel thresholding technique. Furthermore, various feature datasets of cancerous and non cancerous mammographs are calculated and used for classification as either benign or malignant. The performance evaluation is made using Region of Convergence (ROC) graph and Overall Performance (OP) rate.

Keywords: Q learning; reinforcement learning; multilevel thresholding; SVMs; support vector machines; ROC; region of convergence; cancerous mammograms; tumour classification; image segmentation; medical imaging; breast cancer; benign tumours; malignant tumours.

DOI: 10.1504/IJBET.2011.044413

International Journal of Biomedical Engineering and Technology, 2011 Vol.7 No.4, pp.339 - 352

Published online: 21 Jan 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article