Extraction of tumor region in color images using wavelets

In this paper, a wavelet-based method is proposed to extract the tumor regi on from a color medical image. In the past, the diagnosis of tumors usually relied upon the tissue section stained by methylene blue. However, traditi onal approaches have relied primarily on manual processes and erroneous ass essment may arise from subjective judgement and eye measurement. Instead of the previous method, we propose an improved radial search technique to inc rease the accuracy of the extraction of the tumor border. Experimental resu lts show that the proposed method is fast and can extract the tumor region successfully. Moreover, the results of this research will be helpful in the development of a computer-aided image analysis system that will automate v isual feature extraction for tumor evaluation. (C) 2000 Elsevier Science Lt d. All rights reserved.