Machine learning using AMKSOM for Identifying Tumor around Edema region

In most of the part of India, the statistical data reveals that people die because of inaccurate and last stage predictions of disease. An automatic system identifying tumor and differentiating edema from tumor can effectively help neurosurgeon for follow-up and pre-surgical evaluation planning. Therefore, we have developed efficient algorithm which predict early biomarkers and exemplify brain disease. We have integrated adaptive moving mapping unsupervised mapping(AMSOM) with FK-means and feature extraction for improved data reduction with efficient clustering of tissues. We have carried out algorithm using real time data from hospital and reshaped to 256×256 pixel size for processing. The image analysis reveals discriminative brain patterns associated to the brain tumor detection improving mean square error 14%, psnr with 12% and accuracy with 6%. Here proposed method implemented gives better detection and evaluation yielding promising results for the MRI 1.5 Tesla T1 images with 22 features. Therefore, it will be used by radiologists in the study of brain aging and disease.