Machine learning base Brain tumour detection and classification using image toolbox

Image Processing

This paper presents a novel framework for braintumor diagnosis and its grade classification based on higherorder statistical texture features namely kurtosis and skewnessalong with selected morphological features. These features wereextracted from segmented tumorous T2-weighted brain MRimages, in order to distinguish high grade (HG) tumor from lowgrade (LG) tumor. Tumor classification is carried out usingSupport vector machine (SVM) classifier with k-fold cross-validation. This work also compares the performance of the SVMwith linear discriminant analysis (LDA) and Naives Bayesclassifiers. Our proposed feature set achieved sensitivity,specificity and accuracy of 100% with SVM (linear kernel) whileclassifying brain tumor MR images as LG/HG. Magnetic resonance imaging (MRI) is a techniquewhich is used for the evaluation of the brain tumor in medicalscience. In this report, a methodology to study and classify theimage de-noising filters such as Median filter, Adaptive filter, Averaging filter, Un-sharp masking filter and Gaussian filter isused to remove the additive noises present in the MRI images i.e. Gaussian, Salt & pepper noise and speckle noise. A novel idea is proposed for successful identification of the brain tumor using SVM (linear kernal). Efficient classification of the MRIs is done using Naïve Bayes Classifier and Support Vector Machine (SVM) so as to provide accurate prediction and classification.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.