• Amna P Asgar
• Anita Joseph
• Anjana Annmariya Chandy
• Anna Xavier
• Shyama Sreekumar

The conventional method of brain tumor detection and classification is by human inspection of magnetic resonant images of brain. But it is difficult when large amount of data is to be diagnosed and to be reproducible and also large data needs to be stored in the memory of analyst for immediate remedy suggestion or for continuous monitoring of the tumor at the required last stages. Robust computerized segmentation algorithms can help physicians to detect and classify brain tumor in lesser time with greater accuracy. In this paper, the images are initially pre-processed using median filter. We developed an algorithm for stripping the skull before segmentation process. K means segmentation is done for effective segmentation followed by feature extraction by Gray Level Co-occurrence matrix. Kurtosis, Skewness, Contrast, Entropy, Variance, Standard deviation, Homogeneity, Root Mean Square, Inverse difference moment and Correlation are the various features extracted from Gray Level Co-occurrence matrix. The probabilistic neural network classifier is trained for classification of images into normal, benign or malignant. Finally the area of tumor growth is calculated from which the grade of tumor is identified. The analysis of different features obtained from Gray Level Co-occurrence matrix is carried out to find out the features which can classify the images with more accuracy and helps in removing unnecessary features. The performance of the proposed technique is evaluated using confusion matrix from which 100 % Sensitivity, 88.88% Accuracy, 75% Specificity, 83.33% Positive Predictive Value and 100% Negative Predictive Value is obtained.

Classification, K-means Segmentation, Gray Level Co-occurrence matrix, Feature Analysis, Probabilistic Neural Network

"FEATURE ANALYSIS FOR BRAIN TUMOR DETECTION USING PROBABILISTIC NEURAL NETWORK CLASSIFIER", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 5, page no.326-330, May - 2019, Available :http://www.jetir.org/papers/JETIRCN06061.pdf

"FEATURE ANALYSIS FOR BRAIN TUMOR DETECTION USING PROBABILISTIC NEURAL NETWORK CLASSIFIER", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 5, page no. pp326-330, May - 2019, Available at : http://www.jetir.org/papers/JETIRCN06061.pdf