Author(s): Dina Aboul Dahab | Samy S. A. Ghoniemy | Gamal M. Selim
Journal: International Journal of Image Processing and Visual Communication
ISSN 2319-1724
Volume: 1;
Issue: 2;
Start page: 1;
Date: 2012;
VIEW PDF
DOWNLOAD PDF 
Original page
Keywords: Probabilistic Neural Network | Edge detection | image segmentation | brain tumor detection and identification
ABSTRACT
In this paper, modified image segmentationtechniques were applied on MRI scan images in order to detectbrain tumors. Also in this paper, a modified Probabilistic NeuralNetwork (PNN) model that is based on learning vectorquantization (LVQ) with image and data analysis andmanipulation techniques is proposed to carry out an automatedbrain tumor classification using MRI-scans. The assessment ofthe modified PNN classifier performance is measured in terms ofthe training performance, classification accuracies andcomputational time. The simulation results showed that themodified PNN gives rapid and accurate classification comparedwith the image processing and published conventional PNNtechniques. Simulation results also showed that the proposedsystem out performs the corresponding PNN system presented in[30], and successfully handle the process of brain tumorclassification in MRI image with 100% accuracy when thespread value is equal to 1. These results also claim that theproposed LVQ-based PNN system decreases the processing timeto approximately 79% compared with the conventional PNNwhich makes it very promising in the field of in-vivo brain tumordetection and identification.
Journal: International Journal of Image Processing and Visual Communication
ISSN 2319-1724
Volume: 1;
Issue: 2;
Start page: 1;
Date: 2012;
VIEW PDF
DOWNLOAD PDF Original pageKeywords: Probabilistic Neural Network | Edge detection | image segmentation | brain tumor detection and identification
ABSTRACT
In this paper, modified image segmentationtechniques were applied on MRI scan images in order to detectbrain tumors. Also in this paper, a modified Probabilistic NeuralNetwork (PNN) model that is based on learning vectorquantization (LVQ) with image and data analysis andmanipulation techniques is proposed to carry out an automatedbrain tumor classification using MRI-scans. The assessment ofthe modified PNN classifier performance is measured in terms ofthe training performance, classification accuracies andcomputational time. The simulation results showed that themodified PNN gives rapid and accurate classification comparedwith the image processing and published conventional PNNtechniques. Simulation results also showed that the proposedsystem out performs the corresponding PNN system presented in[30], and successfully handle the process of brain tumorclassification in MRI image with 100% accuracy when thespread value is equal to 1. These results also claim that theproposed LVQ-based PNN system decreases the processing timeto approximately 79% compared with the conventional PNNwhich makes it very promising in the field of in-vivo brain tumordetection and identification.
