Please use this identifier to cite or link to this item:
|Title:||ARTIFICIAL .NEURAL NETWORK AND TEXTURE BASED ANALYSIS OF BRAIN IMAGES|
|Keywords:||ELECTRONICS AND COMPUTER ENGINEERING;ARTIFICIAL NEURAL NETWORK;TEXTURE BASED ANALYSIS;BRAIN IMAGES|
|Abstract:||Medical imaging covers a- major application- area providing significant assistance in medical diagnosis. The development of these systems leads to valuable diagnostic tools that may largely assist physicians in the identification of tumors or malignant formations. Systems capable to discriminate : among various Tumor categories aim to, improve expert's ability to identify.. abnormalities (e.g.. cancerous regions) in . tissue , while decreasing the need for aggressive intervention and enhancing the . capability to, make accurate diagnosis. Different modalities in biomedical imaging, like CT scanners, provide detailed cross-sectional views of the human anatomy.. Advances in Artificial Neural Network (ANN) may contribute, to the design and development of new computational- tools to analyze.. multidimensional medical images. ANN and Texture based analysis on Brain images deals with analyzing Brain Images (specifically dealing with Tumor stage recognition and Tumor segmentation) using Neural network techniques and textural• content of the image. Main contribution in the thesis is done in, Edge Detection technique for Textured and non-Textured 'image, and use of, Kohonen, Neural Network for detection of stages of Tumor. The algorithm for Edge detection of Non-textured Image has proved to be good for some images, while- Edge Detection of Textured image, proved to be .better than-commercial packages available. Kohonen Neural network has been used for long in pattern recognition. The techniques that were applied on pattern has been tested for Brain Images after some processing and-is found that Kohonen Neural network can. successfully detect Tumor Stage'once it gets trained. Further the tumor that was not used in training can be associated with some closely related stage. The net' 'ork successfully recognizes noisy image also. Java Applet is chosen for the implementations. of all algorithms because, _Java is a Platform independent language, besides Applets . are most suitable for Internet. The results are tested and verified in Windows 2000 platform though it will support Unix/Linux also.|
|Research Supervisor/ Guide:||Anand, R. S.|
|Appears in Collections:||MASTERS' THESES (E & C)|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.