Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/15257
Title: Certain improvements in micro array gene analysis using novel neural network approach for prediction of cancer in colon
Researcher: Venkatesh E T
Guide(s): Thangaraj P
Keywords: Neural Network algorithm, artificial neural network, linear vector quantization, microarray, gene analysis, Kent Ridge Biomedical Data Repository
Upload Date: 20-Jan-2014
University: Anna University
Completed Date: 
Abstract: Over the last few years, microarray gene analysis has emerged as a tool for managing thousands of gene expression levels in parallel. This technology has been successfully used to determine disease states and responses to stimuli between cells. Experimental techniques such as oligonucleotide allow comparison of thousands of genes simultaneously. There are several techniques available for microarray data analysis and mining. The most elementary techniques used are based on individual gene analysis, such as fold approach, t-test rule and the Bayesian framework. The research investigates the existing framework for cancer prediction using microarray gene expression data with the help of Artificial Neural Network (ANN) classifier, to improve the prediction accuracy. For testing the accuracy of ANN, colon cancer datasets obtained from Kent Ridge Biomedical Data Repository are used. Classification accuracy results obtained from existing data mining algorithms were in the range of 53.23% to 85.48%. The enhanced LVQ, called Linear Vector Quantization was able to classify the colon cancer dataset with an accuracy of 90.32% compared to 87.1% of regular Learning Vector Quantization. The proposed Neural Network with Genetic Algorithm classification method was able to achieve its objectives, with the classification accuracy increasing by 5.68% over the proposed LVQ method. In the proposed Neural Network method of classification, accuracy was increased by 5.7% over the technique proposed by Furey et al, by 15.4% over that proposed by Ben Dor et al and by 8.9% over that proposed by Nguyen et al. Compared to Naïve Bayes, CART and SMO, the proposed algorithm displays very good sensitivity of 90% and above, leading to perfect prediction of cancer in colon. newline newline newline
Pagination: xix, 120
URI: http://hdl.handle.net/10603/15257
Appears in Departments:Faculty of Science and Humanities

Files in This Item:
File Description SizeFormat 
01_title.pdfAttached File36.88 kBAdobe PDFView/Open
02_certificates.pdf658.19 kBAdobe PDFView/Open
03_abstract.pdf24.52 kBAdobe PDFView/Open
04_acknowledgement.pdf16.43 kBAdobe PDFView/Open
05_contents.pdf48.53 kBAdobe PDFView/Open
06_chapter 1.pdf123.14 kBAdobe PDFView/Open
07_chapter 2.pdf91.18 kBAdobe PDFView/Open
08_chapter 3.pdf482.14 kBAdobe PDFView/Open
09_chapter 4.pdf796.83 kBAdobe PDFView/Open
10_chapter 5.pdf778.43 kBAdobe PDFView/Open
11_chapter 6.pdf791.23 kBAdobe PDFView/Open
12_chapter 7.pdf165.14 kBAdobe PDFView/Open
13_references.pdf55.23 kBAdobe PDFView/Open
14_publications.pdf16.31 kBAdobe PDFView/Open
15_vitae.pdf14.57 kBAdobe PDFView/Open


Items in Shodhganga are protected by copyright, with all rights reserved, unless otherwise indicated.