Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/13808
Title: Hardware reconfigurable neural network based electronic nose system for early diagnosis of sterile body fluid infections
Researcher: Subadra M
Guide(s): Rajamani V
Keywords: Neural network
Electronic nose system
Sterile body fluid
E-coli
Urinary Tract Infection
Upload Date: 9-Dec-2013
University: Anna University
Completed Date: 05/08/2011
Abstract: In the midst of infectious diseases, detection of microorganisms in Sterile Body Fluids (SBF) has vital diagnostic and therapeutic implications. Sepsis and Urinary Tract Infections are such kinds which continue to be a major problem for neonates. The results of diagnostic studies of these diseases shall be informed without delay to the concerned physician for initiating accurate treatment. This research proposes a rapid (lt3 hours) and hand held diagnostic tool by employing Electronic nose system. Out of various microorganisms causing neonatal sepsis and Urinary Tract Infection (UTI), this research focuses on differentiating E.coli from other microorganisms. This research investigation also tries to identify the growth phase of E.coli as it cannot be identified by the conventional gold standard method. In order to carry out this study, cultured microbial samples were collected and the head space of each sample was analysed by Metal Oxide Semiconductor (MOS) sensor array. Four different pattern recognition procedures have been deployed; they are classified as linear: Principal Component Analysis (PCA) and three non linear : Multi Layer Perceptron (MLP), Principal Component Analysis hybrid (PCANN) and Support Vector Machine (SVM). In comparison with the results of previous work, it is found that in this work, the overall sensitivity 99.44% and the overall Specificity 100% are achieved comparatively in a shorter duration (lt3 hours). This study successfully demonstrated the feasibility of E-nose as a diagnostic tool for early diagnosis of sterile body fluid infections together with the possibility of recognising the growth phase. After investigation it has been realised that potential does exist for early diagnosis which may help save the neonates by providing them an appropriate antibiotic at the earliest. As the MOS sensory array can be fabricated as a chip, it is now assured that PARC (ANN) can also be put into the chip, thereby a prototype can be formed for devising hand held diagnostic tool.
Pagination: xxviii, 195p.
URI: http://hdl.handle.net/10603/13808
Appears in Departments:Faculty of Information and Communication Engineering

Files in This Item:
File Description SizeFormat 
01_title.pdfAttached File49.73 kBAdobe PDFView/Open
02_certificates.pdf912.01 kBAdobe PDFView/Open
03_abstract.pdf17.94 kBAdobe PDFView/Open
04_acknowledgement.pdf16.11 kBAdobe PDFView/Open
05_contents.pdf74.77 kBAdobe PDFView/Open
06_chapter 1.pdf238.43 kBAdobe PDFView/Open
07_chapter 2.pdf492.38 kBAdobe PDFView/Open
08_chapter 3.pdf1.3 MBAdobe PDFView/Open
09_chapter 4.pdf5.03 MBAdobe PDFView/Open
10_chapter 5.pdf698.92 kBAdobe PDFView/Open
11_chapter 6.pdf2.08 MBAdobe PDFView/Open
12_chapter 7.pdf24.75 kBAdobe PDFView/Open
13_appendices 1 to 3.pdf145.09 kBAdobe PDFView/Open
14_references.pdf72.23 kBAdobe PDFView/Open
15_publications.pdf17.33 kBAdobe PDFView/Open
16_vitae.pdf11.88 kBAdobe PDFView/Open


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