Title page for ETD etd-02232005-181427


Type of Document Master's Thesis
Author Bomma, Pallavi
Author's Email Address pbomma1@lsu.edu
URN etd-02232005-181427
Title Computer-Aided Diagnosis Tool for the Detection of Cancerous Nodules in X-Ray Images
Degree Master of Science in Electrical Engineering (M.S.E.E.)
Department Electrical & Computer Engineering
Advisory Committee
Advisor Name Title
Jerry Trahan Committee Chair
Bahadir K. Gunturk Committee Member
Hsiao-Chun Wu Committee Member
Keywords
  • roc analysis
  • artificial neural networks
  • medical imaging
Date of Defense 2005-01-14
Availability unrestricted
Abstract
This thesis involves development of a computer-aided diagnosis (CAD) tool for the detection of cancerous nodules in X-ray images. Both cancerous and non-cancerous regions appear with little distinction on an X-ray image. For accurate detection of cancerous nodules, we need to differentiate the cancerous nodules from the non-cancerous. We developed an artificial neural network to differentiate them. Artificial neural networks (ANN) find a large application in the area of medical imaging. They work in a manner rather similar to the brain and have good decision making criteria when trained appropriately. We trained the neural network by the backpropagation algorithm and tested it with different images from a database of thoracic radiographs (chest X-rays) of dogs from the LSU Veterinary Medical Center.

If we give X-ray images directly as input to the ANN, it incurs substantial complexity and training time for the network to process the images. A pre-processing stage involving some image enhancement techniques helps to solve the problem to a certain extent. The CAD tool developed in this thesis works in two stages. We pre-process the digitized images (by contrast enhancement, thresholding, filtering, and blob analysis) obtained after scanning the X-rays and then separate the suspected nodule areas (SNA) from the image by a segmentation process. We then input enhanced SNAs to the backpropagation-trained ANN. When given these enhanced SNAs, the neural network recognition accuracy, compared to unprocessed images as inputs, improved from 70% to 83.33%.

Files
  Filename       Size       Approximate Download Time (Hours:Minutes:Seconds) 
 
 28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access 
  Bomma_thesis.pdf 2.11 Mb 00:09:46 00:05:01 00:04:23 00:02:11 00:00:11

Browse All Available ETDs by ( Author | Department )

If you have questions or technical problems, please Contact LSU-ETD Support.