Introduction: Bone cancer is a common type of cancer in canines. There are different types of bone cancer of which osteosarcoma is the most common corresponding to more than 90% of all tumors in large breed dogs. It usually appears in the limbs and grows fast and is highly metastatic. Imaging methods such as x-ray, computed tomography (CT), magnetic resonance imaging (MRI), and biopsy are used primarily for diagnosing this disease. These medical imaging techniques are most prevalent and are being extensively used in diagnostic imaging, but they are not suitable as a pre-screening method due to some disadvantages; including high expense, high dose of radiation, and keeping the patient (canine) motionless during the imaging procedures. Also this type cancer has high metastatic potential and may not be diagnosed using these techniques before it has spread to other body parts.
Objectives: This research study explores the possibility of using thermographic images as a pre-screening tool for diagnosis of bone cancer in dogs, the advantages include being non invasive, cheaper, faster, no radiation, real time imaging, portable and available for computer aided image analysis.
Results: Experiments were performed in several stages with different sample sizes and combinations of images. Research started by dividing all the images into two groups, cancer and no-cancer. Results were analyzed and further divisions of images were made based on body parts, hair type, and body area and then more experiments were conducted. These experiments were analyzed to identify how images should be divided to improve the results. All the experiments were performed with color normalization using temperature data provided by the Long Island Veterinary Specialists. After analyzing the initial experimental results it was determined that grouping by separate body parts provided better results than with other division of images. For this the images were first divided into four groups according to body parts (Elbow/Knee, Full Limb, Shoulder/Hip and Wrist). Each of the groups was then further divided into three sub-groups according to views (Anterior, Lateral and Posterior). Thermographic patterns of normal and abnormal dogs were analyzed using feature extraction and pattern classification tools. Texture features, spectral feature and histogram features were extracted from the thermograms and were used for pattern classification. Three different pattern classification tools, CVIP-FEPC, Partek Discovery Suite and Artificial Neural Network were used to classify the thermograms. The best classification success rate in canine bone cancer detection was 90% with sensitivity of 100% and specificity of 80% produced by anterior view of full-limb region with nearest neighbor classification method and normRGB-lum color normalization method with CVIP-FEPC as pattern classification method. Three experiments with each data normalization method that provide best result for each experimental setup were chosen and were used for Partek Discovery Suite. Only those features that were used on those experiments were considered. Nearest neighbor along with discrimininant analysis were used as pattern classification tools. The result was almost same as the one obtain from CVIP-FEPC in some case one or two sample being misclassified. This shows that the pattern classification algorithms that were used in CVIP-FEPC were sufficient for pattern classification for this research. The features extracted using CVIP-FEPC was then used to classify images using artificial neural network. The features values that were not normalized were considered. Principal component transform was used before these were fetched into artificial neural network. Only the principal components that contain 98% of the variance were considered and rests were discarded. The best classification success rate of 100% with 100% sensitivity and 100% specificity was obtained for anterior view of full limb region with artificial neural network as the pattern classification method.
Conclusion: Our results show that it is possible to use thermographic imaging as a pre-screening tool for detection of canine bone cancer. We have also determined that separation of images considering factors like body parts, camera view, body area making a homogeneous data set has high effect on success rate.
|Advisor:||Umbaugh, Scott E.|
|Commitee:||LeAnder, Robert, Noble, Bradley|
|School:||Southern Illinois University at Edwardsville|
|Department:||Electrical and Computer Engineering|
|School Location:||United States -- Illinois|
|Source:||MAI 54/03M(E), Masters Abstracts International|
|Subjects:||Information Technology, Electrical engineering|
|Keywords:||Canine bone cancer, Cvip-fepc, Cviptools, Feature extraction, Pattern classification, Thermographic images|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be