A Comparison of Four Texture Algorithms: Detecting Early Signs of Osteoarthritis in Knee Trabecular Bone

Information

• Visibility: hidden
• Publication Type: Ongoing Master Thesis
• Workgroup(s)/Project(s):
  • Visualization Group
• Date: ongoing
• Date (Start): 15. July 2017
• Matrikelnummer: 1227706
• First Supervisor:
  • Eduard Gröller
  • Renata Raidou
• Keywords: Knee Osteoarthritis Detection

Abstract

This master thesis aims to provide an in-depth comparison of four texture algorithms in their capacity of recognizing early signs of osteoarthritis in 2D radiographs of the knee trabecular bone. Given the fractal properties of the trabecular bone, two fractal-based algorithms (VOT and Bone Structure Value) that try to characterize the complexity of the underlying structure of the bone are presented. The third algorithm (based on Shannon’s Entropy) stems from the information theory and aims to describe the bone structure in terms of information complexity. The last algorithm is based on the co-occurrence matrix of an image and describes the texture in terms of Haralick features. First, a motivation for the detection of early OA is given. Second, a detailed description and mathematical background of the algorithms are presented. Third, the data sets (MOST, Portugal) used for classification and validation tests are introduced. Fourth, the results of each algorithm are discussed and their independent capacity of not discriminating between trabecular bones with early signs of OA and healthy bones. Fifth, a discussion is given on the influence of resolution, pixel spacing and x-ray doses on the results of the algorithms. Finally, some ideas about an optimal feature selection are introduced.

Additional Files and Images

Weblinks

No further information available.

BibTeX

@runmasterthesis{Oancea2018OAK,
  title =      "A Comparison of Four Texture Algorithms: Detecting Early
               Signs of Osteoarthritis in Knee  Trabecular Bone",
  author =     "Stefan Ovidiu Oancea",
  year =       "2018",
  abstract =   "This master thesis aims to provide an in-depth comparison of
               four texture algorithms in their capacity of recognizing
               early signs of osteoarthritis in 2D radiographs of the knee
               trabecular bone. Given the fractal properties of the
               trabecular bone, two fractal-based algorithms (VOT and Bone
               Structure Value) that try to characterize the complexity of
               the underlying structure of the bone are presented. The
               third algorithm (based on Shannon’s Entropy) stems from
               the information theory and aims to describe the bone
               structure in terms of information complexity. The last
               algorithm is based on the co-occurrence matrix of an image
               and describes the texture in terms of Haralick features.
               First, a motivation for the detection of early OA is given.
               Second, a detailed description and mathematical background
               of the algorithms are presented. Third, the data sets (MOST,
               Portugal) used for classification and validation tests are
               introduced. Fourth, the results of each algorithm are
               discussed and their independent capacity of not
               discriminating between trabecular bones with early signs of
               OA and healthy bones. Fifth, a discussion is given on the
               influence of resolution, pixel spacing and x-ray doses on
               the results of the algorithms. Finally, some ideas about an
               optimal feature selection are introduced.",
  keywords =   "Knee Osteoarthritis Detection",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2018/Oancea2018OAK/",
}