Deep Learning Architectures for vessel Segmentation in 2D and 3D Biomedical Images

Mario Zusag
Deep Learning Architectures for vessel Segmentation in 2D and 3D Biomedical Images
[Bachelor Thesis] [image]

Information

Abstract

The aim of this thesis is to describe deep learning approaches for vessel segmentation in 2 and 3-dimensional biomedical images and the results achieved from these approaches on specific sets of data. The first chapter introduces the objective of this thesis, describes the data, which was used for the training, gives a short overview of machine learning and covers some theoretical aspects of artificial neural networks and especially of convolutional neural networks. The second chapter describes methods that were used for achieving the segmentation in 2 and 3 dimensions, like preprocessing of the images, algorithmic approaches, and general project set-up. The third and final chapter focuses on the results of methods described in chapter 2, contains personal advice for future approaches for improving the algorithm’s results and discusses the results. The thesis provides the theory, code snippets for the most fundamental part of the algorithms’ implementations and shows graphical, as well as numerical results of the approaches.

Additional Files and Images

Additional images and videos

Additional files

Weblinks

No further information available.

BibTeX

@bachelorsthesis{Zusag-2017-Bach,
  title =      "Deep Learning Architectures for vessel Segmentation in 2D
               and 3D Biomedical Images",
  author =     "Mario Zusag",
  year =       "2017",
  abstract =   "The aim of this thesis is to describe deep learning
               approaches for vessel segmentation in 2 and 3-dimensional
               biomedical images and the results achieved from these
               approaches on specific sets of data. The first chapter
               introduces the objective of this thesis, describes the data,
               which was used for the training, gives a short overview of
               machine learning and covers some theoretical aspects of
               artificial neural networks and especially of convolutional
               neural networks. The second chapter describes methods that
               were used for achieving the segmentation in 2 and 3
               dimensions, like preprocessing of the images, algorithmic
               approaches, and general project set-up. The third and final
               chapter focuses on the results of methods described in
               chapter 2, contains personal advice for future approaches
               for improving the algorithm’s results and discusses the
               results. The thesis provides the theory, code snippets for
               the most fundamental part of the algorithms’
               implementations and shows graphical, as well as numerical
               results of the approaches.",
  month =      aug,
  address =    "Favoritenstrasse 9-11/186, A-1040 Vienna, Austria",
  school =     "Institute of Computer Graphics and Algorithms, Vienna
               University of Technology",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2017/Zusag-2017-Bach/",
}