Feature Extraction with Intrinsic Distortion Correction in Celiac Disease Imagery: No Need for Rasterization

• Published in: Medical Computer Vision. Large Data in Medical Imaging pp. 196 - 204
• Main Authors: Gadermayr, Michael, Uhl, Andreas, Vécsei, Andreas
• Format: Book Chapter
• Language: English
• Published: Cham Springer International Publishing 2014
• Series: Lecture Notes in Computer Science
• Subjects: Celiac Disease; Distortion Correction (DC); Distortion Types; Rotation Invariant Local Binary Pattern; Undistorted Image
• ISBN: 9783319055299; 3319055291
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-05530-5_19

In the fields of computer aided celiac disease diagnosis, wide-angle endoscopy lenses are employed which introduce a significant degree of barrel type distortion. In recent studies on celiac disease classification, distortion correction techniques are investigated which use interpolation techniques, in order to maintain the rasterization of the image. Subsequent feature extraction is based on the new images. We introduce a generic feature extraction methodology with intrinsic distortion correction, which does not include this rasterization for features which do not need a regular grid. As distortion correction turned out to be disadvantageous in most cases, we aim in investigating the (negative) effect of the applied rasterization. In our experiments, the omission of rasterization actually turns out to be advantageous. This fact is an incentive for developing more features, which are not based on a regular grid.