Pyramid Histograms of Motion Context with Application to Angiogram Video Classification
Due to poor image quality as well as the difficulty of modeling the non-rigid heart motion, motion information has rarely been used in the past for angiogram analysis. In this paper we propose a new motion feature for the purpose of classifying angiogram videos according to their viewpoints. Specifically, local motion content of the video around the anatomical structures cardiac vessels is represented using the so-called “motion context”, a motion histogram representation in polar coordinates within a local patch. The global motion layout is captured as pyramid histograms of the motion context (PHMC) in a manner similar to that proposed by the Spatial Pyramid Kernel . The PHMC is a robust representation of the motion features in a video sequence. Through experiments on a large database of angiograms obtained from both diseased and control subjects, we show that our technique consistently outperforms state-of-the-art methods in the angiogram classification test.
KeywordsRight Coronary Artery Left Coronary Artery Shape Context Coronary Artery Tree Vessel Centerline
Unable to display preview. Download preview PDF.
- 1.Bosch, A., Zisserman, A., Munoz, X.: Representing shape with a spatial pyramid kernel. In: Proceedings of the 6th ACM International Conference on Image and Video Retrieval, CIVR 2007, pp. 401–408. ACM, New York (2007)Google Scholar
- 2.Sonka, M., et al.: Robust simultaneous detection of coronary borders in complex images. IEEE Trans. Medical Imaging, 151–161 (1995)Google Scholar
- 3.Schaap, M., et al.: Standardized evaluation methodology and reference database for evaluating coronary artery centerline extraction algorithms. Medical image Analysis, 701–714 (2009)Google Scholar
- 5.Sato, Y., Nakajima, S., Shiraga, N., Atsumi, H., Yoshida, S., Koller, T., Gerig, G., Kikinis, R.: Three-dimensional multi-scale line filter for segmentation and visualization of curvilinear structures in medical images. IEEE Medical Image Analysis, 143–168 (1998)Google Scholar
- 6.Perfetti, R., Ricci, E., Casali, D., Costantini, G.: A cnn based algorithm for retinal vessel segmentation. In: ICC 2008: Proceedings of the 12th WSEAS International Conference on Circuits, pp. 152–157. World Scientific and Engineering Academy and Society (WSEAS), Stevens Point (2008)Google Scholar
- 7.Haris, K., Efstratiadis, S., Maglaveras, N., Pappas, C., Gourassas, J., Louridas, G.: Model-based morphological segmentation and labeling of coronary angiograms. IEEE-TMI (10), 1003–1015 (1999)Google Scholar
- 8.Medis medical imaging systems, Inc., http://www.medis.nl/index.htm
- 9.Syeda-Mahmood, T., Beymer, D., Wang, F., Mahmood, A., Lundstrom, R., Shafee, N., Holve, T.: Automatic selection of keyframes from angiogram videos. In: International Conference on Pattern Recognition (ICPR 2010), Turkey (2010)Google Scholar
- 12.Grauman, K., Darrell, T.: The pyramid matching kernel: Discriminative classification with sets of image features. In: ICCV (2005)Google Scholar