Interactive Border Contour with Automatic Tracking Algorithm Selection for Medical Images
Contour tracking methods such as Live Wire and Riverbed have been widely used for several medical imaging applications such as tissue and tumor segmentation. Several variations of these methods have been proposed, but none of them is better than the others for all kind of tasks and image modalities. In this paper, we propose an interactive framework for medical image segmentation with four different contour tracking methods with an intuitive interface for visualizing and choosing the best option for each contour segment. The framework also includes a classifier which indicates the most appropriated method in both automatically and semi-automatically fashions. Our experiments employ a robot user which simulates the human behavior and was able to indicate the correct method for 67% of the segments.
KeywordsContour tracking Machine learning Image processing
The authors would like to thank FAPESP (2016/21591-5) for funding.
- 4.Falcão, A.X., da Cunha, B.S.: Multiscale shape representation by the image foresting transform. In: Proceedings of SPIE, vol. 4322, pp. 1091–1100 (2001)Google Scholar
- 7.Falcão, A., Cunha, B., Lotufo, R.: Design of connected operators using the image foresting transform. In: Proceedings of SPIE, vol. 4322, pp. 468–479 (2001)Google Scholar
- 16.Wenzel, F., Galy-Fajou, T., Deutsch, M., Kloft, M.: Bayesian nonlinear support vector machines for big data. In: Ceci, M., Hollmén, J., Todorovski, L., Vens, C., Džeroski, S. (eds.) ECML PKDD 2017. LNCS (LNAI), vol. 10534, pp. 307–322. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-71249-9_19CrossRefGoogle Scholar