Fault-Tolerant Medical Image Interpretation
Novel medical imaging techniques have been introduced over the last decade. They provide for the non-invasive study of internal structures and their dynamic behavior because of the large bandwidth characteristics of the new sensors. The processing and automatic interpretation of such images, however, lags far behind. We suggest herein a synergetic approach where novel techniques derived from artificial intelligence (AI), computer vision (CV) and neural networks (NN) could be integrated towards robust and automatic image interpretation. Such image interpretation would be relevant for the analysis of internal organs and/or tissue and to the understanding of time-varying (dynamic) imagery. Within the medical area it is important that such analysis be fault-tolerant (low sensitivity) to noisy data, occlusion/overlap, geometric distortions, and still be efficient.
KeywordsAssociative Memory Markov Random Fields Scale Space Range Image Response Vector
Unable to display preview. Download preview PDF.
- Rosenfeld, A. (Ed.) (1984), Multiresolution Image Processing and Analysis, Springer-Verlag.Google Scholar
- Pizer, S. (1989), Multiscale methods and the segmentation of medical images, this volume.Google Scholar
- Uhr, L. (1988), Parallel Computer Vision, Academic Press.Google Scholar
- Sood, A. and Shah, M. (1987), Scale space technique to finding primitives in images with application to road following, Proceedings SPIE—Applications of Artificial Intelligence V.Google Scholar
- Witkin, A.P. (1983), Scale space filtering, Proceedings of IJCAI.Google Scholar
- Yuille, A.L. and Poggio, T. (1983), Fingerprints theorems for zero crossings, MIT AI memo 730.Google Scholar
- Al-Hujaze, E. and Sood, A. (1988), Range data description based on multiple characteristics, Proceedings 1988 Goddard Conference on Space Applications of Artificial Intelligence.Google Scholar