Wydział Elektrotechniki, Automatyki, Informatyki i Elektroniki AGH Studia doktoranckie: Biocybernetyka i Inżynieria Biomedyczna
Przedmiot:	Medical image analysis Analiza obrazów medycznych
Kiedy:	sem. 5
Wymiar:	30 h
Prowadzący:	prof. dr hab. inż. Ryszard Tadeusiewicz
Miejsce pracy	WEAIiE
Forma zajęć:	seminarium
Treści kształcenia:	Problems presented and discussed during the lectures: 1. Sources of medical images and its general characteristics X-ray images CT images NMR images USG images 2. Compression of medical images and its storing in medical databases 3. General properties of medical images 4. Preprocessing of medical images and its general enhancement Simple low pass and high pass filtering of the images Nonlinear filtering and mathematical morphology algorithms (e.g. median and top heat transformation) for medical images cleaning and improvement Spectral transformations as a tools for increasing visibility of morphological details on medical images Image enhancement by means of contrast regulation Binaryzation of the images Subtractive angiography and other methods of image enrichment by using of multi-images transformations Hough transform in medical applications 5. Medical images and web technologies 6. Hospital Information Systems for medical images storing and distribution 7. Algorithms and methods of goal oriented processing of medical images Segmentation of medical images Thinning and skeletonization of the objects on the images Example of special kind processing of the medical images: straightening transformation 8. Examples of parameters extracted from some medical images and methods of its analysis in selected medical problems Shape description on medical images Texture description on medical images
	9. Pattern recognition, clustering and classification applied to selected medical images General model of medical images recognition Acquisition of the data for recognition Features space formation Decision functions as measures of patterns similarity Decision making and answer generation Transformations helping in decision functions formation Clustering as the symmetrical problem for classification and pattern recognition Methods based on decomposition Methods base on aggregation Complex iterative methods · Simple methods of pattern recognition (e.g. k-neighbors algorithm and similar metrics-based methods) in medical applications Methods of pattern recognition based on approximation of regions in features space Linear approximation of bounders between regions SVM as simple and effective method of pattern recognition Nonlinear methods and nonlinear transformations in region modeling Neural networks for medical images recognition Bayesian and other probabilistic methods of pattern recognition in medical applications Syntactic and other structural methods in medical image recognition 10. Automatic understanding of the medical images as new paradigm used for advanced computer aiding of medical investigations General overview of image understanding concept Two directional way of data flow during solving the image understanding problem Language description of the image as a key to understanding procedures Types of languages used in syntactic approach to the image description problems 11. Image understanding methods in applications to medical diagnosis Graph Image Language Techniques Supporting Radiological, Hand Image Cognitive Interpretations The characteristic of the analyzed data Syntactic description of wrist radiograms Graph language describing the wrist structure Selected results Conclusion 12. Picture grammars in classification - example of 3D coronary vessels visualizations The classification problem Functional characteristics of coronary arteries Graph-based formalisms in the semantic description of coronary vessels Characteristics of the image data Graph-based description of coronary arteries Semantic analysis of coronary structure Results Conclusion