Abstract: Medical images are increasingly being used within healthcare for diagnosis, planning treatment, guiding treatment and monitoring disease progression.
Technically, medical imaging mainly processes uncertain, missing, ambiguous, complementary, inconsistent, redundant contradictory, distorted data and information has a strong structural character. As a general approach, the understanding of any image involves the matching of features extracted from the image with pre-stored models. The production of a high-level symbolic model requires the representation of knowledge about the objects to be modeled, their relationships, and how and when to use the information stored within the model.
This presentation reports new (semi)automated methods for the segmentation and classification of images of cerebral structures using soft computing techniques (e.g. fuzzy logic, genetic algorithms), information fusion and specific domain knowledge. Fuzzy logic acts as a unified framework for representing and processing both numerical and symbolic information (“hybridization”), as well as structural information constituted mainly by spatial relationships in biomedical imaging. Our applications are mainly for the segmentation of brain structures for magnetic resonance (MR) and CT (computer tomography) images, based both on atlas and real data. Promising results show the superiority of this knowledge-based approach over best traditional techniques in terms of segmentation errors. The classification of different cerebral structures is made by implementing rules yielded both by domain literature and by medical experts. In this paper we also propose a new method for medical images segmentation in the framework of neutrosophic logic and Dezert-Smarandache Theory, combined with fuzzy k-means algorithm.
Though the proposed methodology has been implemented and successfully used for model-driven in the domain of MR and CT imaging, the deployed methods are generic and applicable to any structure that can be defined by expert knowledge and morphological image analysis.
Another branch of applications of soft-computing is illustrated in image registration and data fusion by using different techniques, e.g. linear transformations and wavelet functions. An useful application was implemented in PET-CT imaging, both for a single PET-CT scanner and for separate PET and CT images. We first applied a mutual information based registration algorithm and then fused the PET and CT images using the 2í-Granular Support Vector Machine. As a general result, the fused image contains the properties of both PET and CT images and is an efficient tool for clinical diagnosis and therapy.

Brief Biography of the Speaker:
Prof. dr. eng. Hariton Costin graduated from the Technical University in Iasi, Romania, Faculty of Electronics and Telecommunications (1980). He obtained the Ph.D degree in Applied Informatics and a MBA diploma both from the same university in Iasi, and now is full professor and head of the Biomedical Instrumentation Department at the University of Medicine and Pharmacy, Faculty of Medical Bioengineering, Iasi, Romania, (www.umfiasi.ro ; www.bioinginerie.ro). Also, he is senior researcher at the Romanian Academy – Iasi Branch, Institute of Computer Science, within the Image Processing and Pattern Recognition Lab, (http://www.iit.tuiasi.ro/personal/h_costin.html), where he studies image processing and analysis by using Artificial Intelligence methods and data fusion.
Prof. Costin’s competence areas include: medical electronics, biosignal and image processing and analysis, artificial intelligence (soft-computing, expert systems), hybrid systems, HCI (human-computer interfaces), e-health and telemedicine.
His scientific activity can be resumed by about 100 published papers, 7 books, 4 book chapters, 3 patents, 2 national awards.
His research activity is illustrated by 28 research reports within Romanian Academy, a position of technical manager within EU/FP5/INES 2001-32316 project for a telemedicine application, he is director of the most complex Romanian pilot telemedical centre in Iasi, and director for 8 national granted projects in bioengineering and biomedical image processing / analysis. He was invited postdoc researcher at the University of Science and Technology of Lille (France, 2002, in medical imaging) and gave invited talks at international conferences.
Prof. Costin is a member of the IEEE - Engineering in Medicine & Biology Society (EMBS) and of other 8 scientific societies.