Decision support systems based on pattern recognition for evaluating of histopathologic images

ABSTRACTMASTER THESISDECISION SUPPORT SYSTEMS BASED ON PATTERN RECOGNITION FOREVALUATING OF HISTOPATHOLOGIC IMAGESSuat TORAMANFirat UniversityGraduate School of Natural and Applied ScienceDepartment of Computer and Electronics Education2006, Page: 58Computer technology is developed very rapidly. This development is related to the areaof medicine, industry and communication, too. Especially, medicine area profits so much fromthese kinds of developments. From many previous work, it is seen that the computer baseddecision support systems are used by the physicians because of its fast computation ability andits ability of researching enormous data comparatively.In this thesis, decision support systems that accelerate the medical decision process arerepresented to help the physicians for more accurate diagnosis. These decision support systemshave the ability of segmenting the histopathology images, computing the area of the region ofinterest and counting the number of cells in a given region. For fulfilling these operations, fuzzyc-means clustering, Kohonen self organizing maps, Shannon entropy and fractal analysismethods are used. All these methods are used for segmenting the region of interest in a givenhistopathologic image. Then various image filtering and image improving methods are appliedfor cell number counting and cell classification. Physician controlled experiments onhistopathologic images have shown that the proposed decision support systems performsuccessful results.Key words: Pattern recognition, Fuzzy c-means, Shannon entropy, Kohonen networks, Fractalanalysis, Decision trees, Decision support systems, Histopathology.