Regionální segmentace biomedicínských obrazových dat - laboratorní úloha

Abstract

Improving health applications for the detection of pathological tissues is also a logical step in today's world of computing technology and the application of artificial intelligence to everyday life. Accurate detection of pathological features plays an essential role in the planning of surgical procedures and the associated post-operative condition. Therefore, on segmentation applications the emphasis is laided precision. This thesis deals with this issue, specifically regional segmentation. The essence of this work is a comparative analysis of regional segmentation methods based on thresholding. It is a comparison of conventional Otsu method and unconventional soft method based on fuzzy set decomposition in combination with local statistical aggregation. The output of this work are segmentation analyzes depending on the dynamic effect of image noise intensity. For analysis purposes, images from CT and MRI were selected. The second output of the thesis is software "Regional Segmentation 1.0.0", which is used to perform segmentation with subsequent evaluation. The application has a well-arranged graphical - user interface that was created in MATLAB R2015a by MathWorks. The third output is a laboratory task on the topic of regional segmentation, which will be used for educational purposes.