Utilization of Medical Image Soft Segmentation Based on Fuzzy Sets Classification Process Modified by Local Aggregation Approach

Medical image segmentation has been a challenging task for a long time. In the current age, we are overcrowded by medical image data acquired from various sources, such as CT, MR, ultrasound and many others. We usually need to perform segmentation, detection and extraction of objects of interest for further processing. This process includes quantification of parameters to determine a clinical evaluation. There are multiregional segmentation methods that allow for differentiation of individual morphological objects. However, the commonly used hard thresholding approaches lack of robustness in noisy environment leading to an incorrect pixel classification. Image segmentation based on fuzzy set theory brings much more effective alternative for image thresholding gained by local aggregation, making this method more noise resistive. We consciously performed a comparative analysis of articular cartilage and blood vessels segmentation. It was an obvious method utilization in which the native image features are badly recognizable and the objects features are well observed.