3D SEGMENTATION AND FEATURE EXTRACTION OF CLSM SCANNED NUCLEI USING EVOLUTIONARY SNAKES

This paper presents a novel approach for quantifying chromatin in cell nuclei by using 3D texture analysis on PI dyed prostate nuclei. This was rendered possible by the use of the proposed 3D segmentation method for extraction of the volumetric data. The method is a generalization of evolutionary snake approach to the case of 3D objects. The segmentation is followed by 3D texture feature extraction performed on a group of normal and malignant prostate nuclei. Student´s t-test results show a statistically significant differentiation between malignant and normal cells in many of the 3D texture features used showing the effectiveness of the proposed approach for quantifying nuclei state of malignancy