Megakaryocyte Images Segmentation Using Wavelet Transform and Modified Fuzzy C-means Techniques

The image segmentation is crucial and effective step in the description of megakaryocyte cell because it allows to identify the main details (cytoplasm and nucleus) and processed it separately from unnecessary details (background). This paper is proposed a state of art technique for the automatic segmentation for megakaryocyte structure image. The proposed method consists of several steps and it combines different tools: wavelet transform technique and modified fuzzy c-means technique. The first step is data collection, data will be collected as photograph of microscope sample which illustrates megakaryocytes of subject’s sample. In this work, 25 samples (aspirates stained by gemsa stain) are collected from 10 recruited patients. Wavelet transform technique was used to differentiate between image content and noise as pre-processing step, also to prepare image to the next step. Implementation proposed modified fuzzy c-means technique is the third step. The modification of fuzzy c-means technique deals uncertainty of the image details based on measure of fuzziness and Euclidean distance via fuzzy inference technique. Based on quantitative evaluations tools (sensitivity, specificity and accuracy) the proposed method exhibits remarkably high level of accuracy compared with Otsu’s segmentation method in the case of detection nucleus and cytoplasm (8% and 24% respectively).