1 SEMI AUTOMATED SEGMENTATION OF CHROMOSOMES IN METAPHASE CELLS Mousami V. Munot*, Dr. Madhuri A. Joshi $ , Priyanka Mandhawkar † * Assistant Prof. Dept of E&TC, Pune Institute of ComputerTechnology, India, mousami_vm@yahoo.co.in $ Dean R&D, Prof. Dept of Electronics,College of Engineering, Pune, India, punemajoshi@gmail.com; † Student, Dept of E&TC, Pune Iunstitute of Computer Technology, India, mandavkarpriyanka@gmail.com ; Keywords: Metaphase, Karyotyping, Random Walker Segmentation. Abstract Image segmentation plays a crucial role in many medical imaging applications by automating or facilitating the delineation of anatomical structures and other regions of interest. Since the birth of the automated karyotyping systems by the aid of computers, building a fully automated chromosome analysis system has been an ultimate goal. Along with many other challenges, accurate segmentation of the chromosomes has been a major challenge especially due to the non rigid nature of the chromosomes. The earlier reported approaches for the segmentation have limited success as they are sensitive to scale variation, experimented only on gray images, unable to segment the clusters and the highly bent chromosomes. This work, describes an effective approach of segmentation of chromosomes in Metaphase images using Random Walker Algorithm [RWA] which is yet unexplored and not reported in the literature. The efforts are also done to compare the results with traditional methods so as to prove the efficiency of the implemented RWA algorithm. The algorithm is tested on publically available database and has shown encouraging and acceptable results. 1. Introduction Human chromosome analysis is an essential task in cytogenetics, especially in prenatal screening and genetic syndrome diagnosis, cancer pathology research and environmentally induced mutagen dosimetry [1, 4]. Metaphase is the stage of cell division at which the chromosomes are most suitable for analysis. [5,9]. A normal human diploid cell contains 22 pairs of chromosomes, autosomes of classes 1-22 and 2 sex chromosomes, either XX or XY. The karyotype displays chromosomes in standard positions based on their length, centromeric index and band pattern. The process of manual karyotyping is usually carried out by expert clinicians who initially identifies each chromosome in the picture using hierarchical chromosome identification and then finally using expert knowledge view the pictures, identify the chromosomes, cut and place them in their specified locations in the karyotype and classifies them into smaller groups [6]. The visual inspection is thus tedious, time consuming, laborious and an expensive procedure. Hence, many attempts have been made to automate the process of karyotyping [5, 10]. Automated Karyotyping systems allows countless clinical advantages such as interactive and graphical environment, faster in the accomplishment of the samples, allowing quality printing, being self explanatory, better interpretation of the image, and it still makes possible the storage of the information in a database for future analysis [15]. The basic steps in any automated digital image processing or pattern recognition systems would be composed of Pre- processing, Segmentation, feature extraction and Classification. The results of the automated systems though encouraging, have limited success, still needing human interaction. This is mainly due to non rigid nature of chromosomes leading to unpredictable shape variability thus requiring special efforts to segment and classify the chromosomes to their respective classes. 2. Related Work Segmentation is the process of dividing the image into segments, each of which has meaning to the human observer. It is a technique to separate or differentiate the objects which are of interest from non objects or background [8, 12]. In chromosome analysis, it is desired to segment the image into background and chromosome pixels and to divide further the chromosome pixels into individual chromosome type pixels. Segmenting the chromosome image into background and chromosome is a fairly straightforward task usually accomplished by thresholding. However, dividing the chromosome pixels into individual chromosomes types is quite difficult since the chromosomes often touch or overlap [2, 11]. A wide assortment of segmentation methods and approaches have been experimented and explored for the separation of chromosome from the metaphase images. In general chromosome intensities are brighter than the neighbouring background, although the background surface is not globally uniform and so adaptive thresholding is reported to be an effective segmentation method. But the method fails, creating holes inside the chromosomes when a number of pixels in the foreground are darker than the neighbouring foreground pixels [08]. Fuzzy set theory has been applied to chromosome segmentation in which fuzzy binary relations are