Embryonic Image Enhancement Based on Genetic Algorithm and Generic Filter Milad Momeni, Zahra Hossieni Nezhad, Mohsen Ebrahimi Moghaddam Dept. of Computer Science and Engineering Shahid Beheshti University; G.C Tehran, Iran e-mail: mi.momeni@mail.sbu.ac.ir, z.hosseininezhad@mail.sbu.ac.ir, m_moghadam@sbu.ac.ir Abstract—Monitoring the development of embryos in In vitro fertilization (IVF) clinical procedures is essential to prevent missing crucial events, so embryos should be kept a couple of days in a Time-Lapse. Time-Lapse is a heating incubator with an embedded camera. The captured images by these devices are usually influenced by many several different kinds of factors. This paper employs a robust, fault tolerant and reliable method to enhance the quality of embryonic images using composite filter which is a combination of several types of filters. Likewise, a comparison is presented between our method and some other enhancement techniques. Finally, the experimental results showed this method could efficiently enhance the quality of captured embryos’ images. Keywords-enhancement; embryo; genetic algorithm; generic filter; filter bank I. INTRODUCTION In vitro fertilization (IVF) is a new type of medical assisted reproduction technology (ART) that enables many infertile couples to have successful pregnancies and conceive a child [1]. In order to achieve this goal, the embryologist should have a good observation on the factors impacting results to choose the most promising embryos. Most of these factors could be estimated by embryos’ images like cell mitosis detection, pronuclear (PN) scoring, PN envelope fading, cell lineage analysis [2-4]. In this study, we use microscopy images of developing embryos taken at regular time intervals from Time-Lapse. These microscopy images usually suffer from low quality images and affected by environmental conditions such as insufficient light levels, noise and other things during image acquisition. Image enhancement is considered as one of the most essential and important techniques that improves the quality of the image to prepare better input for processing the image [5]. So, first and one of the most important phases of medical images analysis is image enhancement. The goal of this process is to clarify images for supervisor viewing, adjusting contrast, blur and noise removal and also revealing details. In the context of the medical image enhancement, many works have been proposed for such an aim. Enhancement based on 2D empirical mode decomposition is a work on frequency domain filters to have a high contrast image [6] also an effective edge detector morphological filter [7] is proposed to sharpen and deblur digital medical images. fractional differentials were used [8] to adjust the fractional order considering the dynamic gradient feature of the image to extract the edges more accurate and enhance them better. MH-FIL [9] is one of the several number of techniques based on histogram to enhance the contrast of the images using homomorphic filters. To enhance the low resolution, a versatile edge preserver uses different types of guided filter [10]. Each of these methods do their best in specific type of domain but they would have many shortcomings in other types of medical images. In this paper, we propose a more comprehensive method to enhance the embryonic image by genetic algorithm. This method uses genetic algorithm as a meta-heuristic to generate a proper composite filter using a set of filter from a filter bank. The outline of this paper is arranged as follows: Firstly, we describe our method then proposed method experimental results are presented and we draw a comparison between the results of this method and some other methods. Finally, a fair conclusion has been made. II. PROPOSED METHOD In This work, we present a new evolutionary method to enhance the Embryo cell image taken from the Time-Lapse device. Using genetic algorithm, a set of filters were combined and created a composite filter. The reason of using such evolutionary method is to specify the filter type and its parameters. General flowchart of the proposed method is illustrated in Fig. 1. First, a preprocessing phase including resizing, converting to grayscale is applied. Then, as will be explained further in the next sub-section, a genetic algorithm is prepared for selecting the best combination of filters. Figure 1. Flowchart of the general method. 141 2017 3rd International Conference on Frontiers of Signal Processing 978-1-5386-1038-1/17/$31.00 ©2017 IEEE