A proposed Image Processing Framework for Diagnosis of Anemia with Providing Proper Nutrition A. G. Sedki S. A. Shaban D. L. Elsheweikh Department of Home Economics Department of Computer Science Faculty of Specific Education Faculty of Specific Education Mansoura, Egypt. Mansoura, Egypt. amalgamil@mans.edu.eg soh.shabaan@gmail.com dalialotfy686@gmail.com Abstract— Anemia is one of the most widespread diseases which people face in developing countries. The traditional method used to detect anemia is a blood sample test, but this method is costly and requires a lot of time. This paper aims to design an automatic, fast, and reliable system for detecting anemia for adults whether a man or a woman. Also, it aims to form daily healthy meals for adults after calculating their daily nutritional requirements using the food exchange system. The image processing techniques are used in the proposed system for analysis of the anterior conjunctiva of the eye, tongue, and nails. Features are extracted by many methods such as color coherence vector and local binary pattern. For classification process weighted Euclidean distance and k-nearest neighbor methods are used. Nutritional requirements are computed according to weight, height, age, and physical activity. Also, the proposed system offers a set of nutritional advice for healthy adults. The proposed system achieved a high performance level when compared with the results of laboratory analysis. Keywords-: Anemia; Image Processing; Food Exchange System; Color Coherence Vector; Local Binary Pattern; Gray Level Co-occurrence Matrix. I. INTRODUCTION A healthy balanced diet should contain a sufficient amount of iron to meet the needs of a person's body. Taking small amounts of iron, folic acid, and nutrients that boost iron absorption are the main factors for the high prevalence of anemia [1]. The level of blood Hemoglobin (Hb) is used for measuring anemia. Hb is a red blood cell protein that carries oxygen to all of the body cells [2]. The World Health Organization (WHO) statistics refer to people nearly 1.62 billion all over the world suffer from anemia. Anemia has the highest prevalence in countries in which income is low and middle [3]. The WHO organization defines anemia as Hb < 13 g/dL in adult men, <12 g/dL in non-pregnant adult women, and <11 g/dL in pregnant adult women [4]. Anemia has a bad effect on health, such as lethargy, difficulty in concentrating, and fatigue, which reduces economic productivity and life quality [5]. There are some methods to check for anemia or not. One of them is through a medical examination in a laboratory and others through consultation of a doctor [6]. For the rapid detection of anemia, the doctor examines the colors of the anterior conjunctiva of the eye, tongue, and fingernails. In the absence of anemia, the color of the anterior conjunctiva of the eye is full or almost complete redness. However, in the case of anemia, it is pale [7]. A healthy tongue should be pink and covered with small nodules (papillae). Changes in color and texture of the tongue are signs of anemia [8]. In the absence of anemia, the nails color is pink. However, in the case of anemia, it is white or pale [9]. The main problem to detect anemia in some developing countries is the cost, which makes the people ignored to go to a laboratory or a doctor. Therefore, there is a need for other inexpensive, fast and accurate methods for detection of anemia. Computer-aided diagnosis tool uses image processing techniques to help in increasing the diagnosis accuracy and also its speed [10]. So image processing in the field of medicine assists medical practitioners in diagnosing patient’s disease [11]. There are few systems have developed for detection of anemia. K-means algorithm had proposed for detection of anemia based on the eye analysis by Sevani et al. [12]. Automated detection of anemia by analyzing the anterior conjunctiva pallor of the eye through a non-invasive visual method had designed by Tamir et al. [13]. Nail image processing for different disease diagnoses had presented by Indi et al. [14]. Nail color changes had used using the support vector machine to diagnose the disease by Nithya et al. [15]. C4.5 algorithms had used for analyzing human hand nail to identify many diseases by Gunge et al. [16]. An automated tongue diagnostic system on mobile-enabled platforms had proposed by Tania et al. [17]. The pallor sites of conjunctiva and tongue had analyzed by classifiers called Azure-based Generalized flow for Medical Image Classification (AGMIC) for anemia screening by Roychowdhury et al. [18]. From the above mentioned, it is clear that the detection of anemia can base on the analysis of anterior conjunctiva pallor of the eye, tongue, and nails. Therefore, this paper aims to design an automated, quick, simple, reliable, and inexpensive system for the detection of anemia in adults by using the image processing technique. And also, it aims to form daily healthy meals from Egyptian food items, especially those who suffered from anemia according to their daily nutritional requirements. II. THE PROPOSED FRAMEWORK The proposed system consists of two stages. The first stage, is the diagnosis, and the second stage, is the nutrition. In the diagnosis stage, the anterior conjunctiva pallor of the eye, tongue, and nails are analyzed to know whether anemia or not. International Journal of Computer Science and Information Security (IJCSIS), Vol. 18, No. 7, July 2020 22 https://sites.google.com/site/ijcsis/ ISSN 1947-5500