IJSTE - International Journal of Science Technology & Engineering | Volume 1 | Issue 11 | May 2015 ISSN (online): 2349-784X All rights reserved by www.ijste.org 338 Content Based Medical Image Retrieval using Artificial Neural Network Preetika D’Silva P. Bhuvaneswari PG Student Associate Professor Department of Electronics and Communication Department of Electronics and Communication Rajarajeswari College of Engineering, Bangalore Rajarajeswari College of Engineering, Bangalore Abstract Large database of images in the field of medicine requires proper systems that will help in accurate diagnostics and their efficient management. Content based medical image retrieval is a system that helps to browse, explore, find, and retrieve images similar to the query image with minimal user input. In this paper we propose a system that will retrieve all medical images that matches the query image. Shape and texture features are extracted from the pre-processed medical images for creating the medical database. Once the medical database is created, the features of the query image are extracted and are used by the neural network to train it. Euclidean distance between the database features and the query features are computed, ranked and we label the relevant images from the initial retrieved images. Then the feed forward back propagation neural network is used finally to retrieve the similar medical images. We have taken X-ray images of hand, foot, chest, head and ankle. The precision and recall values for the retrieval system using only texture features, using only shape features and using combined texture and shape features are calculated and compared. Keywords: Content based Medical Image Retrieval, Euclidean Distance GLCM, Neural network, Zernike ________________________________________________________________________________________________________ I. INTRODUCTION People can create, then process and store images on the internet using many available resources. This has produced the requirement for a way to manage and search these images. Hence, finding efficient image retrieval mechanisms from large resources has become an extensive area of interest to researchers [2] [11]. Searching and retrieving images from a huge database of images is performed by a method called as image retrieval. In today‘s modern age, virtually all spheres of human life including hospitals, surveillance, crime prevention, commerce, architecture engineering, fashion and graphic design, journalism, academics, government, and historical research utilize images for efficient services. Image data are integrated and stored in a system called an image database. Image data consists of the raw images and information extracted from images by automated or computer assisted image analysis. [2] Content based medical image retrieval (CBMIR) [3] [6] [7] [8] is the digital image searching problem in huge database that utilizes the contents of image themselves rather than depending on the textual information. Hospitals generate many medical images like X-ray, CT, and MRI of different parts of the human body contain semantic information. We can use this information to retrieve images. Under such situation, medical image retrieval has changed from earlier text based method to the content based method or a combination of the two methods. [3] This paper is organized as follows: The proposed system is described in Section II, Shape feature extraction is described in Section III, Texture feature extraction is described in Section IV, Euclidean distance is described in Section V, Artificial neural network is described in Section VI. In Section VII we describe the implementation results, in Section VIII we describe the performance evaluation and in Section IX we draw the respective conclusion. II. PROPOSED SYSTEM The fig.1 shows the block diagram of the proposed system. Firstly, the medical images are pre-processed. Then features are extracted from them and stored in a database. In the testing phase, a query image is pre-processed and different features are extracted. Then Euclidean distance between the query features and the database features are calculated and the images are sorted and labelled. These labelled features along with the database features are then used by the feed forward neural network to retrieve the medical images.s