Medical applications rely upon suitable interpretation of medical images and benefit from gaining access to visually similar images. In this paper we present a novel and computationally simple approach for retrieval of images from medical image database. The proposed approach consists of extracting feature vectors from medical images of different categories & modalities by exploiting the higher degree of correlation that exists amongst the pixel values of a medical image. Retrieval of images in this approach is based upon correlation which is classified into two types: Intra Pixel Correlation Correlation that exists among pixels in an image and statistical correlation, also termed as Inter Pixel Correlation that exists among extracted feature vectors of query and Target (database) Images. A transformation of the image pixels to a bit map is done based on Intra Pixel Correlation. Medical images ranging across various categories utilize shape as the fundamental distinguishing criterion for identifying the specific category, which is ensured in this approach by the incorporation of edge detector (Canny) after mapping into bit-plane domain. Inter-pixel correlation accounts for quantifying the similarity between image matrices of query and target (database) images. This provides better refinement and henceforth retrieves images with higher degree of recall and precision rate. Experimental results illustrate that the proposed approach is distinct and efficient compared to the existing techniques. © 2011 IEEE.