Wavelet transform for cardiac image retrieval

In this chapter, we present an overview of use of various wavelet transforms in content-based image retrieval method for diagnosis aid in cardiac fields. We start by describing the wavelet properties that are the most important for biocardiac applications. The retrieval performance analysis is provided for various orthogonal, biorthogonal, and Gabor wavelet transforms. The feature vector, used to measure similarity, is derived by computing the energy and standard deviation from the various wavelet filtered coefficients. The relevance similarity is ranked according to the closest similar measures computed by the Manhattan distance. The performance of all the wavelet transforms is evaluated on three different publicly available cardiac image databases, namely NEMA, OASIS, and EXACT09. The performance in terms of average retrieval precision (ARP) and average retrieval rate (ARR) is compared for all the wavelet transforms.