Cluster based-image descriptors and fractional hybrid optimization for medical image retrieval

Nowadays health information management (HIM) is a challenging area of research. In HIM, retrieving, storing and interpreting the information regarding the health of patients are considered as the significant stages. As a consequence, retrieving the earlier records of the case, based on the current information of patients helps in assisting medical practitioners in recognition of patients with similar problems and the curing process. On focusing this as an important objective of this study, an image retrieval system is proposed which utilizes visual features to describe the contents of the image. Initially, the input images associated with cases of patients are considered as input. Then, the features, such as Correlogram, LGP, wavelet moments and mean, variance, skew, kutoiss from BFC of the image are detected by the exploitation of image descriptors, and they are stored in the feature database. Then, various weights are allocated to every feature, and the Fractional hybrid optimization is proposed by merging fractional brain storm optimization (FBSO) with fractional lion algorithm (FLA) for optimal weight score generation. The simulation is done with six forms of medical images and the parameters, such as recall, precision and f-measure are utilized for distinguishing the performance of the conventional methods.