This paper presents a classification-driven biomedical image retrieval framework based on image filtering and similarity fusion by employing supervised learning techniques. In this framework, the probabilistic outputs of a multiclass support vector machine (SVM) classifier as category prediction of query and database images are exploited at first to filter out irrelevant images, thereby reducing the search space for similarity matching. Images are classified at a global level according to their modalities based on different low-level, concept, and keypoint-based features. It is difficult to find a unique feature to compare images effectively for all types of queries. Hence, a query-specific adaptive linear combination of similarity matching approach is proposed by relying on the image classification and feedback information from users. Based on the prediction of a query image category, individual precomputed weights of different features are adjusted online. The prediction of the classifier may be inaccurate in some cases and a user might have a different semantic interpretation about retrieved images. Hence, the weights are finally determined by considering both precision and rank order information of each individual feature representation by considering top retrieved relevant images as judged by the users. As a result, the system can adapt itself to individual searches to produce query-specific results. Experiment is performed in a diverse collection of 5 000 biomedical images of different modalities, body parts, and orientations. It demonstrates the efficiency (about half computation time compared to search on entire collection) and effectiveness (about 10%-15% improvement in precision at each recall level) of the retrieval approach.

<span>Data mining is the process of finding structure of data from large data sets. With this process, the decision makers can make a particular decision for further development of the real-world problems. Several data clusteringtechniques are used in data mining for finding a specific pattern of data. The K-means method isone of the familiar clustering techniques for clustering large data sets.  The K-means clustering method partitions the data set based on the assumption that the number of clusters are fixed.The main problem of this method is that if the number of clusters is to be chosen small then there is a higher probability of adding dissimilar items into the same group. On the other hand, if the number of clusters is chosen to be high, then there is a higher chance of adding similar items in the different groups. In this paper, we address this issue by proposing a new K-Means clustering algorithm. The proposed method performs data clustering dynamically. The proposed method initially calculates a threshold value as a centroid of K-Means and based on this value the number of clusters are formed. At each iteration of K-Means, if the Euclidian distance between two points is less than or equal to the threshold value, then these two data points will be in the same group. Otherwise, the proposed method will create a new cluster with the dissimilar data point. The results show that the proposed method outperforms the original K-Means method.</span>

No abstract

No abstract

No abstract