Case retrieval in medical databases by fusing heterogeneous information.

Department of Image et Traitement de l'Information, Institut Telecom/Telecom Bretagne, F-29200 Brest, France.
IEEE transactions on medical imaging 01/2011; 30(1):108-18. DOI: 10.1109/TMI.2010.2063711
Source: PubMed

ABSTRACT A novel content-based heterogeneous information retrieval framework, particularly well suited to browse medical databases and support new generation computer aided diagnosis (CADx) systems, is presented in this paper. It was designed to retrieve possibly incomplete documents, consisting of several images and semantic information, from a database; more complex data types such as videos can also be included in the framework. The proposed retrieval method relies on image processing, in order to characterize each individual image in a document by their digital content, and information fusion. Once the available images in a query document are characterized, a degree of match, between the query document and each reference document stored in the database, is defined for each attribute (an image feature or a metadata). A Bayesian network is used to recover missing information if need be. Finally, two novel information fusion methods are proposed to combine these degrees of match, in order to rank the reference documents by decreasing relevance for the query. In the first method, the degrees of match are fused by the Bayesian network itself. In the second method, they are fused by the Dezert-Smarandache theory: the second approach lets us model our confidence in each source of information (i.e., each attribute) and take it into account in the fusion process for a better retrieval performance. The proposed methods were applied to two heterogeneous medical databases, a diabetic retinopathy database and a mammography screening database, for computer aided diagnosis. Precisions at five of 0.809 ± 0.158 and 0.821 ± 0.177, respectively, were obtained for these two databases, which is very promising.

  • [Show abstract] [Hide abstract]
    ABSTRACT: Content-based image retrieval (CBIR) is an image search technique that utilises visual features as search criteria; it has potential clinical applications in evidence-based diagnosis, physician training, and biomedical research. Graph-based CBIR techniques have high accuracy when retrieving images by the similarity of the spatial arrangement of their constituent objects but these techniques were initially designed for single-modality images and have limited retrieval capabilities when multi-modality images, such as combined positron emission tomography and computed tomography (PET-CT), are considered. In this paper, we present a graph-based CBIR approach for multimodality images that integrates modality-specific features on graph vertices and adapts a well-established graph similarity scheme to account for varying vertex feature sets. Furthermore, we propose a graph pruning method that removes redundant edges using the spatial proximity of image regions. We evaluated our work using two simulated data sets, consisting of 2D liver shapes and 3D whole-body lymphoma images. In our experiments we achieved a higher level of retrieval precision using our graph method when compared to conventional graph-based retrieval, demonstrating that our proposed method enabled new capabilities and improved multi-modality CBIR.
    Computer-Based Medical Systems (CBMS), 2012 25th International Symposium on; 01/2012
  • [Show abstract] [Hide abstract]
    ABSTRACT: To realize the medical semantic annotation of mammogram, a semantic modeling approach for calcifications in mammogram based on hierarchical Bayesian network was proposed. Firstly, support vector machines was used to map low-level image feature into feature semantics, then high-level semantic was captured through feature semantic fusion using Bayesian network, finally semantic model was established. To validate the method, the model was applied to annotate the semantic information of mammograms. In this experiment, we chose 142 images as training set and 50 images as testing set, the results showed that the precision ratio of malignant samples is 81.48%, and benign samples is 73.91%.
    Strategic Technology (IFOST), 2011 6th International Forum on; 01/2011
  • Source

Full-text (2 Sources)

Available from
May 16, 2014