Do we need annotation experts? A case study in celiac disease classification.

Inference of clinically-relevant findings from the visual appearance of images has become an essential part of processing pipelines for many problems in medical imaging. Typically, a sufficient amount labeled training data is assumed to be available, provided by domain experts. However, acquisition of this data is usually a time-consuming and expensive endeavor. In this work, we ask the question if, for certain problems, expert knowledge is actually required. In fact, we investigate the impact of letting non-expert volunteers annotate a database of endoscopy images which are then used to assess the absence/presence of celiac disease. Contrary to previous approaches, we are not interested in algorithms that can handle the label noise. Instead, we present compelling empirical evidence that label noise can be compensated by a sufficiently large corpus of training data, labeled by the non-experts.

On the role of correlation and abstraction in cross-modal multimedia retrieval.

The problem of cross-modal retrieval from multimedia repositories is considered. This problem addresses the design of retrieval systems that support queries across content modalities, for example, using an image to search for texts. A mathematical formulation is proposed, equating the design of cross-modal retrieval systems to that of isomorphic feature spaces for different content modalities. Two hypotheses are then investigated regarding the fundamental attributes of these spaces. The first is that low-level cross-modal correlations should be accounted for. The second is that the space should enable semantic abstraction. Three new solutions to the cross-modal retrieval problem are then derived from these hypotheses: correlation matching (CM), an unsupervised method which models cross-modal correlations, semantic matching (SM), a supervised technique that relies on semantic representation, and semantic correlation matching (SCM), which combines both. An extensive evaluation of retrieval performance is conducted to test the validity of the hypotheses. All approaches are shown successful for text retrieval in response to image queries and vice versa. It is concluded that both hypotheses hold, in a complementary form, although evidence in favor of the abstraction hypothesis is stronger than that for correlation.

Latent Dirichlet allocation models for image classification.

Two new extensions of latent Dirichlet allocation (LDA), denoted topic-supervised LDA (ts-LDA) and class-specific-simplex LDA (css-LDA), are proposed for image classification. An analysis of the supervised LDA models currently used for this task shows that the impact of class information on the topics discovered by these models is very weak in general. This implies that the discovered topics are driven by general image regularities, rather than the semantic regularities of interest for classification. To address this, ts-LDA models are introduced which replace the automated topic discovery of LDA with specified topics, identical to the classes of interest for classification. While this results in improvements in classification accuracy over existing LDA models, it compromises the ability of LDA to discover unanticipated structure of interest. This limitation is addressed by the introduction of css-LDA, an LDA model with class supervision at the level of image features. In css-LDA topics are discovered per class, i.e., a single set of topics shared across classes is replaced by multiple class-specific topic sets. The css-LDA model is shown to combine the labeling strength of topic-supervision with the flexibility of topic-discovery. Its effectiveness is demonstrated through an extensive experimental evaluation, involving multiple benchmark datasets, where it is shown to outperform existing LDA-based image classification approaches.

Holistic context models for visual recognition.

A novel framework to context modeling based on the probability of co-occurrence of objects and scenes is proposed. The modeling is quite simple, and builds upon the availability of robust appearance classifiers. Images are represented by their posterior probabilities with respect to a set of contextual models, built upon the bag-of-features image representation, through two layers of probabilistic modeling. The first layer represents the image in a semantic space, where each dimension encodes an appearance-based posterior probability with respect to a concept. Due to the inherent ambiguity of classifying image patches, this representation suffers from a certain amount of contextual noise. The second layer enables robust inference in the presence of this noise by modeling the distribution of each concept in the semantic space. A thorough and systematic experimental evaluation of the proposed context modeling is presented. It is shown that it captures the contextual "gist" of natural images. Scene classification experiments show that contextual classifiers outperform their appearance-based counterparts, irrespective of the precise choice and accuracy of the latter. The effectiveness of the proposed approach to context modeling is further demonstrated through a comparison to existing approaches on scene classification and image retrieval, on benchmark data sets. In all cases, the proposed approach achieves superior results.

Learning pit pattern concepts for gastroenterological training.

In this article, we propose an approach to learn the characteristics of colonic mucosal surface structures, the so called pit patterns, commonly observed during high-magnification colonoscopy. Since the discrimination of the pit pattern types usually requires an experienced physician, an interesting question is whether we can automatically find a collection of images which most typically show a particular pit pattern characteristic. This is of considerable practical interest, since it is imperative for gastroenterological training to have a representative image set for the textbook descriptions of the pit patterns. Our approach exploits recent research on semantic image retrieval and annotation. This facilitates to learn a semantic space for the pit pattern concepts which eventually leads to a very natural formulation of our task.