Caetano Traina Junior

Color and Texture Influence on Computer-Aided Diagnosis of Dermatological Ulcers

This study presents an analysis of classification techniques for Computer-Aided Diagnosis (CAD) regarding ulcerated lesions. We focus on determining influence of both color and texture in the automated image classification and its implication. To do so, we assayed a dataset of dermatological ulcers containing five variations in terms of tissue composition of lesion skin: granulation (red), fibrin (yellow), callous (white), necrotic (black), and a mix of the previous variations (mixed). Every image was previously labelled by experts regarding this red-yellow-black-white-mixed model. We employed specially designed color and texture extractors to represent the dataset images, namely: Color Layout, Color Structure, Scalable Color, Edge Histogram, Haralick, and Texture-Spectrum. The first three are color feature extractors and the last three are texture extractors. Following, we employed the Symmetrica Uncert Attribute Eval method to determine the features suitable for image classification. We tested a set of classifiers that follows distinct paradigms over the selected features, achieving an accuracy ratio of up to 77% in terms of images correctly classified, with the area under the receiver operating characteristic (ROC) curve up to 0.84. The classification performance and the selected features enabled us to determine that texture features were more predominant than color in the entire classification process.

BREATH: Heat Maps Assisting the Detection of Abnormal Lung Regions in CT Scans

Computed Tomography (CT) scans are often employed to diagnose lung diseases, as abnormal tissue regions may indicate whether proper treatment is required. However, detecting specific regions containing abnormalities in a CT scan demands time and effort of specialists. Moreover, different parts of a single lung image may present both normal and abnormal characteristics, what makes inaccurate the classification of a single lung as healthy (normal) or not. In this paper we propose the BREATH method, capable of detecting abnormalities in lung tissue regions, highlighting them by means of a heat map visualization. The method starts by segmenting lung tissues using a superpixel-based approach, followed by the training of a statistical model to represent normal tissues and, finally, the generation of a heat map showing abnormal regions that require attention from the physicians. We validated our statistical model using a dataset with 246 lung CT scans, where 40 are healthy and the remaining present varying diseases. Experimental results show that BREATH is accurate for lung segmentation with F-Measure of up to 0.99. The statistical modeling of healthy and abnormal lung regions has shown almost no overlap, and the detection of superpixels containing abnormalities presented precision values higher than 86%, for all values of recall. These values support our claim that the heat map representation of BREATH for the abnormal detection can be used as an intuitive method to assist physicians during the diagnosis.

Proceedings of the 7th International Conference on Similarity Search and Applications - SISAP

7th International Conference on Similarity Search and Applications (SISAP). Los Cabos, Mexico. 29-31 october 2014.Dataand model-driven computer simulations for understanding spatio-temporal dynamics of emerging phenomena are increasingly critical in various application domains, from predicting geo-temporal evolution of epidemics to helping reduce energy footprints of buildings leading to more sustainable building systems and architectural designs. These simulations track 10s or 100s of inter-dependent parameters, spanning multiple information layers and spatio-temporal frames, affected by complex dynamic processes operating at different resolutions. Consequently, the key characteristics of data sets and models relevant to these data-intensive simulations often include the following: (a) voluminous, (b) multi-variate, (c) multi-resolution, (d) spatio-temporal, and (e) inter-dependent. While very powerful and highly modular and flexible simulation software exists, because of the volume and complexity of the simulation data, the varying spatial and temporal scales at which the key transmission processes operate and relevant observations are made, today experts lack the means to adequately and systematically interpret observations, understand the underlying processes, and re-use of existing simulation results in new settings. In this talk, I will introduce computational challenges that arise from the need to process, index, search, and analyze, in a scalable manner, large volumes of temporal data resulting from data-intensive simulations and present some solutions.

Clustering Methods for Moderate-to-High Dimensionality Data

Traditional clustering methods are usually inefficient and ineffective over data with more than five or so dimensions. In Sect. 2.3 of the previous chapter, we discuss the main reasons that lead to this fact. It is also mentioned that the use of dimensionality reduction methods does not solve the problem, since it allows one to treat only the global correlations in the data. Correlations local to subsets of the data cannot be identified without the prior identification of the data clusters where they occur. Thus, algorithms that combine dimensionality reduction and clustering into a single task have been developed to look for clusters together with the subspaces of the original space where they exist. Some of these algorithms are briefly described in this chapter. Specifically, we first present a concise survey on the existing algorithms, and later we discuss three of the most relevant ones. Then, in order to help one to evaluate and to compare the algorithms, we conclude the chapter by presenting a table to link some of the most relevant techniques with the main desirable properties that any clustering technique for moderate-to-high dimensionality data should have. The general goal is to identify the main strategies already used to deal with the problem, besides the key limitations of the existing techniques.

Related Work and Concepts

This chapter presents the main background knowledge relevant to the book. Sections 2.1 and 2.2 describe the areas of processing complex data and knowledge discovery in traditional databases. The task of clustering complex data is discussed in Sect. 2.3, while the task of labeling such kind of data is described in Sect. 2.4. Section 2.5 introduces the MapReduce framework, a promising tool for large scale data analysis, which has been proven to offer one valuable support to the execution of data mining algorithms in a parallel processing environment. Section 2.6 concludes the chapter.