João Batista Neto

Improving the ranking quality of medical image retrieval using a genetic feature selection method

In this paper, we take advantage of single-valued functions that evaluate rankings to develop a family of feature selection methods based on the genetic algorithm approach, tailored to improve the accuracy of content-based image retrieval systems. Experiments on three image datasets, comprising images of breast and lung nodules, showed that developing functions to evaluate the ranking quality allows improving retrieval performance. This approach produces significantly better results than those of other fitness function approaches, such as the traditional wrapper and than filter feature selection algorithms.

Segmentation of Large Images with Complex Networks

Image segmentation is still a challenging issue in pattern recognition. Among the various segmentation approaches are those based on graph partitioning, which present some drawbacks, one being high processing times. With the recent developments on complex networks theory, pattern recognition techniques based on graphs have improved considerably. The identification of cluster of vertices can be considered a process of community identification according to complex networks theory. Since data clustering is related with image segmentation, image segmentation can also be approached via complex networks. However, image segmentation based on complex networks poses a fundamental limitation which is the excessive numbers of nodes in the network. This paper presents a complex network approach for large image segmentation that is both accurate and fast. To that, we incorporate the concept of super pixels, to reduce the number of nodes in the network. We evaluate our method for both synthetic and real images. Results show that our method can outperform other graph-based methods both in accuracy and processing times.

Medical image retrieval and analysis by Markov random fields and multi-scale fractal dimension.

Many Content-based Image Retrieval (CBIR) systems and image analysis tools employ color, shape and texture (in a combined fashion or not) as attributes, or signatures, to retrieve images from databases or to perform image analysis in general. Among these attributes, texture has turned out to be the most relevant, as it allows the identification of a larger number of images of a different nature. This paper introduces a novel signature which can be used for image analysis and retrieval. It combines texture with complexity extracted from objects within the images. The approach consists of a texture segmentation step, modeled as a Markov Random Field process, followed by the estimation of the complexity of each computed region. The complexity is given by a Multi-scale Fractal Dimension. Experiments have been conducted using an MRI database in both pattern recognition and image retrieval contexts. The results show the accuracy of the proposed method in comparison with other traditional texture descriptors and also indicate how the performance changes as the level of complexity is altered.

Feature selection with equalized salience measures and its application to segmentation

Segmentation is a crucial step in computer vision in which texture plays an important role. The existence of a large amount of methods from which texture can be computed is, sometimes, a hurdle to overcome when it comes to modeling solutions for texture-based segmentation. Following the excellence of the natural vision system and its generality, this work has adopted a feature selection method based on salience of synaptic connections of a Multilayer Perceptron neural network. Unlike traditional approaches, this paper introduces an equalization scheme to salience measures which contributed to significantly improve the selection of the most suitable features and, hence, yield better segmentation. The proposed method is compared with exhaustive search according to the Jeffrey-Matusita distance criterion. Segmentation for images of natural scenes has also been provided as a probable application of the method.

An Adjustable Error Measure for Image Segmentation Evaluation

Due to the subjective nature of the segmentation process, quantitative evaluation of image segmentation methods is still a difficult task. Humans perceive image objects in different ways. Consequently, human segmentations may come in different levels of refinement, ie, under- and over-segmentations. Popular segmentation error measures in the literature (Arbelaez and OCE) are supervised methods (also called empirical discrepancy methods), in which error is computed by comparing objects in segmentations with a reference (ground-truth) image produced by humans. Since reference images can be many, the key issue for a segmentation error measure is to be consistent in the presence of both under- and over-segmentation. In general, the term consistency refers to the ability of the error measure to be low, when comparing similar segmentations, or high, when faced with different segmentations, while capturing under-or over-segmentations. In this paper we propose a new object-based empirical discrepancy error measure, called Adjustable Object-based Measure (AOM). We introduce a penalty parameter which gives the method the ability to be more (or less) responsive in the presence of over-segmentation. Hence, we extend the notion of consistency so as to include the applications need in the process. Some applications require segmentation to be extremely accurate, hence under-or over-segmentation should be well penalised. Others, do not. By changing the penalty parameter, AOM can deliver more consistent results not only in reference to the under-or over-segmentation issue alone, but also according to the nature of the application. We compare our method with Arbelaez (used as standard measure in the benchmark of Berkeley Segmentation Image Dataset) and OCE. Our results show that AOM not only is more consistent in the presence of over-segmentation, but is also faster to compute. Unlike Arbelaez and OCE, AOM also satisfies the metric axiom of symmetry.

Class-specific metrics for multidimensional data projection applied to CBIR

Content-based image retrieval is still a challenging issue due to the inherent complexity of images and choice of the most discriminant descriptors. Recent developments in the field have introduced multidimensional projections to burst accuracy in the retrieval process, but many issues such as introduction of pattern recognition tasks and deeper user intervention to assist the process of choosing the most discriminant features still remain unaddressed. In this paper, we present a novel framework to CBIR that combines pattern recognition tasks, class-specific metrics, and multidimensional projection to devise an effective and interactive image retrieval system. User interaction plays an essential role in the computation of the final multidimensional projection from which image retrieval will be attained. Results have shown that the proposed approach outperforms existing methods, turning out to be a very attractive alternative for managing image data sets.

Silhouette-based feature selection for classification of medical images

Classification is an important task for computer-aided diagnosis systems (CADs). However, many classifiers may not perform well, presenting poor generalization and high computational cost, especially when dealing with high-dimensional datasets. Thus, feature selection can greatly mitigate these problems. In this paper, we propose two filter-based feature selection algorithms that calculate the simplified silhouette statistic as evaluation function: the silhouette-based greedy search (SiGS) and the silhouette-based genetic algorithm search (SiGAS). Silhouette statistic is used to guide the search for features that provide better class separability. Experiments performed on three datasets have shown that the SiGAS algorithm overcomes traditional filter algorithms, such as CFS, FCBF and reliefF. It also outperforms a similar algorithm, kNNGAS, based on genetic algorithm that minimizes the classification error of k-nearest neighbors. Additionally, results have shown that SiGAS produces better accuracy than SiGS.

Improving Non-local Video Denoising with Local Binary Patterns and Image Quantization

The most challenging aspect of video and image denoising is to preserve texture and small details, while filtering out noise. To tackle such problem, we present two novel variants of the 3D Non-Local Means (NLM3D) which are suitable for videos and 3D images. The first proposed algorithm computes texture patterns for each pixel by using the LBP-TOP descriptor to modify the NLM3D weighting function. It also uses MSB (Most Significant Bits) quantization to improve robustness to noise. The second proposed algorithm filters homogeneous and textured regions differently. It analyses the percentage of non-uniform LBP patterns of a region to determine whether or not the region exhibits textures and/or small details. Quantitative and qualitative experiments indicate that the proposed approaches outperform well known methods for the video denoising task, especially in the presence of textures and small details.

Projection-Based Image Retrieval Using Class-Specific Metrics

Content-based image classification/retrieval based on image descriptors has become an essential component in most database systems. However, most existing systems do not provide mechanisms that enable interactive multi-objective queries, hampering the user experience. In this paper we present a novel methodology capable of accomplishing multi-objective searches while still being interactive. Our approach relies on a combination of class-specific metrics and multidimensional projection to devise an effective and interactive image retrieval system. Besides allowing visual exploration of image data sets, the provided results and comparisons show that the proposed approach outperforms existing methods, turning out to be a very attractive alternative for managing image data sets.

Patient-Perceived Barriers to Accessing Cleft Care at a Tertiary Referral Center in São Paulo, Brazil

Background: In low- and middle-income countries, poor access to care can result in delayed surgical repair of orofacial clefts leading to poor functional outcomes. Even in Brazil, an upper middle-income country with free comprehensive cleft care, delayed repair of orofacial clefts commonly occurs. This study aims to assess patient-perceived barriers to cleft care at a referral center in São Paulo. Methods: A 29-item questionnaire assessing the barriers to care was administered to 101 consecutive patients (or their guardians) undergoing orofacial cleft surgery in the Plastic Surgery Department in Hospital das Clínicas, in São Paulo, Brazil, between February 2016 and January 2017. Results: A total of 54.4% of patients had their first surgery beyond the recommended time frame of 6 months for a cleft lip or cleft lip and palate and 18 months for a cleft palate. There was a greater proportion of isolated cleft palates in the delayed group (66.7% vs 33.3%). Almost all patients had a timely diagnosis, but delays occurred from diagnosis to repair. The mean number of barriers reported for each patient was 3.8. The most frequently cited barriers related to lack of access to care include (1) lack of hospitals available to perform the surgery (54%) and (2) lack of availability of doctors (51%). Conclusion: Delays from diagnosis to treatment result in patients receiving delayed primary repairs. The commonest patient-perceived barriers are related to a lack of access to cleft care, which may represent a lack of awareness of available services.