RESUMO

BACKGROUND AND OBJECTIVE: Convolutional Neural Networks (CNNs) can provide excellent results regarding the segmentation of blood vessels. One important aspect of CNNs is that they can be trained on large amounts of data and then be made available, for instance, in image processing software. The pre-trained CNNs can then be easily applied in downstream blood vessel characterization tasks, such as the calculation of the length, tortuosity, or caliber of the blood vessels. Yet, it is still unclear if pre-trained CNNs can provide robust, unbiased, results in downstream tasks involving the morphological analysis of blood vessels. Here, we focus on measuring the tortuosity of blood vessels and investigate to which extent CNNs may provide biased tortuosity values even after fine-tuning the network to a new dataset under study. METHODS: We develop a procedure for quantifying the influence of CNN pre-training in downstream analyses involving the measurement of morphological properties of blood vessels. Using the methodology, we compare the performance of CNNs that were trained on images containing blood vessels having high tortuosity with CNNs that were trained on blood vessels with low tortuosity and fine-tuned on blood vessels with high tortuosity. The opposite situation is also investigated. RESULTS: We show that the tortuosity values obtained by a CNN trained from scratch on a dataset may not agree with those obtained by a fine-tuned network that was pre-trained on a dataset having different tortuosity statistics. In addition, we show that improving the segmentation accuracy does not necessarily lead to better tortuosity estimation. To mitigate the aforementioned issues, we propose the application of data augmentation techniques even in situations where they do not improve segmentation performance. For instance, we found that the application of elastic transformations was enough to prevent an underestimation of 8% of blood vessel tortuosity when applying CNNs to different datasets. CONCLUSIONS: The results highlight the importance of developing new methodologies for training CNNs with the specific goal of reducing the error of morphological measurements, as opposed to the traditional approach of using segmentation accuracy as a proxy metric for performance evaluation.

Assuntos

RESUMO

Prostate cancer (PCa) is a pervasive condition that is manifested in a wide range of histologic patterns in biopsy samples. Given the importance of identifying abnormal prostate tissue to improve prognosis, many computerized methodologies aimed at assisting pathologists in diagnosis have been developed. It is often argued that improved diagnosis of a tissue region can be obtained by considering measurements that can take into account several properties of its surroundings, therefore providing a more robust context for the analysis. Here we propose a novel methodology that can be used for systematically defining contextual features regarding prostate glands. This is done by defining a Gland Context Network (GCN), a representation of the prostate sample containing information about the spatial relationship between glands as well as the similarity between their appearance. We show that such a network can be used for establishing contextual features at any spatial scale, therefore providing information that is not easily obtained from traditional shape and textural features. Furthermore, it is shown that even basic features derived from a GCN can lead to state-of-the-art classification performance regarding PCa. All in all, GCNs can assist in defining more effective approaches for PCa grading.

Assuntos

RESUMO

Blood leakage from the vessels in the eye is the hallmark of many vascular eye diseases. One of the preclinical mouse models of retinal blood leakage, the very-low-density-lipoprotein receptor deficient mouse (Vldlr-/-), is used for drug screening and mechanistic studies. Vessel leakage is usually examined using Fundus fluorescein angiography (FFA). However, interpreting FFA images of the Vldlr-/- model is challenging as no automated and objective techniques exist for this model. A pipeline has been developed for quantifying leakage intensity and area including three tasks: (i) blood leakage identification, (ii) blood vessel segmentation, and (iii) image registration. Morphological operations followed by log-Gabor quadrature filters were used to identify leakage regions. In addition, a novel optic disk detection algorithm based on graph analysis was developed for registering the images at different timepoints. Blood leakage intensity and area measured by the methodology were compared to ground truth quantifications produced by two annotators. The relative difference between the quantifications from the method and those obtained from ground truth images was around 10% ± 6% for leakage intensity and 17% ± 8% for leakage region. The Pearson correlation coefficient between the method results and the ground truth was around 0.98 for leakage intensity and 0.94 for leakage region. Therefore, we presented a computational method for quantifying retinal vascular leakage and vessels using FFA in a preclinical angiogenesis model, the Vldlr-/- model.

Assuntos

RESUMO

Many real-world systems can be studied in terms of pattern recognition tasks, so that proper use (and understanding) of machine learning methods in practical applications becomes essential. While many classification methods have been proposed, there is no consensus on which methods are more suitable for a given dataset. As a consequence, it is important to comprehensively compare methods in many possible scenarios. In this context, we performed a systematic comparison of 9 well-known clustering methods available in the R language assuming normally distributed data. In order to account for the many possible variations of data, we considered artificial datasets with several tunable properties (number of classes, separation between classes, etc). In addition, we also evaluated the sensitivity of the clustering methods with regard to their parameters configuration. The results revealed that, when considering the default configurations of the adopted methods, the spectral approach tended to present particularly good performance. We also found that the default configuration of the adopted implementations was not always accurate. In these cases, a simple approach based on random selection of parameters values proved to be a good alternative to improve the performance. All in all, the reported approach provides subsidies guiding the choice of clustering algorithms.

Assuntos

RESUMO

Studies regarding knowledge organization and acquisition are of great importance to understand areas related to science and technology. A common way to model the relationship between different concepts is through complex networks. In such representations, networks' nodes store knowledge and edges represent their relationships. Several studies that considered this type of structure and knowledge acquisition dynamics employed one or more agents to discover node concepts by walking on the network. In this study, we investigate a different type of dynamics adopting a single node as the "network brain." Such a brain represents a range of real systems such as the information about the environment that is acquired by a person and is stored in the brain. To store the discovered information in a specific node, the agents walk on the network and return to the brain. We propose three different dynamics and test them on several network models and on a real system, which is formed by journal articles and their respective citations. The results revealed that, according to the adopted walking models, the efficiency of self-knowledge acquisition has only a weak dependency on topology and search strategy.

RESUMO

Complex networks have been widely used to model biological systems. The concept of accessibility has been proposed recently as a means to organize the nodes of complex networks as belonging to its border or center. Such an approach paves the way to investigating how the functional and structural properties of nodes vary with their respective position in the networks. In this work, we approach such a problem in a biological context applying border detection to Protein-Protein Interaction networks from four organisms of the Mycoplasma genus. We found evidence that the borderness of proteins bears a relation with their spatial organization and molecular function specificity.

Assuntos

RESUMO

Adsorption processes are responsible for detection of cancer biomarkers in biosensors (and immunosensors), which can be captured with various principles of detection. In this study, we used a biosensor made with nanostructured films of polypyrrole and p53 antibodies, and image analysis of scanning electron microscopy data made it possible to correlate morphological changes of the biosensor with the concentration of cells containing the cancer biomarker p53. The selectivity of the biosensor was proven by distinguishing images obtained with exposure of the biosensor to cells containing the biomarker from those acquired with cells that did not contain it. Detection was confirmed with cyclic voltammetry measurements, while the adsorption of the p53 biomarker was probed with polarization-modulated infrared reflection absorption (PM-IRRAS) and a quartz crystal microbalance (QCM). Adsorption is described using the Langmuir-Freundlich model, with saturation taking place at a concentration of 100 Ucells/mL. Taken together, our results point to novel ways to detect biomarkers or any type of analyte for which detection is based on adsorption as is the case of the majority of biosensors.

Assuntos

RESUMO

Linearity is an important and frequently sought property in electronics and instrumentation. Here, we report a method capable of, given a transfer function (theoretical or derived from some real system), identifying the respective most linear region of operation with a fixed width. This methodology, which is based on least squares regression and systematic consideration of all possible regions, has been illustrated with respect to both an analytical (sigmoid transfer function) and a simple situation involving experimental data of a low-power, one-stage class A transistor current amplifier. Such an approach, which has been addressed in terms of transfer functions derived from experimentally obtained characteristic surface, also yielded contributions such as the estimation of local constants of the device, as opposed to typically considered average values. The reported method and results pave the way to several further applications in other types of devices and systems, intelligent control operation, and other areas such as identifying regions of power law behavior.

RESUMO

In the search for a cure for many muscular disorders it is often necessary to analyze muscle fibers under a microscope. For this morphological analysis, we developed an image processing approach to automatically analyze and quantify muscle fiber images so as to replace today's less accurate and time-consuming manual method. Muscular disorders, that include cardiomyopathy, muscular dystrophies, and diseases of nerves that affect muscles such as neuropathy and myasthenia gravis, affect a large percentage of the population and, therefore, are an area of active research for new treatments. In research, the morphological features of muscle fibers play an important role as they are often used as biomarkers to evaluate the progress of underlying diseases and the effects of potential treatments. Such analysis involves assessing histopathological changes of muscle fibers as indicators for disease severity and also as a criterion in evaluating whether or not potential treatments work. However, quantifying morphological features is time-consuming, as it is usually performed manually, and error-prone. To replace this standard method, we developed an image processing approach to automatically detect and measure the cross-sections of muscle fibers observed under microscopy that produces faster and more objective results. As such, it is well-suited to processing the large number of muscle fiber images acquired in typical experiments, such as those from studies with pre-clinical models that often create many images. Tests on real images showed that the approach can segment and detect muscle fiber membranes and extract morphological features from highly complex images to generate quantitative results that are readily available for statistical analysis.

Assuntos