A novel transfer learning approach for the classification of histological images of colorectal cancer

Colorectal cancer (CRC) is the second most diagnosed cancer in the United States. It is identified by histopathological evaluations of microscopic images of the cancerous region, relying on a subjective interpretation. The Colorectal Histology dataset used in this study contains 5000 images, made available by the University Medical Center Mannheim. This approach proposes the automatic identification of eight types of tissues found in CRC histopathological evaluation. We apply Transfer Learning from architectures of Convolutional Neural Networks (CNNs). We modify the structures of CNNs to extract features from the images and input them to well-known machine learning methods: Naive Bayes, Multilayer Perceptron, k-Nearest Neighbors, Random Forest, and Support Vector Machine (SVM). We evaluated 108 extractor–classifier combinations. The one that achieved the best results is DenseNet169 with SVM (RBF), reaching an Accuracy of 92.083% and F1-Score of 92.117%. Therefore, our approach is capable of distinguishing tissues found in CRC histopathological evaluation.

     

Broadmark: A Testing Framework for Broad-Phase Collision Detection Algorithms

Research in the area of collision detection permeates most of the literature on simulations, interaction and agents planning, being commonly regarded as one of the main bottlenecks for large-scale systems. To this day, despite its importance, most subareas of collision detection lack a common ground to test and validate solutions, reference implementations and widely accepted benchmark suites. In this paper, we delve into the broad-phase of collision detection systems, providing both an open-source framework, named Broadmark, to test, compare and validate algorithms, and an in-deep analysis of the main techniques used so far to tackle the broad-phase problem. The technical challenges of building this framework from the software and hardware perspectives are also described. Within our framework, several original and state-of-the-art implementations of CPU and GPU algorithms are bundled, alongside three benchmark scenes to stress algorithms under several conditions. Furthermore, the system is designed to be easily extensible. We use our framework to bring out an extensive performance comparison among assembled solutions, detailing the current CPU and GPU state-of-the-art on a common ground. We believe that Broadmark encompasses the principal insights and tools to derive and evaluate novel algorithms, thus greatly facilitating discussion about successful broad-phase collision detection solutions.     

Optimization of the Acceptance of Prebiotic Beverage Made from Cashew Nut Kernels and Passion Fruit Juice

The aim of this research was to develop a prebiotic beverage from a hydrosoluble extract of broken cashew nut kernels and passion fruit juice using response surface methodology in order to optimize acceptance of its sensory attributes. A 2² central composite rotatable design was used, which produced 9 formulations, which were then evaluated using different concentrations of hydrosoluble cashew nut kernel, passion fruit juice, oligofructose, and 3% sugar. The use of response surface methodology to interpret the sensory data made it possible to obtain a formulation with satisfactory acceptance which met the criteria of bifidogenic action and use of hydrosoluble cashew nut kernels by using 14% oligofructose and 33% passion fruit juice.     

New computational solution to quantify synthetic material porosity from optical microscopic images

This paper presents a new computational solution to quantify the porosity of synthetic materials from optical microscopic images. The solution is based on an artificial neuronal network of the multilayer perceptron type and a backpropagation algorithm is used for training. To evaluate this new solution, 40 sample images of a synthetic material were analysed and the quality of the results was confirmed by human visual analysis. In addition, these results were compared with ones obtained with a commonly used commercial system confirming their superior quality and the shorter time needed. The effect of images with noise was also studied and the new solution showed itself to be more reliable. The training phase of the new solution was analysed confirming that it can be performed in a very easy and straightforward manner. Thus, the new solution demonstrated that it is a valid and adequate option for researchers, engineers, specialists and other professionals to quantify the porosity of materials from microscopic images in an automatic, fast, efficient and reliable manner.     

Automatic Detection of COVID-19 Infection Using Chest X-Ray Images Through Transfer Learning

The new coronavirus (COVID-19), declared by the World Health Organization as a pandemic, has infected more than 1 million people and killed more than 50 thousand. An infection caused by COVID-19 can develop into pneumonia, which can be detected by a chest X-ray exam and should be treated appropriately. In this work, we propose an automatic detection method for COVID-19 infection based on chest X-ray images. The datasets constructed for this study are composed of 194 X-ray images of patients diagnosed with coronavirus and 194 X-ray images of healthy patients. Since few images of patients with COVID-19 are publicly available, we apply the concept of transfer learning for this task. We use different architectures of convolutional neural networks (CNNs) trained on ImageNet, and adapt them to behave as feature extractors for the X-ray images. Then, the CNNs are combined with consolidated machine learning methods, such as k-Nearest Neighbor, Bayes, Random Forest, multilayer perceptron (MLP), and support vector machine (SVM). The results show that, for one of the datasets, the extractor-classifier pair with the best performance is the MobileNet architecture with the SVM classifier using a linear kernel, which achieves an accuracy and an F1-score of 98.5%. For the other dataset, the best pair is DenseNet201 with MLP, achieving an accuracy and an F1-score of 95.6%. Thus, the proposed approach demonstrates efficiency in detecting COVID-19 in X-ray images.      

Hepatitis B surface antigen carrier state in hepatosplenic schistosomiasis

The prevalence of hepatitis B surface antigen (HBsAg) was studied in 103 cases of hepatosplenic schistosomiasis (HSS), 134 control cases with a variety of illnesses including hepatointestinal schistosomiasis, and 600 blood donors, in an area endemic for both schisfosomiasis and viral hepatitis. The patients with HSS proved to be persistent carriers for HBsAg in a significantly higher proportion than the other two groups of cases. The HSS cases who were carriers of HBsAg had more clinical signs of chronic liver disease and strikingly more chronic inflammation of the portal spaces on liver biopsy. It is suggested that abnormal immunological responses in patients with HSS makes them more susceptible to become carriers of HBsAg and that the addition of this injurious factor makes their basic disease worse, and may be responsible for the development of cirrhosis in some cases.     

Digital twin modeling for predictive maintenance of gearboxes in floating offshore wind turbine drivetrains

This paper presents a multi-degree of freedom torsional model of drivetrain system as the digital twin model for monitoring the remaining useful lifetime of the drivetrain components. An algorithm is proposed for the model identification, which receives the torsional response and estimated values of rotor and generator torques, and calculates the drivetrain dynamic properties, e.g. eigenvalues, and torsional model parameters. The applications of this model in prediction of gearbox remaining useful lifetime is discussed. The proposed method is computationally fast, and can be implemented by integrating with the current turbine control and monitoring system without a need for a new system and sensors installation. A test case, using 5 MW reference drivetrain, has been demonstrated.

     

Ultrastructural and antimicrobial impacts of allyl isothiocyanate incorporated in cellulose, β-cyclodextrin,and carbon nanotubes nanocomposites

Antimicrobial nanobio packaging with controlled-release of active compounds is one of the most promising versions of active packaging. In this study, the antimicrobial effect of allyl isothiocyanate (AIT), complexed or not with β-cyclodextrin (βCD), and carbon nanotubes (CNTs) in cellulose films against Salmonella choleraesuis and Listeria innocua was investigated. Structural changes caused by AIT in bacteria were evaluated by transmission electron microscopy (TEM), as well as changes in the surface of the films, which were analyzed by atomic force microscopy and scanning electron microscopy. The active films inhibited the growth of both bacteria and caused damage to the cell membrane. The presence of inclusion complex and CNTs resulted in structural changes in films, such as the formation of lumps and reduction of roughness, respectively. Complexation of AIT with βCD and the use of CNTs increased the retention of the antimicrobial agent, which is desired to promote its controlled diffusion and consequently increase the preservative action of the film. However, when considering the use of βCD inclusion complexes, caution is necessary to prevent detrimental changes in the films' surface.