Deep learning model for fully automated breast cancer detection system from thermograms

Breast cancer is one of the most common diseases among women worldwide. It is considered one of the leading causes of death among women. Therefore, early detection is necessary to save lives. Thermography imaging is an effective diagnostic technique which is used for breast cancer detection with the help of infrared technology. In this paper, we propose a fully automatic breast cancer detection system. First, U-Net network is used to automatically extract and isolate the breast area from the rest of the body which behaves as noise during the breast cancer detection model. Second, we propose a two-class deep learning model, which is trained from scratch for the classification of normal and abnormal breast tissues from thermal images. Also, it is used to extract more characteristics from the dataset that is helpful in training the network and improve the efficiency of the classification process. The proposed system is evaluated using real data (A benchmark, database (DMR-IR)) and achieved accuracy = 99.33%, sensitivity = 100% and specificity = 98.67%. The proposed system is expected to be a helpful tool for physicians in clinical use.

A Systematic Literature Review of Cutting Tool Wear Monitoring in Turning by Using Artificial Intelligence Techniques

In turning operations, the wear of cutting tools is inevitable. As workpieces produced with worn tools may fail to meet specifications, the machining industries focus on replacement policies that mitigate the risk of losses due to scrap. Several strategies, from empiric laws to more advanced statistical models, have been proposed in the literature. More recently, many monitoring systems based on Artificial Intelligence (AI) techniques have been developed. Due to the scope of different artificial intelligence approaches, having a holistic view of the state of the art on this subject is complex, in part due to a lack of recent comprehensive reviews. This literature review therefore presents 20 years of literature on this subject obtained following a Systematic Literature Review (SLR) methodology. This SLR aims to answer the following research question: “How is the AI used in the framework of monitoring/predicting the condition of tools in stable turning condition?” To answer this research question, the “Scopus” database was consulted in order to gather relevant publications published between 1 January 2000 and 1 January 2021. The systematic approach yielded 8426 articles among which 102 correspond to the inclusion and exclusion criteria which limit the application of AI to stable turning operation and online prediction. A bibliometric analysis performed on these articles highlighted the growing interest of this subject in the recent years. A more in-depth analysis of the articles is also presented, mainly focusing on six AI techniques that are highly represented in the literature: Artificial Neural Network (ANN), fuzzy logic, Support Vector Machine (SVM), Self-Organizing Map (SOM), Hidden Markov Model (HMM), and Convolutional Neural Network (CNN). For each technique, the trends in the inputs, pre-processing techniques, and outputs of the AI are presented. The trends highlight the early and continuous importance of ANN, and the emerging interest of CNN for tool condition monitoring. The lack of common benchmark database for evaluating models performance does not allow clear comparisons of technique performance.

A Systematic and Comparative Analysis of Four Major Classes of DFT Functionals to Compute Linear and Nonlinear Optical Properties of Benchmark Molecules

This study spotlights the fundamental insights about the systematic and comparative analysis of four famous hybrid classes of density functional theory (DFT) methods and their efficacy to calculate the linear and nonlinear optical (NLO) polarizabilities. For this study, urea and para-nitroaniline ([Formula: see text]-NA) molecular geometries are used as prototypes to calculate their linear and NLO properties. For comparative purposes, these molecules are often used as reference organic molecules for determination of NLO response properties and there is a dire need for such a benchmark database to be utilized by the researchers. We report systematically a range of functionals including hybrid (B3LYP, PBE1PBE, BH and HLYP), meta-hybrid (M06, M06-2X, M06-HF, M06-L), long-range corrected (CAM-B3LYP, LC-BLYP, LC-B97D, LC-B97D3) and functional with dispersion correction ([Formula: see text]B97, [Formula: see text]B97X, [Formula: see text]B97XD, HSEH1PBE). These groups are evaluated and their efficiency to calculate linear and NLO properties is graphically compared with each other. Overall, there are less deviations among different functionals for calculating dipole moments of [Formula: see text]-NA and urea while these deviations enhance as one moves from dipole moment to linear polarizability and nonlinear hyperpolarizabilities. In general, if we look at the trends, the polarizability values of B3LYP, M06-L, CAM-B3LYP and HSEH1PBE are relatively large and can be compared with each other. The dispersion corrected and long-range corrected functionals show more systematic deviations. For instance, among dispersion corrected functionals, the amplitudes of dipole moments, linear polarizability and NLO polarizabilities show an increasing trend as [Formula: see text]. It is also important to note that LC-B97D and LC-B97D3 of long-range corrected functional have observed exactly the same values of all the calculated parameters. A good agreement is being observed in static first and second hyperpolarizabilities of urea (B3LYP, M06-L, M06 and HSEH1PBE) and [Formula: see text]-NA (B3LYP, M06, M06-L, CAM-B3LYP and HSEH1PBE). Thus, we believe that the current investigation will provide the benchmark data of reference NLO molecules at different methods for theoretical community and molecular level insights for experimental community to design better NLO materials for hi-tech NLO applications.

A Review on Speech Emotion Recognition Using Deep Learning and Attention Mechanism

Emotions are an integral part of human interactions and are significant factors in determining user satisfaction or customer opinion. speech emotion recognition (SER) modules also play an important role in the development of human–computer interaction (HCI) applications. A tremendous number of SER systems have been developed over the last decades. Attention-based deep neural networks (DNNs) have been shown as suitable tools for mining information that is unevenly time distributed in multimedia content. The attention mechanism has been recently incorporated in DNN architectures to emphasise also emotional salient information. This paper provides a review of the recent development in SER and also examines the impact of various attention mechanisms on SER performance. Overall comparison of the system accuracies is performed on a widely used IEMOCAP benchmark database.

Per-Covid-19: A Benchmark Database For Covid-19 Percentage Prediction From CT-scans

Covid-19 infection recognition is very important step in the fighting against the new pandemic Covid-19. In fact, many methods have been used to recognize the Covid-19 infection including Reverse transcription polymerase chain reaction (RT-PCR), X-ray scan and CT-scan. In addition to the recognition of the Covid-19 infection, CT-scans can provide more important information about the evolution of this disease and its severity. With the extensive number of Covid-19 infections, estimating the Covid-19 percentage can help the intensive care to free up the resuscitation beds for the critical cases and follow other protocol for less severity cases. In this paper, we propose Covid-19 percentage estimation database. Moreover, we evaluate the performance of three Covolutional Neural Network (CNN) architectures which are ResneXt-50, Densenet-161 and Inception-v3. For the three CNN architectures, we use two loss functions which are MSE and Dynamic Huber. In addition, two pretrained scenarios are investigated (ImageNet pretrained models and X-ray pretrained models). The evaluated approaches achieved promising results, where Inception-v3 with using Dynamic Huber loss function and X-ray pretrained model achieved the best performance.

Multisensor Thermal Infrared and Microwave Land Surface Temperature Algorithm Intercomparison

To ensure optimal and consistent algorithm usage within climate studies utilizing satellite-derived Land Surface Temperature (LST) datasets, an algorithm intercomparison exercise was undertaken to assess the various operational and scientific LST retrieval algorithms in use. This study was focused on several LST products including single-sensor products for AATSR, Terra-MODIS, SEVIRI, SSM/I and SSMIS; a Climate Date Record (CDR), which is a combined dataset drawing from AATSR, SLSTR and MODIS; and finally a merged low Earth orbit/geostationary product using data from AATSR, MODIS and SEVIRI. Therefore, the analysis included 14 algorithms: seven thermal infrared algorithms and seven microwave algorithms. The thermal infrared algorithms include five split-window coefficient-based algorithms, one optimal estimation algorithm and one single-channel inversion algorithm, with the microwave focusing on linear regression and neural network methods. The algorithm intercomparison assessed the performance of the retrieval algorithms for all sensors using a benchmark database. This approach was chosen due to the lack of sufficient in situ validation sites globally and the bias this limited set engendered on the training of particular algorithms. A simulated approach has the ability to test all parameters in a consistent, fair manner at a global scale. The benchmark database was constructed from European Centre for Medium-Range Weather Forecasts Re-analysis 5 (ERA5) atmospheric data, Combined ASTER and MODIS Emissivity for Land (CAMEL) infrared emissivity data, and Tool to Estimate Land Surface Emissivities at Microwave frequencies (TELSEM) emissivity data for the period of 2013–2015. The best-performing algorithms had biases of under 0.2 K and standard deviations of approximately 0.7 K. These results were consistent across multiple sensors. Areas of improvement, such as coefficient banding, were found for all algorithms as well as lines for further inquiry that could improve the global and regional performance.

r2SCAN-3c: An Efficient “Swiss Army Knife” Composite Electronic-Structure Method

The recently proposed second revision of the SCAN meta-GGA density-functional approximation (DFA) {Furness et al., J. Phys. Chem. Lett. 2020, 11, 8208-8215, termed r²SCAN} is used to construct an efficient composite electronic-structure method termed r²SCAN-3c, expanding the "3c'' series (hybrid: HSE/PBEh-3c, GGA: B97-3c, HF: HF-3c) to themGGA level. To this end, the unaltered r²SCAN functional is combined with a tailor-made <br>triple-zeta Gaussian AO-basis as well as with refitted D4 and gCP corrections for London-dispersion and basis-set superposition error. The performance of the new method is evaluated for the GMTKN55 thermochemical database covering large parts of chemical space with about 1500 <br>data points, as well as additional benchmarks for noncovalent interactions, organometallic reactions, lattice energies of organic molecules and ices, as well as for the adsorption on polar salt and non-polar coinage-metal surfaces. These comprehensive tests reveal a spectacular performance and robustness of r²SCAN-3c for reaction energies and noncovalent interactions in molecular and periodic systems, as well as outstanding conformational energies, and consistent structures. At just one tenth of the cost, r²SCAN-3c provides one of the best results of all semi-local DFT/QZ methods ever tested for the GMTKN55 benchmark database. Specifically for reaction and conformational energies as well as for noncovalent interactions, the new method outperforms hybrid-DFT/QZ approaches, compared to which the computational savings are even larger (factor 100-1000).<br>In relation to other "3c'' methods, r²SCAN-3c by far surpasses the accuracy of its predecessor B97-3c at only about twice the cost. The perhaps most relevant remaining systematic deviation of r²SCAN-3c is due to self-interaction-error, owing to its mGGA nature. However, SIE is notably reduced compared to other (m)GGAs, as is demonstrated for several examples. After all, this remarkably efficient and robust method is chosen as our new group default, replacing previous low-level DFT and partially even expensive high-level methods in most standard applications for systems with up to several hundreds of atoms.<br><br>