Effectiveness of Deep Learning Models for Brain Tumor Classification and Segmentation

A brain tumor is a mass or growth of abnormal cells in the brain. In children and adults, brain tumor is considered one of the leading causes of death. There are several types of brain tumors, including benign (non-cancerous) and malignant (cancerous) tumors. Diagnosing brain tumors as early as possible is essential, as this can improve the chances of successful treatment and survival. Considering this problem, we bring forth a hybrid intelligent deep learning technique that uses several pre-trained models (Resnet50, Vgg16, Vgg19, U-Net) and their integration for computer-aided detection and localization systems in brain tumors. These pre-trained and integrated deep learning models have been used on the publicly available dataset from The Cancer Genome Atlas. The dataset consists of 120 patients. The pre-trained models have been used to classify tumor or no tumor images, while integrated models are applied to segment the tumor region correctly. We have evaluated their performance in terms of loss, accuracy, intersection over union, Jaccard distance, dice coefficient, and dice coefficient loss. From pre-trained models, the U-Net model achieves higher performance than other models by obtaining 95% accuracy. In contrast, U-Net with ResNet-50 outperforms all other models from integrated pre-trained models and correctly classified and segmented the tumor region.     

Learning from demonstration in robots: Experimental comparison of neural architectures

Robots have played an important role in the automation of computer aided manufacturing. The classical robot control implementation involves an expensive key step of model-based programming. An intuitive way to reduce this expensive exercise is to replace programming with machine learning of robot actions from demonstration where a (learner) robot learns an action by observing a demonstrator robot performing the same. To achieve this learning from demonstration (LFD) different machine learning techniques such as Artificial Neural Networks (ANN), Genetic Algorithms, Hidden Markov Models, Support Vector Machines, etc. can be used. This piece of work focuses exclusively on ANNs. Since ANNs have many standard architectural variations divided into two basic computational categories namely the recurrent networks and feed-forward networks, representative networks from each have been selected for study, i.e. Feed Forward Multilayer Perceptron (FF) network for feed-forward networks category and Elman (EL), and Nonlinear Autoregressive Exogenous Model (NARX) networks for the recurrent networks category. The main objective of this work is to identify the most suitable neural architecture for application of LFD in learning different robot actions. The sensor and actuator streams of demonstrated action are used as training data for ANN learning. Consequently, the learning capability is measured by comparing the error between demonstrator and corresponding learner streams. To achieve fairness in comparison three steps have been taken. First, Dynamic Time Warping is used to measure the error between demonstrator and learner streams, which gives resilience against translation in time. Second, comparison statistics are drawn between the best, instead of weight-equal, configurations of competing architectures so that learning capability of any architecture is not forced handicap. Third, each configuration's error is calculated as the average of ten trials of all possible learning sequences with random weight initialization so that the error value is independent of a particular sequence of learning or a particular set of initial weights. Six experiments are conducted to get a performance pattern of each architecture. In each experiment, a total of nine different robot actions were tested. Error statistics thus obtained have shown that NARX architecture is most suitable for this learning problem whereas Elman architecture has shown the worst suitability. Interestingly the computationally lesser MLP gives much lower and slightly higher error statistics compared to the computationally superior Elman and NARX neural architectures, respectively.     

A new region aware invisible robust blind watermarking approach

Multimedia Tools and Applications - The multimedia revolution has made a strong impact on our society. The explosive growth of Internet access to this digital information has generated new...     

Heat transfer enhancement in hydromagnetic dissipative flow past a moving wedge suspended by H2O-aluminum alloy nanoparticles in the presence of thermal radiation

This letter investigates the two dimensional magnetohydrodynamic Falkner Skan flow of water saturated with AA7072–H₂O and AA7075–H₂O nanoparticles over a moving wedge. Influence of viscous dissipation and thermal radiation is also under consideration. The model is reduced into nondimensional form by using defined self-similar transformations. Then, numerical study (Runge-Kutta numerical scheme) is carried out for solution purpose. The effects of pertinent flow parameters are discussed in the flow regimes by means of graphical aid. Graphical results for special cases (static wedge and when the movement of flow and wedge is in opposite direction) of the model are also elucidated graphically. It is noted that fluid velocity decreases for volumetric fraction and magnetic force favor the velocity. Significant effects of thermal radiation and nanoparticles volume fraction pointed for thermal filed and the influence of Eckert number is almost inconsequential. For more radiative fluid heat transfer coefficient decreases and nanoparticles volumetric fraction favor the local rate of heat transfer. In the presence of AA7075 nanoparticles, rate of heat transfer is quite rapid as compared to that of AA7072 nanoparticles. To validate the present computations, some present results are compared with already existing results in the literature and found better agreement between them. Finally, major points of the study ingrained in the last section of the letter.     

Toxicity and Metabolism of Layered Double Hydroxide Intercalated with Levodopa in a Parkinson’s Disease Model

Layered hydroxide nanoparticles are generally biocompatible, and less toxic than most inorganic nanoparticles, making them an acceptable alternative drug delivery system. Due to growing concern over animal welfare and the expense of in vivo experiments both the public and the government are interested to find alternatives to animal testing. The toxicity potential of zinc aluminum layered hydroxide (ZAL) nanocomposite containing anti-Parkinsonian agent may be determined using a PC 12 cell model. ZAL nanocomposite demonstrated a decreased cytotoxic effect when compared to levodopa on PC12 cells with more than 80% cell viability at 100 μg/mL compared to less than 20% cell viability in a direct levodopa exposure. Neither levodopa-loaded nanocomposite nor the un-intercalated nanocomposite disturbed the cytoskeletal structure of the neurogenic cells at their IC₅₀ concentration. Levodopa metabolite (HVA) released from the nanocomposite demonstrated the slow sustained and controlled release character of layered hydroxide nanoparticles unlike the burst uptake and release system shown with pure levodopa treatment.     

A Coordinated Navigation Strategy for Multi-Robots to Capture a Target Moving with Unknown Speed

The paper proposes an algorithm for multi-robot coordination and navigation in order to intercept a target at a long distance. For this purpose, a limit cycle based algorithm using a neural oscillator with phase differences is proposed. The state of target is unknown, under the assumption that it is stationary or in motion with constant unknown speed along a straight line. Using the proposed algorithm, a group of robots is intended to move towards the target in such a way that the robots surround it. While moving to the target, self-collision between the robots is avoided. Moreover, a collision avoidance with static obstacles as well as dynamic target is realized. The robots reach the target at a desired distance, keeping uniformly distributed angles around the target. The algorithm is further extended so that a static interception point for the target can be estimated in place of pursuing a dynamic target, which is referred to as a virtual target in this paper. In other words, the robots move towards the virtual target instead of the actual target. The robots ultimately encircle the actual target when they arrive at the virtual target. The effectiveness of the proposed method is verified through simulation results.     

Insight into the binding of copper(II) by non-toxic biodegradable material (Oryza sativa): effect of modification and interfering ions

The present studies are focused on the use of non-toxic biodegradable straw from Oryza sativa in its simple and modified forms for the binding of copper(II) ions. A relatively new “green” method was adopted for modification with urea under microwaves. The studies have been performed by using the aqueous solution of Cu(II) ions with and without the presence of Cd(II) and Pb(II) as interfering ions. FTIR analysis showed the presence of oxygen- and nitrogen-containing functional groups in simple and modified materials. The emergence of new bands and shifts in the peaks confirmed the modification. The kinetics of the process was studied using the commonly employed mathematical models. Although Elovich model seemed to fit yet coefficient of determination did not reinforce it. Pseudo-second-order model was found to explain the kinetics of the binding of metal ions by simple and modified straw. The equilibrium was studied using the non-linear approach. Based on root mean square error values, it was found that Langmuir model was the most suitable model, followed by Temkin model. Surface areas were compared for single and multi-metal systems. The effect of pH was also studied. Under the studied set of conditions, the modification of straw caused a decrease in the equilibrium time of contact and increase in the biosorption capacities. The presence of other ions decreased the capacities drastically due the competition to bind with the materials.     

Faster dynamic spatial partitioning in OpenSimulator

OpenSimulator has emerged as one of the leading tools to help researchers, developers, and practitioners working in the field of virtual worlds since it is an open-source alternative to second life, the state of the art in virtual worlds. The grid mode of OpenSimulator is highly scalable, and it places no restriction on the number of cooperating OpenSimulator instances, each of which may simulate activity in an arbitrary number of regions. However, like second life, it suffers from both over-provision and under-provision of resources due to static allocation of regions to instances and the lack of an expansion and contraction model which adjusts resource allocation according to workload. We have used OpenSimulator to implement dynamic scalability which is an integral part of our novel infrastructure presented in earlier work. This paper reports timing analysis of the basic capabilities of OpenSimulator that are used to re-locate regions to additional simulators. The focus has been on a conservative extension to OpenSimulator using existing methods. To overcome serious performance issues during reassignment of a region, we present two extended region removal methods. Comparison of timing information for both existing and extended strategies is provided on both a network of Windows systems and a cluster of Linux nodes.