Residual-Based Multi-Stage Deep Learning Framework for Computer-Aided Alzheimer's Disease Detection.

Alzheimer's Disease (AD) poses a significant health risk globally, particularly among the elderly population. Recent studies underscore its prevalence, with over 50% of elderly Japanese facing a lifetime risk of dementia, primarily attributed to AD. As the most prevalent form of dementia, AD gradually erodes brain cells, leading to severe neurological decline. In this scenario, it is important to develop an automatic AD-detection system, and many researchers have been working to develop an AD-detection system by taking advantage of the advancement of deep learning (DL) techniques, which have shown promising results in various domains, including medical image analysis. However, existing approaches for AD detection often suffer from limited performance due to the complexities associated with training hierarchical convolutional neural networks (CNNs). In this paper, we introduce a novel multi-stage deep neural network architecture based on residual functions to address the limitations of existing AD-detection approaches. Inspired by the success of residual networks (ResNets) in image-classification tasks, our proposed system comprises five stages, each explicitly formulated to enhance feature effectiveness while maintaining model depth. Following feature extraction, a deep learning-based feature-selection module is applied to mitigate overfitting, incorporating batch normalization, dropout and fully connected layers. Subsequently, machine learning (ML)-based classification algorithms, including Support Vector Machines (SVM), Random Forest (RF) and SoftMax, are employed for classification tasks. Comprehensive evaluations conducted on three benchmark datasets, namely ADNI1: Complete 1Yr 1.5T, MIRAID and OASIS Kaggle, demonstrate the efficacy of our proposed model. Impressively, our model achieves accuracy rates of 99.47%, 99.10% and 99.70% for ADNI1: Complete 1Yr 1.5T, MIRAID and OASIS datasets, respectively, outperforming existing systems in binary class problems. Our proposed model represents a significant advancement in the AD-analysis domain.

Silicon nanoparticles: A promising approach for control of Pseudomonas aeruginosa biofilms.

Objectives: The current study aimed to investigate the control and treatment of biofilm-producing isolates of Pseudomonas aeruginosa using silicon nanoparticles (SiNPs). Materials and Methods: Biofilm-producing isolates of P. aeruginosa were recovered from various food samples and identified through fluorescent green colony formation on selective and differential media, as well as the amplification of oprI and oprL genes. Tube methods, Congo-red agar method, and scanning electron microscopy (SEM) were used to study biofilm phenotypes. The effect of SiNPs was evaluated by broth dilution assay. Results: The biofilm assay revealed that these isolates formed biofilms on glass surfaces within 72 hr of incubation. Scanning electron micrographs showed that the biofilm communities were composed of multicellular clusters of P. aeruginosa encased in matrix material. However, these isolates were unable to form biofilms on SiNPs-coated surfaces. The results showed that the planktonic isolates of P. aeruginosa were comparatively sensitive to the antibacterial properties of SiNPs, with minimum inhibitory concentration (MIC) ranging from 100 to 200 µg/ml. Contrarily, the biofilms were found to be 500 times more tolerant to the highest concentration of SiNPs (MIC of 500 µg/ml) and were more resistant. Under static conditions, the sedimentation of SiNPs resulted in their ineffectiveness. However, under shaking conditions, the biofilms were effectively dispersed and the cells were lysed. The results showed that SiNPs were effective against both the planktonic and the metabolically inactive forms of P. aeruginosa. Conclusion: This study suggests that SiNPs could be a useful tool for preventing the formation of biofilms and removing pre-existing biofilms.

Floating Fog: extending fog computing to vast waters for aerial users.

There are thousands of flights carrying millions of passengers each day, having three or more Internet-connected devices with them on average. Usually, onboard devices remain idle for most of the journey (which can be of several hours), therefore, we can tap on their underutilized potential. Although these devices are generally becoming more and more resourceful, for complex services (such as related to machine learning, augmented/virtual reality, smart healthcare, and so on) those devices do not suffice standalone. This makes a case for multi-device resource aggregation such as through femto-cloud. As our first contribution, we present the utility of femto-cloud for aerial users. But for that sake, a reliable and faster Internet is required (to access online services or cloud resources), which is currently not the case with satellite-based Internet. That is the second challenge we try to address in our paper, by presenting an adaptive beamforming-based solution for aerial Internet provisioning. However, on average, most of the flight path is above waters. Given that, we propose that beamforming transceivers can be docked on stationery ships deployed in the vast waters (such as the ocean). Nevertheless, certain services would be delay-sensitive, and accessing their on-ground servers or cloud may not be feasible (in terms of delay). Similarly, certain complex services may require resources in addition to the flight-local femto-cloud. That is the third challenge we try to tackle in this paper, by proposing that the traditional fog computing (which is a cloud-like but localized pool of resources) can also be extended to the waters on the ships harboring beamforming transceivers. We name it Floating Fog. In addition to that, Floating Fog will enable several new services such as live black-box. We also present a cost and bandwidth analysis to highlight the potentials of Floating Fog. Lastly, we identify some challenges to tackle the successful deployment of Floating Fog.