ATARI: A Graph Convolutional Neural Network Approach for Performance Prediction in Next-Generation WLANs.

IEEE 802.11 (Wi-Fi) is one of the technologies that provides high performance with a high density of connected devices to support emerging demanding services, such as virtual and augmented reality. However, in highly dense deployments, Wi-Fi performance is severely affected by interference. This problem is even worse in new standards, such as 802.11n/ac, where new features such as Channel Bonding (CB) are introduced to increase network capacity but at the cost of using wider spectrum channels. Finding the best channel assignment in dense deployments under dynamic environments with CB is challenging, given its combinatorial nature. Therefore, the use of analytical or system models to predict Wi-Fi performance after potential changes (e.g., dynamic channel selection with CB, and the deployment of new devices) are not suitable, due to either low accuracy or high computational cost. This paper presents a novel, data-driven approach to speed up this process, using a Graph Neural Network (GNN) model that exploits the information carried in the deployment's topology and the intricate wireless interactions to predict Wi-Fi performance with high accuracy. The evaluation results show that preserving the graph structure in the learning process obtains a 64% increase versus a naive approach, and around 55% compared to other Machine Learning (ML) approaches when using all training features.

High-fat diet protects the blood-brain barrier in an Alzheimer's disease mouse model.

Type 2 diabetes (T2D) is associated with increased risk of Alzheimer's disease (AD). There is evidence for impaired blood-brain barrier (BBB) in both diseases, but its role in the interplay between them is not clear. Here, we investigated the effects of high-fat diet (HFD), a model for T2D, on the Tg2576 mouse model of AD, in regard to BBB function. We showed that HFD mice had higher weight, more insulin resistance, and higher serum HDL cholesterol levels, primarily in Tg2576 mice, which also had higher brain lipids content. In terms of behavior, Tg2576 HFD mice were less active and more anxious, but had better learning in the Morris Water Maze compared to Tg2576 on regular diet. HFD had no effect on the level of amyloid beta 1-42 in the cortex of Tg2576 mice, but increased the transcription level of insulin receptor in the hippocampus. Tg2576 mice on regular diet demonstrated more BBB disruption at 8 and 12 months accompanied by larger lateral ventricles volume in contrast to Tg2576 HFD mice, whose BBB leakage and ventricular volume were similar to wild-type (WT) mice. Our results suggest that in AD, HFD may promote better cognitive function through improvements of BBB function and of brain atrophy but not of amyloid beta levels. Lipid metabolism in the CNS and peripheral tissues and brain insulin signaling may underlie this protection.

VALIDATION OF AN ALGORITHM FOR NONMETALLIC INTRAOCULAR FOREIGN BODIES' COMPOSITION IDENTIFICATION BASED ON COMPUTED TOMOGRAPHY AND MAGNETIC RESONANCE IMAGING.

PURPOSE: To validate and evaluate the accuracy of an algorithm for the identification of nonmetallic intraocular foreign body composition based on computed tomography and magnetic resonance imaging. METHODS: An algorithm for the identification of 10 nonmetallic materials based on computed tomography and magnetic resonance imaging has been previously determined in an ex vivo porcine model. Materials were classified into 4 groups (plastic, glass, stone, and wood). The algorithm was tested by 40 ophthalmologists, which completed a questionnaire including 10 sets of computed tomography and magnetic resonance images of eyes with intraocular foreign bodies and were asked to use the algorithm to identify their compositions. Rates of exact material identification and group identification were measured. RESULTS: Exact material identification was achieved in 42.75% of the cases, and correct group identification in 65%. Using the algorithm, 6 of the materials were exactly identified by over 50% of the participants, and 7 were correctly classified according to their groups by over 75% of the materials. DISCUSSION: The algorithm was validated and was found to enable correct identification of nonmetallic intraocular foreign body composition in the majority of cases. This is the first study to report and validate a clinical tool allowing intraocular foreign body composition based on their appearance in computed tomography and magnetic resonance imaging, which was previously impossible.

Thrombin Activity and Thrombin Receptor in Rat Glioblastoma Model: Possible Markers and Targets for Intervention?

High-grade gliomas constitute a group of aggressive CNS cancers that have high morbidity and mortality rates. Despite extensive research, current therapeutic approaches enable survival beyond 2 years in rare cases only. Thrombin and its main CNS target, protease-activated receptor-1, have been implicated in tumor progression and brain edema. Our aim was to study protease-activated receptor-1 (PAR-1) protein expression and thrombin-like activity levels in both in vitro and in vivo models of glioblastoma and correlate them with the volume of the surrounding edema. We measured the presence of PAR-1 protein using fluorescence immunohistochemistry and assessed thrombin activity in various glial and non-glial cell lines and in a CNS-1 glioma rat model using a thrombin-specific fluorescent assay. Thrombin activity was found to be highly elevated in various high-grade glioma cell lines as well as in non-glial malignant cell lines. In the CNS-1 glioma model, the level of PAR-1 fluorescence in the tumor was significantly elevated compared to adjacent regions of reactive gliosis or distant brain areas. The elevated level of thrombin activity observed in the high-grade glioma positively correlated with tumor-induced brain edema. In conclusion, thrombin is secreted from glioma cells and PAR-1 may be a new biological marker for high-grade gliomas.

Magnetic resonance imaging and computed tomography for the detection and characterization of nonmetallic intraocular foreign bodies.

PURPOSE: To perform a comprehensive comparative analysis of nonmetallic intraocular foreign bodies (IOFBs) using computed tomography (CT) and magnetic resonance imaging (MRI). METHODS: An ex vivo model of porcine eyes was used to study IOFBs consisting of 10 different materials: plastic, eyeglass lens, bottle glass, windshield glass, porcelain, gravel stone, concrete, wood, thorn, and pencil graphite. The study included 30 eyes with IOFBs and 6 control eyes. Each eye was scanned by CT and MRI. Images were analyzed by three-dimensional viewing software to determine distinguishing characteristics for each material. RESULTS: Analysis of MRI and CT scans yielded distinguishing characteristics for each of the 10 materials, and this information was integrated into a clinical algorithm that enables their distinction. More materials were identified by MRI than by CT, and smaller IOFB size was associated with lower detectability. Review of CT and head-coil MRI scans by masked specialists yielded a 95% agreement rate and allowed detection of most IOFBs. CONCLUSION: Magnetic resonance imaging was superior to CT in IOFB detection. Using these modalities, a set of distinguishing characteristics was established for the identification of the 10 studied materials. We recommend MRI to be part of the evaluation of patients with a suspected IOFB, after CT to rule out metallic IOFBs.

Antibody-conjugated, dual-modal, near-infrared fluorescent iron oxide nanoparticles for antiamyloidgenic activity and specific detection of amyloid-ß fibrils.

Amyloid-ß (Aß) peptide is the main fibrillar component of plaque deposits found in brains affected by Alzheimer's disease (AD) and is related to the pathogenesis of AD. Passive anti-Aß immunotherapy has emerged as a promising approach for the therapy of AD, based on the administration of specific anti-Aß monoclonal antibodies (aAßmAbs) to delay Aß aggregation in the brain. However, the main disadvantage of this approach is the required readministration of the aAßmAbs at frequent intervals. There are only a few reports describing in vitro study for the immobilization of aAßmAbs to nanoparticles as potential targeting agents of Aß aggregates. In this article, we report the immobilization of the aAßmAb clone BAM10 to near-infrared fluorescent maghemite nanoparticles for the inhibition of Aß40 fibrillation kinetics and the specific detection of Aß40 fibrils. The BAM10-conjugated iron oxide nanoparticles were well-characterized, including their immunogold labeling and cytotoxic effect on PC-12 (pheochromocytoma cell line). Indeed, these antibody-conjugated nanoparticles significantly inhibit the Aß40 fibrillation kinetics compared with the same concentration, or even five times higher, of the free BAM10. This inhibitory effect was confirmed by different assays such as the photo-induced crosslinking of unmodified proteins combined with sodium dodecyl sulfate- polyacrylamide gel electrophoresis. A cell viability assay also confirmed that these antibody-conjugated nanoparticles significantly reduced the Aß40-induced cytotoxicity to PC-12 cells. Furthermore, the selective labeling of the Aß40 fibrils with the BAM10-conjugated near-infrared fluorescent iron oxide nanoparticles enabled specific detection of Aß40 fibrils ex vivo by both magnetic resonance imaging and fluorescence imaging. This study highlights the immobilization of the aAßmAb to dual-modal nanoparticles as a potential approach for aAßmAb delivery, eliminating the issue of readministration, and contributes to the development of multifunctional agents for diagnosis and therapy of AD.