The neuroprotective properties of bis(7)-tacrine, a novel dimeric acetylcholinesterase (AChE) inhibitor, on glutamate-induced excitotoxicity were investigated in primary cultured cerebellar granule ...neurons (CGNs). Exposure of CGNs to 75 mum glutamate resulted in neuronal apoptosis as demonstrated by Hoechst staining, TUNEL, and DNA fragmentation assays. The bis(7)-tacrine treatment (0.01-1 mum) on CGNs markedly reduced glutamate-induced apoptosis in dose- and time-dependent manners. However, donepezil and other AChE inhibitors, even at concentrations of inhibiting AChE to the similar extents as 1 mum bis(7)-tacrine, failed to prevent glutamate-induced excitotoxicity in CGNs; moreover, both atropine and dihydro-beta-erythroidine, the cholinoreceptor antagonists, did not affect the anti-apoptotic properties of bis(7)-tacrine, suggesting that the neuroprotection of bis(7)-tacrine appears to be independent of inhibiting AChE and cholinergic transmission. In addition, ERK1/2 and p38 pathways, downstream signals of N-methyl-d-aspartate (NMDA) receptors, were rapidly activated after the exposure of glutamate to CGNs. Bis(7)-tacrine inhibited the apoptosis and the activation of these two signals with the same efficacy as the coapplication of PD98059 and SB203580. Furthermore, using fluorescence Ca(2+) imaging, patch clamp, and receptor-ligand binding techniques, bis(7)-tacrine was found effectively to buffer the intracellular Ca(2+) increase triggered by glutamate, to reduce NMDA-activated currents and to compete with (3)HMK-801 with an IC(50) value of 0.763 mum in rat cerebellar cortex membranes. These findings strongly suggest that bis(7)-tacrine prevents glutamate-induced neuronal apoptosis through directly blocking NMDA receptors at the MK-801-binding site, which offers a new and clinically significant modality as to how the agent exerts neuroprotective effects.
This paper introduces a convolutional neural network (CNN) semantic re-ranking system to enhance the performance of sketch-based image retrieval (SBIR). Distinguished from the existing approaches, ...the proposed system can leverage category information brought by CNNs to support effective similarity measurement between the images. To achieve effective classification of query sketches and high-quality initial retrieval results, one CNN model is trained for classification of sketches, another for that of natural images. Through training dual CNN models, the semantic information of both the sketches and natural images is captured by deep learning. In order to measure the category similarity between images, a category similarity measurement method is proposed. Category information is then used for re-ranking. Re-ranking operation first infers the retrieval category of the query sketch and then uses the category similarity measurement to measure the category similarity between the query sketch and each initial retrieval result. Finally, the initial retrieval results are re-ranked. The experiments on different types of SBIR datasets demonstrate the effectiveness of the proposed re-ranking method. Comparisons with other re-ranking algorithms are also given to show the proposed method's superiority. Further, compared to the baseline systems, the proposed re-ranking approach achieves significantly higher precision in the top ten different SBIR methods and datasets.
Antibiotic contamination attracts growing concerns because of their deleterious effects on the ecosystem and human health. In this study, 43 antibiotics in wastewater from a variety of sources and ...water of the Yangtze River in Chongqing City in western China were measured. Thirty compounds were detected, and their concentrations were highest in leachates from the municipal solid waste treatment facilities (landfills and incineration plants) with total concentrations of 3584–57,106 ng/L. The total concentrations in influents of municipal and industrial wastewater treatment plants (WWTPs) were comparable (401–7994 ng/L versus 640–8945 ng/L). The concentrations in raw sewage from swine farms (with a total of 10,219–39,195 ng/L) and poultry farms (1419–36,027 ng/L) were noticeably higher than those from other farms (54.0–5516 ng/L). Fluoroquinolones were the dominant antibiotics contributing over 50% in all the sources, and sulfonamides and imidazole fungicides contributed 3.2–34%, whereas tetracyclines and macrolides had minor contributions. The overall antibiotic removal rates were highest in solid waste treatment facilities (88% on average), comparable between municipal and industrial WWTPs (61%), and lowest in animal farms (39%). The mass loads to the investigated municipal WWTPs via influent wastewater ranged from 7.80 to 1531 kg/year (53.2–2482 μg/day per capital). The influent mass loads to the industrial WWTPs and farms were 3.7–50 kg/year and 0.9–5437 g/year, respectively. We estimated that the mass inventories of antibiotics from these sources to the environment via effluent discharges were approximately 2044 kg for municipal WWTPs, 61 kg for industrial WWTPs, and 34 kg for animal farms in the whole city. Antibiotic concentrations in the Yangtze River water were substantially low (< 492 ng/L, with a mean of 57.8 ng/L) suggesting dissipation during the movement.
VTE has emerged as a major public health problem. However, data on VTE burden in China are seldom reported.
This study collected data on patients with a principal diagnosis of VTE, pulmonary embolism ...(PE), or DVT by using the International Classification of Diseases, 10th Revision, from 90 hospitals across China. The trends in hospitalization rates, mortality, length of stay (LOS), and comorbidities from 2007 to 2016 were analyzed.
In total, 105,723 patients with VTE were identified. For patients with VTE, the age- and sex-adjusted hospitalization rate increased from 3.2 to 17.5 per 100,000 population, and in-hospital mortality decreased from 4.7% to 2.1% (P < .001). The mean LOS declined from 14 to 11 days (P < .001). In addition, the data in 2016 showed that the hospitalization rate of VTE was higher in elderly male patients (male patients vs female patients, 155.3 vs 125.4 per 100,000 population in patients aged ≥ 85 years; P < .001) and in northern China (north vs south, 18.4 vs 13.4 per 100,000 population; P < .001). Higher mortality rates were found in patients with cancer and Charlson Comorbidity Index scores > 2. Similar trends were also observed in patients with PE and those with DVT. The hospitalization rate in China was much lower than that of the United States or selected sites in Canada and Europe, the LOS was much longer, and the in-hospital mortality rates were similar.
The hospitalization rates of VTE increased steadily, and the mortality declined. This study provides important information on the disease burden of VTE in China.
Recently, the efficient deployment and acceleration of powerful vision transformers (ViTs) on resource-limited edge devices for providing multimedia services have become attractive tasks. Although ...early exiting is a feasible solution for accelerating inference, most works focus on convolutional neural networks (CNNs) and transformer models in natural language processing (NLP). Moreover, the direct application of early exiting methods to ViTs may result in substantial performance degradation. To tackle this challenge, we systematically investigate the efficacy of early exiting in ViTs and point out that the insufficient feature representations in shallow internal classifiers and the limited ability to capture target semantic information in deep internal classifiers restrict the performance of these methods. We then propose an early exiting framework for general ViTs termed LGViT, which incorporates heterogeneous exiting heads, namely, local perception head and global aggregation head, to achieve an efficiency-accuracy trade-off. In particular, we develop a novel two-stage training scheme, including end-to-end training and self-distillation with the backbone frozen to generate early exiting ViTs, which facilitates the fusion of global and local information extracted by the two types of heads. We conduct extensive experiments using three popular ViT backbones on three vision datasets. Results demonstrate that our LGViT can achieve competitive performance with approximately 1.8 × speed-up.
Weakly supervised object localization (WSOL) is one of the most popular and challenging tasks in computer vision. This task is to localize the objects in the images given only the image-level ...supervision. Recently, dividing WSOL into two parts (class-agnostic object localization and object classification) has become the state-of-the-art pipeline for this task. However, existing solutions under this pipeline usually suffer from the following drawbacks: 1) they are not flexible since they can only localize one object for each image due to the adopted single-class regression (SCR) for localization; 2) the generated pseudo bounding boxes may be noisy, but the negative impact of such noise is not well addressed. To remedy these drawbacks, we first propose to replace SCR with a binary-class detector (BCD) for localizing multiple objects, where the detector is trained by discriminating the foreground and background. Then we design a weighted entropy (WE) loss using the unlabeled data to reduce the negative impact of noisy bounding boxes. Extensive experiments on the popular CUB-200-2011 and ImageNet-1K datasets demonstrate the effectiveness of our method.
Weakly supervised object localization (WSOL) is one of the most popular and challenging tasks in computer vision. This task is to localize the objects in the images given only the image-level ...supervision. Recently, dividing WSOL into two parts (class-agnostic object localization and object classification) has become the state-of-the-art pipeline for this task. However, existing solutions under this pipeline usually suffer from the following drawbacks: 1) they are not flexible since they can only localize one object for each image due to the adopted single-class regression (SCR) for localization; 2) the generated pseudo bounding boxes may be noisy, but the negative impact of such noise is not well addressed. To remedy these drawbacks, we first propose to replace SCR with a binary-class detector (BCD) for localizing multiple objects, where the detector is trained by discriminating the foreground and background. Then we design a weighted entropy (WE) loss using the unlabeled data to reduce the negative impact of noisy bounding boxes. Extensive experiments on the popular CUB-200-2011 and ImageNet-1K datasets demonstrate the effectiveness of our method.
Recently, the efficient deployment and acceleration of powerful vision transformers (ViTs) on resource-limited edge devices for providing multimedia services have become attractive tasks. Although ...early exiting is a feasible solution for accelerating inference, most works focus on convolutional neural networks (CNNs) and transformer models in natural language processing (NLP).Moreover, the direct application of early exiting methods to ViTs may result in substantial performance degradation. To tackle this challenge, we systematically investigate the efficacy of early exiting in ViTs and point out that the insufficient feature representations in shallow internal classifiers and the limited ability to capture target semantic information in deep internal classifiers restrict the performance of these methods. We then propose an early exiting framework for general ViTs termed LGViT, which incorporates heterogeneous exiting heads, namely, local perception head and global aggregation head, to achieve an efficiency-accuracy trade-off. In particular, we develop a novel two-stage training scheme, including end-to-end training and self-distillation with the backbone frozen to generate early exiting ViTs, which facilitates the fusion of global and local information extracted by the two types of heads. We conduct extensive experiments using three popular ViT backbones on three vision datasets. Results demonstrate that our LGViT can achieve competitive performance with approximately 1.8 \(\times\) speed-up.
