RESUMO
Background: The blood volume of intraparenchymal vessels is reported to be increased in smokers. However, the blood volume can be affected by many confounders besides tobacco exposure. This study aimed to investigate the association between cigarette smoking and pulmonary blood volume after adjusting the related factors in a large cohort of Chinese males. Methods: In this retrospective study, male participants admitted to the First Affiliated Hospital of Xi'an Jiaotong University for annual health assessment between February 2017 and February 2018 were enrolled. All subjects underwent non-contrast chest computed tomography (CT) scans, and 152 subjects underwent a review CT scan 2-3 years later. A three-dimensional approach was employed to segment the lung and intrapulmonary vessels and quantitative CT (QCT) measurements, including lung volume (LV), intrapulmonary vessel volume (IPVV), low-attenuation area <-950 Hounsfield unit (LAA-950 and LAA-950%), and mean lung density (MLD). Linear regression was used to estimate the association between IPVV and the smoking index (SI). A paired t-test was used to compare the QCT parameters between the initial and follow-up CT scans. Results: A total of 656 male participants were enrolled and classified into three subgroups: non-smokers (n=311), current smokers (n=267), and former smokers (n=78). The IPVV of current smokers (134.62±23.96 vs. 120.76±25.52 mL) and former smokers (130.79±25.13 vs. 120.76±25.52 mL) were significantly larger than that of non-smokers (P<0.05). A higher SI was associated with greater IPVV [non-standardized coefficient: 0.167, 95% confidence interval (CI): 0.086-0.248]. For current smokers, the IPVV of the follow-up scan significantly increased compared to its baseline scan (135.49±28.60 vs. 129.73±29.75 mL, t=-2.326, P=0.02), but for the non-smokers and former smokers, the IPVV of the follow-up scan did not increase or decrease compared to the baseline scan (P>0.05). Conclusions: Pulmonary vascular volumes detectable on non-contrast CT are associated with cigarette exposure, and smoking cessation may prevent pulmonary vasculature remodeling.
RESUMO
In this article, two different types of spacer cations, 1,4-butanediamonium (BDA2+) and 2-phenylethylammonium (PEA+) are co-used to prepare the perovskite precursor solutions with the formula of (BDA)1- a (PEA2) a MA4Pb5X16. By simply mixing the two spacer cations, the self-assembled polycrystalline films of (BDA)0.8(PEA2)0.2MA4Pb5X16 are obtained, and BDA2+ is located in the crystal grains and PEA+ is distributed on the surface. The films display a small exciton binding energy, uniformly distributed quantum wells and improved carrier transport. Besides, utilizing mixed spacer cations also induces better crystallinity and vertical orientation of 2D perovskite (BDA)0.8(PEA2)0.2MA4Pb5X16 films. Thus, a power conversion efficiency (PCE) of 17.21% is achieved in the optimized perovskite solar cells with the device structure of ITO/PEDOT:PSS/Perovskite/PCBM/BCP/Ag. In addition, the complementary humidity and thermal stability are obtained, which are ascribed to the enhanced interlayer interaction by BDA2+ and improved moisture resistance by the hydrophobic group of PEA+. The encapsulated devices are retained over 95% or 75% of the initial efficiency after storing 500 h in ambient air under 40 ± 5% relative humidity or 100 h in nitrogen at 60 °C.
RESUMO
The performance of person re-identification (Re-ID) has been seriously effected by the large cross-view appearance variations caused by mutual occlusions and background clutters. Hence learning a feature representation that can adaptively emphasize the foreground persons becomes very critical to solve the person Re-ID problem. In this paper, we propose a simple yet effective foreground attentive neural network (FANN) to learn a discriminative feature representation for person Re-ID, which can adaptively enhance the positive side of foreground and weaken the negative side of background. Specifically, a novel foreground attentive subnetwork is designed to drive the network's attention, in which a decoder network is used to reconstruct the binary mask by using a novel local regression loss function, and an encoder network is regularized by the decoder network to focus its attention on the foreground persons. The resulting feature maps of encoder network are further fed into the body part subnetwork and feature fusion subnetwork to learn discriminative features. Besides, a novel symmetric triplet loss function is introduced to supervise feature learning, in which the intra-class distance is minimized and the inter-class distance is maximized in each triplet unit, simultaneously. Training our FANN in a multi-task learning framework, a discriminative feature representation can be learned to find out the matched reference to each probe among various candidates in the gallery. Extensive experimental results on several public benchmark datasets are evaluated, which have shown clear improvements of our method over the state-of-the-art approaches.