[Comparison of corneal power assessment methods after small incision lenticule extraction]

Objective: To compare the accuracy of different corneal curvature parameters in assessing the corneal refractive status and tracking corneal power changes after small incision lenticule extraction (SMILE). Methods: This prospective cross-sectional study tracked and recorded total corneal curvature parameters measured by different instruments before and three months after SMILE for myopia. These parameters, including total keratometry (TK) from the IOLMaster 700, total corneal refractive power (TCRP) from the Pentacam AXL, real keratometry (RK) from the CASIA 2, and corrected parameters calculated using the Haigis, Shammas, and Maloney methods, were compared with data obtained using the clinical history method (CHM). Surgically induced changes in TK, TCRP, and RK were analyzed and compared with those in spherical equivalent on the corneal plane (ΔSEco). Results: The study included 40 eyes (40 participants). After SMILE, the difference was smallest between TK [(0.08±0.38) D] and CHM values (P>0.05). However, TCRP, RK, K_Haigis, K_Shammas, and K_Maloney were significantly different from CHM data (P<0.05). The width of the 95% limits of agreement of TK (1.49 D) was narrowest, followed by that of RK (1.57 D). Pearson analysis showed that each parameter had a good correlation with CHM data. The differences between the changes in TK, TCRP and RK caused by surgery and ΔSEco were (0.03±0.39) D, (0.17±0.43) D, and (-0.19±0.46) D, respectively. The width of the 95% limits of agreement of ΔTK (1.54 D) was narrowest, and the correlation coefficient of ΔTK (0.951) was highest. Conclusion: The parameter TK of the IOLMaster 700 can provide accurate and objective corneal power evaluation after SMILE.

Clinical Value of Platelet Indices in Premature Coronary Artery Disease

Platelets play an important role in the pathophysiology of coronary artery disease. However, the clinical value of platelet indices in premature coronary heart disease remains largely unknown.Consecutive patients referred for coronary angiography were evaluated (n = 1675). Patients were stratified into premature coronary heart disease (n = 679, age < 55 for male and age < 65 for female), late-onset coronary heart disease (n = 772, age ≥ 55 for male and age ≥ 65 for female), and control (n = 224, age < 55 for male and age < 65 for female). Their clinical and laboratory parameters were collected. The relationship between platelet indices and premature coronary artery disease was analyzed.In univariate analysis, platelet indices showed no significant association with the presence of premature coronary heart disease (P > 0.05). After adjustment for traditional risk factors, mean platelet volume (0.823 [0.683-0.993], P = 0.042) and platelet-large cell ratio (0.976 [0.954-0.999], P = 0.040) were negatively correlated with the presence of premature coronary heart disease. The platelet-to-lymphocyte ratio was statistically significant among different numbers of coronary lesions (P = 0.035). In subgroup analysis, platelet-large cell ratio (1.190 [1.010-1.403], P = 0.038) was an independent risk factor of coronary restenosis after percutaneous coronary intervention.Platelet indices were associated with the prevalence, severity, and coronary restenosis after percutaneous coronary intervention suggesting their possible clinical application in premature coronary heart disease.

Evaluation of the Efficiency of TIMP-2 as a Biomarker for Acute Kidney Injury in Sepsis

The aim of this study was to evaluate the biomarker potential of TIMP-2 in septic-induced acute kidney injury (AKI). Healthy male rats (n=56, age 8-10 weeks, body weight 250-300 g) were randomized into 3 groups: controls (intact rats, n=6), sham-operated (SO, n=24), and sepsis model (cecum ligation and perforation, CLP, n=24). Thirty minutes before and 6, 12, 24, and 48 h after surgery, blood samples were collected to measure serum creatinine, blood urea nitrogen (BUN), and TIMP-2 and the kidneys were isolated for histopathological analysis and Western blotting. The key sepsis-related genes were screened through bioinformatics analysis. In 24 and 48 h after surgery, 2 rats in the SO group reached the diagnostic criteria of AKI (increased levels of serum creatinine and BUN). In the CLP group, serum creatinine in 6 h after the surgery was slightly higher than 30 min before the surgery, but this change did not meet the diagnostic criteria for AKI. In the CLP group, BUN was normal 6 h after the surgery, but increased after 12 h. In more than 50% rats of the CLP group, serum creatinine and BUN significantly increased 12 h after operation, so this can be diagnosed as AKI. In rats of the CLP group, plasma TIMP-2 was elevated 6 h after surgery and increased with time, suggesting that plasma TIMP-2 can be used as an early marker of AKI. Histological examination of the kidneys in this group revealed destruction of the renal tubular structure, swelling of renal tubular epithelium, the disappearance of brush edge and collapse of necrotic epithelial cells, etc., and the degree of damage increased with time. Immunohistochemistry showed that TIMP-2 was expressed in rats of the CLP group at all terms of the experiment. The expression of TIMP-2 and pyroptosis-related proteins (NLRP3, IL-1β, caspase-1, and GSDMD) in the CLP group was higher than in the SO group (p<0.05) and increased with time, suggesting that pyroptosis is involved in AKI. Thus, plasma TIMP-2 is sensitive indicator for the early detection of kidney injury and can be used as an early biomarker of AKI.

Enhanced robust spatial feature selection and correlation filter learning for UAV tracking

Spatial boundary effect can significantly reduce the performance of a learned discriminative correlation filter (DCF) model. A commonly used method to relieve this effect is to extract appearance features from a wider region of a target. However, this way would introduce unexpected features from background pixels and noises, which will lead to a decrease of the filter's discrimination power. To address this shortcoming, this paper proposes an innovative method called enhanced robust spatial feature selection and correlation filter Learning (EFSCF), which performs jointly sparse feature learning to handle boundary effects effectively while suppressing the influence of background pixels and noises. Unlike the ℓ₂-norm-based tracking approaches that are prone to non-Gaussian noises, the proposed method imposes the ℓ_2,1-norm on the loss term to enhance the robustness against the training outliers. To enhance the discrimination further, a jointly sparse feature selection scheme based on the ℓ_2,1 -norm is designed to regularize the filter in rows and columns simultaneously. To the best of the authors' knowledge, this has been the first work exploring the structural sparsity in rows and columns of a learned filter simultaneously. The proposed model can be efficiently solved by an alternating direction multiplier method. The proposed EFSCF is verified by experiments on four challenging unmanned aerial vehicle datasets under severe noise and appearance changes, and the results show that the proposed method can achieve better tracking performance than the state-of-the-art trackers.

[Self-reported quality of life in patients with coronary heart disease and analysis of the associated factors]

Objective: To investigate the quality of life and associated factors in patients with coronary heart disease (CHD) in China. Methods: A cross-sectional study of 25 provinces and cities in China was performed from June to September 2020. A questionnaire was used to collect the socio-demographic and clinical information of patients with CHD, while the European Five-dimensional Quality of Life Scale (EQ-5D) was used to assess the quality of life. Multiple linear regression model was performed to analyze the associated factors. Results: The median age of the 1 075 responders was 60 (52, 67) years, and 797 (74.1%) were men. The EQ-5D and EQ-VAS indices were 0.7 (0.5, 0.8) and 60.0 (40.0, 80.0). Among the five dimensions in the quality of life scale, the frequency of anxiety/depression was the highest (59.8%), while problems in self-care was the lowest (35.8%). In the multiple linear regression model, female, increasing age, obesity, comorbidity(ies), anxiety/depression, social media channels, and receiving the CABG therapy were associated with the lower EQ-5D index (all P<0.05). In addition, increasing age, obesity, comorbidity (ies), depression, anxiety and depression, social media channels, and receiving the CABG therapy were associated with lower EQ-VAS index (all P<0.05). Conclusion: Over half of the patients with CHD in China have a low quality of life, which is related to gender, age, obesity, treatment pathway, the presence or absence of comorbidity (ies), and psychological state. In addition to managing the adverse effects of traditional socio-demographic factors on the quality of life, clinical practices should pay attention to the psychological state of patients. Moreover, establishing a WeChat group for doctor-patient communication could improve the quality of life of CHD patients.

[Ultrasonographic features of retinal pigment epithelial adenoma]

Objective: To investigate the ultrasonographic features of retinal pigment epithelium (RPE) adenoma. Methods: It was a retrospective case series study. The clinical clata of 15 patients (15 eyes) with pathologically confirmed RPE adenoma after local resection of intraocular tumor was collected at Beijing Tongren Hospital, Capital Medical University from November 2013 to October 2019. The general conditions of the patients and the location, size, shape, internal echo features of the lesions in the ocular ultrasound sonogram were analyzed, and the blood flow in the lesions was checked by color Doppler flow imaging (CDFI). Results: Of all the patients included in the study, 7 were male and 8 were female. Their age ranged from 25 to 58 years, with a mean age of (45.7±10.2) years. The most common symptom was vision loss or blurred vision (11 cases). Other symptoms included dark shadows or obscuration in front of the eyes (3 cases) and no symptoms (1 case). A history of previous ocular trauma was present in one case, and the rest of the patients had no history of ocular trauma.The location of tumor growth is scattered. The ultrasonographic features were as follows: the average maximum basal diameter was (8.07±2.75) mm and the average height was (4.02±1.81) mm; the ultrasonographic features mostly demonstrated abruptly elevated dome-shaped echo (6 cases); the lesion edge was not smooth, the internal echo was medium or low, and there could be hollow features (2 cases), with no choroidal depression; and the blood flow signal could be seen in the CDFI lesion, which could lead to retinal detachment and vitreous opacification. Conclusion: The ultrasound imaging features of RPE adenomas mostly demonstrate abruptly elevated dome-shaped echo, unsmooth lesion edge, with no choroidal depression, which may provide valuable evidence for clinical diagnosis and differentiation.

Taming Self-Supervised Learning for Presentation Attack Detection: De-Folding and De-Mixing

Biometric systems are vulnerable to presentation attacks (PAs) performed using various PA instruments (PAIs). Even though there are numerous PA detection (PAD) techniques based on both deep learning and hand-crafted features, the generalization of PAD for unknown PAI is still a challenging problem. In this work, we empirically prove that the initialization of the PAD model is a crucial factor for generalization, which is rarely discussed in the community. Based on such observation, we proposed a self-supervised learning-based method, denoted as DF-DM. Specifically, DF-DM is based on a global-local view coupled with de-folding and de-mixing to derive the task-specific representation for PAD. During de-folding, the proposed technique will learn region-specific features to represent samples in a local pattern by explicitly minimizing the generative loss. While de-mixing drives detectors to obtain the instance-specific features with global information for more comprehensive representation by minimizing the interpolation-based consistency. Extensive experimental results show that the proposed method can achieve significant improvements in terms of both face and fingerprint PAD in more complicated and hybrid datasets when compared with the state-of-the-art methods. When training in CASIA-FASD and Idiap Replay-Attack, the proposed method can achieve an 18.60% equal error rate (EER) in OULU-NPU and MSU-MFSD, exceeding the baseline performance by 9.54%. The source code of the proposed technique is available at https://github.com/kongzhecn/dfdm.

Automatic Learning Rate Adaption for Memristive Deep Learning Systems

As a possible device to further enhance the performance of the hybrid complementary metal oxide semiconductor (CMOS) technology in the hardware, the memristor has attracted widespread attention in implementing efficient and compact deep learning (DL) systems. In this study, an automatic learning rate tuning method for memristive DL systems is presented. Memristive devices are utilized to adjust the adaptive learning rate in deep neural networks (DNNs). The speed of the learning rate adaptation process is fast at first and then becomes slow, which consist of the memristance or conductance adjustment process of the memristors. As a result, no manual tuning of learning rates is required in the adaptive back propagation (BP) algorithm. While cycle-to-cycle and device-to-device variations could be a significant issue in memristive DL systems, the proposed method appears robust to noisy gradients, various architectures, and different datasets. Moreover, fuzzy control methods for adaptive learning are presented for pattern recognition, such that the over-fitting issue can be well addressed. To our best knowledge, this is the first memristive DL system using an adaptive learning rate for image recognition. Another highlight of the presented memristive adaptive DL system is that quantized neural network architecture is utilized, and there is therefore a significant increase in the training efficiency, without the loss of testing accuracy.

Effect of Climate Change on the Potentially Suitable Distribution Pattern of Castanopsis hystrix Miq. in China

Climate warming poses a great threat to ecosystems worldwide, which significantly affects the geographical distribution and suitable growth area of species. Taking Castanopsis hystrix Miq. as the research object, the potentially suitable cultivation regions under present and future climatic emission scenarios in China were predicted based on the MaxEnt model with 360 effective individual distributions and eight environmental variables. The min temperature of coldest month (bio6), precipitation of driest month (bio14), and precipitation of warmest quarter (bio18) are three leading factors affecting the geographical distribution area of C. hystrix Miq. The suitable cultivation regions of C. hystrix Miq. range from 18°-34° N, 89°-122° E in central and southern China and cover an area of 261.95 × 10⁴ km². The spatial pattern of C. hystrix Miq. will migrate to the southern region of low latitudes with a decreasing suitable area when in ssp1-2.6, and to the southwestern region of low latitudes or expand to the northeast region at high latitudes in ssp5-8.5, with an increasing suitable area; no significant change on the spatial pattern in ssp2-2.4. For ssp1-2.6 or ssp2-4.5 climate scenarios, the southern region of high latitudes will be appropriate for introducing and cultivating C. hystrix Miq., and the cultivation area will increase. For ssp5-8.5, its cultivation will increase and expand to the northeast of high-latitude areas slightly.

[Evaluation of uniportal video-assisted thoracoscopic decortication in treatment of drug-resistant tuberculous empyema]

Objective: To examine the safety and efficacy of the uniportal video-assisted thoracoscopic decortication in treatment of drug-resistant tuberculosis empyema. Methods: From January 2018 to December 2020, 122 cases of tuberculous empyema treated by decortication in Department of Surgery, Wuhan Pulmonary Hospital were retrospectively analyzed, including 100 males and 22 females, aged(M(IQR)) 29.5(28.0) years (range: 13 to 70 years). According to the surgical approach and drug resistance, patients with drug-resistant tuberculosis who underwent uniportal video-assisted thoracoscopic decortication were included in group A (n=22), and those who underwent thoracotomy decortication were included in group B (n=28). Drug-sensitive patients who underwent uniportal video-assisted thoracoscopic decortication were included in group C (n=72). There was no statistical difference in the baseline data of the three groups (P>0.05). The operation, early postoperative recovery, and prognosis-related indicators were compared among three groups by Kruskal-Wallis test and χ² test by Mann-Whitney U test and Bonferroni method between groups A and B, groups A and C. Results: The intraoperative blood loss of group A, group B, and group C was 200(475) ml, 300(200) ml, and 225(300) ml, respectively. There was no significant difference in intraoperative hemorrhage (H=2.74, P=0.254) and treatment outcome (χ²=4.76, P=0.575) among the three groups. Compared with group B, the operation time of group A (302.5(187.5) minutes vs. 200.0(60.0) minutes, U=171.0, P=0.007) and postoperative pulmonary reexpansion duration (4.5(3.0) months vs. 3.0 (2.2) months, U=146.5, P=0.032) were longer, and the postoperative drainage duration (9.5(7.8) days vs. 13.0(10.0) days, U=410.0, P=0.044), and the postoperative hospitalization time (12.0(7.8) days vs. 14.5(4.8) days, U=462.2, P=0.020) were shorter. There was no significant difference in complications between group A and group B (63.6%(14/22) vs. 71.4%(20/28), χ²=0.34, P=0.558). Compared with group C, the postoperative drainage duration of group A (9.5(7.8) days vs. 7.0(4.0) days, U=543.5, P=0.031), the postoperative hospitalization time (12.0(7.8) days vs. 9.0(4.0) days, U=533.0, P=0.031) and postoperative pulmonary reexpansion duration (4.5(3.0) months vs. 3.0(2.0) months, U=961.5, P=0.001) were longer. The operation time (302.5(187.5) minutes vs. 242.5(188.8) minutes, U=670.5, P=0.278), and complications (63.6%(14/22) vs. 40.3%(29/72), χ²=3.70, P=0.054) were not different between group A and group C. Conclusions: For drug-resistant tuberculous empyema, the uniportal video-assisted thoracoscopic decortication can achieve the same good therapeutic effect as drug-sensitive tuberculous empyema, and it is as safe as thoracotomy. At the same time, it has the advantage of minimally invasive and can accelerate the early postoperative recovery of patients.

Precise Facial Landmark Detection by Reference Heatmap Transformer

Most facial landmark detection methods predict landmarks by mapping the input facial appearance features to landmark heatmaps and have achieved promising results. However, when the face image is suffering from large poses, heavy occlusions and complicated illuminations, they cannot learn discriminative feature representations and effective facial shape constraints, nor can they accurately predict the value of each element in the landmark heatmap, limiting their detection accuracy. To address this problem, we propose a novel Reference Heatmap Transformer (RHT) by introducing reference heatmap information for more precise facial landmark detection. The proposed RHT consists of a Soft Transformation Module (STM) and a Hard Transformation Module (HTM), which can cooperate with each other to encourage the accurate transformation of the reference heatmap information and facial shape constraints. Then, a Multi-Scale Feature Fusion Module (MSFFM) is proposed to fuse the transformed heatmap features and the semantic features learned from the original face images to enhance feature representations for producing more accurate target heatmaps. To the best of our knowledge, this is the first study to explore how to enhance facial landmark detection by transforming the reference heatmap information. The experimental results from challenging benchmark datasets demonstrate that our proposed method outperforms the state-of-the-art methods in the literature.

Style attention based global-local aware GAN for personalized facial caricature generation

Introduction

Caricature is an exaggerated pictorial representation of a person, which is widely used in entertainment and political media. Recently, GAN-based methods achieved automatic caricature generation through transferring caricature style and performing shape exaggeration simultaneously. However, the caricature synthesized by these methods cannot perfectly reflect the characteristics of the subject, whose shape exaggeration are not reasonable and requires facial landmarks of caricature. In addition, the existing methods always produce the bad cases in caricature style due to the simpleness of their style transfer method.

Methods

In this paper, we propose a Style Attention based Global-local Aware GAN to apply the characteristics of a subject to generate personalized caricature. To integrate the facial characteristics of a subject, we introduce a landmark-based warp controller for personalized shape exaggeration, which employs the facial landmarks as control points to warp image according to its facial features, without requirement of the facial landmarks of caricature. To fuse the facial feature with caricature style appropriately, we introduce a style-attention module, which adopts an attention mechanism, instead of the simple Adaptive Instance Normalization (AdaIN) for style transfer. To reduce the bad cases and increase the quality of generated caricatures, we propose a multi-scale discriminator to both globally and locally discriminate the synthesized and real caricature, which improves the whole structure and realistic details of the synthesized caricature.

Results

Experimental results on two publicly available datasets, the WebCaricature and the CaVINet datasets, validate the effectiveness of our proposed method and suggest that our proposed method achieves better performance than the existing methods.

Discussion

The caricatures generated by the proposed method can not only preserve the identity of input photo but also the characteristic shape exaggeration for each person, which are highly close to the real caricatures drawn by real artists. It indicates that our method can be adopted in the real application.

Universal pre-mixing dry-film stickers capable of retrofitting existing microfluidics

Integrating microfluidic mixers into lab-on-a-chip devices remains challenging yet important for numerous applications including dilutions, extractions, addition of reagents or drugs, and particle synthesis. High-efficiency mixers utilize large or intricate geometries that are difficult to manufacture and co-implement with lab-on-a-chip processes, leading to cumbersome two-chip solutions. We present a universal dry-film microfluidic mixing sticker that can retrofit pre-existing microfluidics and maintain high mixing performance over a range of Reynolds numbers and input mixing ratios. To attach our pre-mixing sticker module, remove the backing material and press the sticker onto an existing microfluidic/substrate. Our innovation centers around the multilayer use of laser-cut commercially available silicone-adhesive-coated polymer sheets as microfluidic layers to create geometrically complex, easy to assemble designs that can be adhered to a variety of surfaces, namely, existing microfluidic devices. Our approach enabled us to assemble the traditional yet difficult to manufacture "F-mixer" in minutes and conceptually extend this design to create a novel space-saving spiral F-mixer. Computational fluid dynamic simulations and experimental results confirmed that both designs maintained high performance for 0.1 < Re < 10 and disparate input mixing ratios of 1:10. We tested the integration of our system by using the pre-mixer to fluorescently tag proteins encapsulated in an existing microfluidic. When integrated with another microfluidic, our pre-mixing sticker successfully combined primary and secondary antibodies to fluorescently tag micropatterned proteins with high spatial uniformity, unlike a traditional pre-mixing "T-mixer" sticker. Given the ease of this technology, we anticipate numerous applications for point-of-care devices, microphysiological-systems-on-a-chip, and microfluidic-based biomedical research.

MIB-ANet: A novel multi-scale deep network for nasal endoscopy-based adenoid hypertrophy grading

Introduction

To develop a novel deep learning model to automatically grade adenoid hypertrophy, based on nasal endoscopy, and asses its performance with that of E.N.T. clinicians.

Methods

A total of 3,179 nasoendoscopic images, including 4-grade adenoid hypertrophy (Parikh grading standard, 2006), were collected to develop and test deep neural networks. MIB-ANet, a novel multi-scale grading network, was created for adenoid hypertrophy grading. A comparison between MIB-ANet and E.N.T. clinicians was conducted.

Results

In the SYSU-SZU-EA Dataset, the MIB-ANet achieved 0.76251 F1 score and 0.76807 accuracy, and showed the best classification performance among all of the networks. The visualized heatmaps show that MIB-ANet can detect whether adenoid contact with adjacent tissues, which was interpretable for clinical decision. MIB-ANet achieved at least 6.38% higher F1 score and 4.31% higher accuracy than the junior E.N.T. clinician, with much higher (80× faster) diagnosing speed.

Discussion

The novel multi-scale grading network MIB-ANet, designed for adenoid hypertrophy, achieved better classification performance than four classical CNNs and the junior E.N.T. clinician. Nonetheless, further studies are required to improve the accuracy of MIB-ANet.