Constraining the Excessive Aggregation of Non-Fullerene Acceptor Molecules Enables Organic Solar Modules with the Efficiency >16.

Translating high-performance organic solar cell (OSC) materials from spin-coating to scalable processing is imperative for advancing organic photovoltaics. For bridging the gap between laboratory research and industrialization, it is essential to understand the structural formation dynamics within the photoactive layer during printing processes. In this study, two typical printing-compatible solvents in the doctor-blading process are employed to explore the intricate mechanisms governing the thin-film formation in the state-of-the-art photovoltaic system PM6:L8-BO. Our findings highlight the synergistic influence of both the donor polymer PM6 and the solvent with a high boiling point on the structural dynamics of L8-BO within the photoactive layer, significantly influencing its morphological properties. The optimized processing strategy effectively suppresses the excessive aggregation of L8-BO during the slow drying process in doctor-blading, enhancing thin-film crystallization with preferential molecular orientation. These improvements facilitate more efficient charge transport, suppress thin-film defects and charge recombination, and finally enhance the upscaling potential. Consequently, the optimized PM6:L8-BO OSCs demonstrate power conversion efficiencies of 18.42% in small-area devices (0.064 cm2) and 16.02% in modules (11.70 cm2), respectively. Overall, this research provides valuable insights into the interplay among thin-film formation kinetics, structure dynamics, and device performance in scalable processing.

Unsupervised and Self-supervised Learning in Low-Dose Computed Tomography Denoising: Insights from Training Strategies.

In recent years, X-ray low-dose computed tomography (LDCT) has garnered widespread attention due to its significant reduction in the risk of patient radiation exposure. However, LDCT images often contain a substantial amount of noises, adversely affecting diagnostic quality. To mitigate this, a plethora of LDCT denoising methods have been proposed. Among them, deep learning (DL) approaches have emerged as the most effective, due to their robust feature extraction capabilities. Yet, the prevalent use of supervised training paradigms is often impractical due to the challenges in acquiring low-dose and normal-dose CT pairs in clinical settings. Consequently, unsupervised and self-supervised deep learning methods have been introduced for LDCT denoising, showing considerable potential for clinical applications. These methods' efficacy hinges on training strategies. Notably, there appears to be no comprehensive reviews of these strategies. Our review aims to address this gap, offering insights and guidance for researchers and practitioners. Based on training strategies, we categorize the LDCT methods into six groups: (i) cycle consistency-based, (ii) score matching-based, (iii) statistical characteristics of noise-based, (iv) similarity-based, (v) LDCT synthesis model-based, and (vi) hybrid methods. For each category, we delve into the theoretical underpinnings, training strategies, strengths, and limitations. In addition, we also summarize the open source codes of the reviewed methods. Finally, the review concludes with a discussion on open issues and future research directions.

Self-supervised deep learning for joint 3D low-dose PET/CT image denoising.

Deep learning (DL)-based denoising of low-dose positron emission tomography (LDPET) and low-dose computed tomography (LDCT) has been widely explored. However, previous methods have focused only on single modality denoising, neglecting the possibility of simultaneously denoising LDPET and LDCT using only one neural network, i.e., joint LDPET/LDCT denoising. Moreover, DL-based denoising methods generally require plenty of well-aligned LD-normal-dose (LD-ND) sample pairs, which can be difficult to obtain. To this end, we propose a self-supervised two-stage training framework named MAsk-then-Cycle (MAC), to achieve self-supervised joint LDPET/LDCT denoising. The first stage of MAC is masked autoencoder (MAE)-based pre-training and the second stage is self-supervised denoising training. Specifically, we propose a self-supervised denoising strategy named cycle self-recombination (CSR), which enables denoising without well-aligned sample pairs. Unlike other methods that treat noise as a homogeneous whole, CSR disentangles noise into signal-dependent and independent noises. This is more in line with the actual imaging process and allows for flexible recombination of noises and signals to generate new samples. These new samples contain implicit constraints that can improve the network's denoising ability. Based on these constraints, we design multiple loss functions to enable self-supervised training. Then we design a CSR-based denoising network to achieve joint 3D LDPET/LDCT denoising. Existing self-supervised methods generally lack pixel-level constraints on networks, which can easily lead to additional artifacts. Before denoising training, we perform MAE-based pre-training to indirectly impose pixel-level constraints on networks. Experiments on an LDPET/LDCT dataset demonstrate its superiority over existing methods. Our method is the first self-supervised joint LDPET/LDCT denoising method. It does not require any prior assumptions and is therefore more robust.

Dual-scale similarity-guided cycle generative adversarial network for unsupervised low-dose CT denoising.

Removing the noise in low-dose CT (LDCT) is crucial to improving the diagnostic quality. Previously, many supervised or unsupervised deep learning-based LDCT denoising algorithms have been proposed. Unsupervised LDCT denoising algorithms are more practical than supervised ones since they do not need paired samples. However, unsupervised LDCT denoising algorithms are rarely used clinically due to their unsatisfactory denoising ability. In unsupervised LDCT denoising, the lack of paired samples makes the direction of gradient descent full of uncertainty. On the contrary, paired samples used in supervised denoising allow the parameters of networks to have a clear direction of gradient descent. To bridge the gap in performance between unsupervised and supervised LDCT denoising, we propose dual-scale similarity-guided cycle generative adversarial network (DSC-GAN). DSC-GAN uses similarity-based pseudo-pairing to better accomplish unsupervised LDCT denoising. We design a Vision Transformer-based global similarity descriptor and a residual neural network-based local similarity descriptor for DSC-GAN to effectively describe the similarity between two samples. During training, pseudo-pairs, i.e., similar LDCT samples and normal-dose CT (NDCT) samples, dominate parameter updates. Thus, the training can achieve equivalent effect as training with paired samples. Experiments on two datasets demonstrate that DSC-GAN beats the state-of-the-art unsupervised algorithms and reaches a level close to supervised LDCT denoising algorithms.

Parallel Binary Image Cryptosystem Via Spiking Neural Networks Variants.

Due to the inefficiency of multiple binary images encryption, a parallel binary image encryption framework based on the typical variants of spiking neural networks, spiking neural P (SNP) systems is proposed in this paper. More specifically, the two basic units in the proposed image cryptosystem, the permutation unit and the diffusion unit, are designed through SNP systems with multiple channels and polarizations (SNP-MCP systems), and SNP systems with astrocyte-like control (SNP-ALC systems), respectively. Different from the serial computing of the traditional image permutation/diffusion unit, SNP-MCP-based permutation/SNP-ALC-based diffusion unit can realize parallel computing through the parallel use of rules inside the neurons. Theoretical analysis results confirm the high efficiency of the binary image proposed cryptosystem. Security analysis experiments demonstrate the security of the proposed cryptosystem.

Radar HRRP Target Recognition Based on Stacked Autoencoder and Extreme Learning Machine.

A novel radar high-resolution range profile (HRRP) target recognition method based on a stacked autoencoder (SAE) and extreme learning machine (ELM) is presented in this paper. As a key component of deep structure, the SAE does not only learn features by making use of data, it also obtains feature expressions at different levels of data. However, with the deep structure, it is hard to achieve good generalization performance with a fast learning speed. ELM, as a new learning algorithm for single hidden layer feedforward neural networks (SLFNs), has attracted great interest from various fields for its fast learning speed and good generalization performance. However, ELM needs more hidden nodes than conventional tuning-based learning algorithms due to the random set of input weights and hidden biases. In addition, the existing ELM methods cannot utilize the class information of targets well. To solve this problem, a regularized ELM method based on the class information of the target is proposed. In this paper, SAE and the regularized ELM are combined to make full use of their advantages and make up for each of their shortcomings. The effectiveness of the proposed method is demonstrated by experiments with measured radar HRRP data. The experimental results show that the proposed method can achieve good performance in the two aspects of real-time and accuracy, especially when only a few training samples are available.

Activatable ultrasmall gold nanorods for "off-on" fluorescence imaging-guided photothermal therapy.

Sensitive and specific fluorescence imaging-guided photothermal therapy (PTT) with high-efficiency is of essential importance and is still a challenge for nanotheranostics. To address these issues, we developed activatable ultrasmall gold nanorods (AUGNRs) to realize "off-on" switched fluorescence imaging-guided efficient PTT. Herein, the GNRs with an ultrasmall small size (∼4 nm) were set as the PTT platform due to their distinct absorption-dominant characteristics, generating an enhanced photothermal conversion efficiency. A near infrared (NIR) dye, Cy5, was conjugated to the surface of the ultrasmall GNRs for fluorescence imaging. Due to the strong localized surface plasmon resonance (LSPR), the fluorescence of Cy5 could be remarkably quenched by the GNRs and show an "off" state under normal conditions. As the AUGNRs are internalized by tumor cells, their ability of fluorescence imaging would be activated by glutathione (GSH) for the reducing action of GSH. Given the higher intracellular GSH concentration in tumor cells, a highly selective intracellular fluorescence imaging pattern was provided by the AUGNRs. As a result, the obtained AUGNRs revealed a uniformly rod-like structure with an aspect ratio of ∼4 and showed an enhanced photothermal conversion efficiency. The in vitro cellular uptake study indicated that the AUGNRs can efficiently enter the tumor cells. It has been demonstrated by in vitro Cy5 release profiles that the AUGNRs could achieve a triggered Cy5 release in response to GSH. The MTT assay and calcein AM/PI co-staining demonstrated that the cancer cells could be effectively killed when exposed to a NIR laser. Our work presents great potential for activated fluorescence imaging-guided PTT with high specificity and efficiency, as a promising method for future clinical cancer diagnostics and treatment.

Prognostic and clinicopathological value of Gli-1 expression in gastric cancer: A meta-analysis.

Glioma associated oncogene-1 (Gli-1) is considered as a strong positive activator of downstream target genes of hedgehog signal pathway in mammalians. However, its diagnostic and prognostic value in gastric cancer remains unclear and controversial. Therefore, a quantitative meta-analysis was conducted to determine the clinical value of Gli-1 in gastric cancer patients. Twelve eligible articles with 886 gastric cancer patients were included in this meta-analysis. The relationship between Gli-1 expression in gastric cancer patients and clinicopathological features and 5-year overall survival (OS) was evaluated using pooled odds ratios (ORs) and hazard ratio (HR) with 95% confidence intervals (CIs). The meta-analysis showed that the upregulated Gli-1 was associated with sample type (gastric cancer tissues) (OR 10.31, 95%CI 7.14-14.88; P = 0.000), differentiation type (OR 3.76, 95%CI 2.55-5.53; P = 0.000), depth of invasion (OR 8.17, 95%CI 3.60-18.55; P = 0.000), lymph node metastasis (OR 3.97, 95%CI 2.73-5.78; P = 0.000) and high TNM stage (OR 3.65, 95%CI 1.89-7.04; P = 0.000). Three studies including 316 patients were assessed for the correlation between Gli-1 and 5-year OS, which indicated that positive Gli-1 expression was associated with poor prognosis in gastric cancer patients (HR 2.14, 95%CI 1.35-3.40; P = 0.001). Little publication bias was identified by funnel plots and Egger's tests. The sensitivity analysis indicated that no study substantially influenced pooled OR/HR. Taken together, Gli-1 is a credible indicator for highly aggressive tumor with poor prognosis in gastric cancer patients.

Relating Doses of Contrast Agent Administered to TIC and Semi-Quantitative Parameters on DCE-MRI: Based on a Murine Breast Tumor Model.

OBJECTIVE: To explore the changes in the time-signal intensity curve(TIC) type and semi-quantitative parameters of dynamic contrast-enhanced(DCE)imaging in relation to variations in the contrast agent(CA) dosage in the Walker 256 murine breast tumor model, and to determine the appropriate parameters for the evaluation ofneoadjuvantchemotherapy(NAC)response. MATERIALS AND METHODS: Walker 256 breast tumor models were established in 21 rats, which were randomly divided into three groups of7rats each. Routine scanning and DCE-magnetic resonance imaging (MRI) of the rats were performed using a 7T MR scanner. The three groups of rats were administered different dosages of the CA0.2mmol/kg, 0.3mmol/kg, and 0.5mmol/kg, respectively; and the corresponding TICs the semi-quantitative parameters were calculated and compared among the three groups. RESULTS: The TICs were not influenced by the CA dosage and presented a washout pattern in all of the tumors evaluated and weren't influenced by the CA dose. The values of the initial enhancement percentage(Efirst), initial enhancement velocity(Vfirst), maximum signal(Smax), maximum enhancement percentage(Emax), washout percentage(Ewash), and signal enhancement ratio(SER) showed statistically significant differences among the three groups (F = 16.952, p = 0.001; F = 69.483, p<0.001; F = 54.838, p<0.001; F = 12.510, p = 0.003; F = 5.248, p = 0.031; F = 9.733, p = 0.006, respectively). However, the values of the time to peak(Tpeak), maximum enhancement velocity(Vmax), and washout velocity(Vwash)did not differ significantly among the three dosage groups (F = 0.065, p = 0.937; F = 1.505, p = 0.273; χ2 = 1.423, p = 0.319, respectively); the washout slope(Slopewash), too, was uninfluenced by the dosage(F = 1.654, p = 0.244). CONCLUSION: The CA dosage didn't affect the TIC type, Tpeak, Vmax, Vwash or Slopewash. These dose-independent parameters as well as the TIC type might be more useful for monitoring the NAC response because they allow the comparisons of the DCE data obtained using different CA dosages.