Accurately detecting defects while reconstructing a high-quality normal background in surface defect detection using unsupervised methods remains a significant challenge. This study proposes an unsupervised method that effectively addresses this challenge by achieving both accurate defect detection and a high-quality normal background reconstruction without noise. We propose an adaptive weighted structural similarity (AW-SSIM) loss for focused feature learning. AW-SSIM improves structural similarity (SSIM) loss by assigning different weights to its sub-functions of luminance, contrast, and structure based on their relative importance for a specific training sample. Moreover, it dynamically adjusts the Gaussian window\'s standard deviation (σ) during loss calculation to balance noise reduction and detail preservation. An artificial defect generation algorithm (ADGA) is proposed to generate an artificial defect closely resembling real ones. We use a two-stage training strategy. In the first stage, the model trains only on normal samples using AW-SSIM loss, allowing it to learn robust representations of normal features. In the second stage of training, the weights obtained from the first stage are used to train the model on both normal and artificially defective training samples. Additionally, the second stage employs a combined learned Perceptual Image Patch Similarity (LPIPS) and AW-SSIM loss. The combined loss helps the model in achieving high-quality normal background reconstruction while maintaining accurate defect detection. Extensive experimental results demonstrate that our proposed method achieves a state-of-the-art defect detection accuracy. The proposed method achieved an average area under the receiver operating characteristic curve (AuROC) of 97.69% on six samples from the MVTec anomaly detection dataset.

BACKGROUND: Image registration is a challenging problem in many clinical tasks, but deep learning has made significant progress in this area over the past few years. Real-time and robust registration has been made possible by supervised transformation estimation. However, the quality of registrations using this framework depends on the quality of ground truth labels such as displacement field.
OBJECTIVE: To propose a simple and reliable method for registering medical images based on image structure similarity in a completely unsupervised manner.
METHODS: We proposed a deep cascade unsupervised deformable registration approach to align images without reliable clinical data labels. Our basic network was composed of a displacement estimation module (ResUnet) and a deformation module (spatial transformer layers). We adopted l 2 $l_2$ -norm to regularize the deformation field instead of the traditional l 1 $l_1$ -norm regularization. Additionally, we utilized structural similarity (ssim) estimation during the training stage to enhance the structural consistency between the deformed images and the reference images.
RESULTS: Experiments results indicated that by incorporating ssim loss, our cascaded methods not only achieved higher dice score of 0.9873, ssim score of 0.9559, normalized cross-correlation (NCC) score of 0.9950, and lower relative sum of squared difference (SSD) error of 0.0313 on CT images, but also outperformed the comparative methods on ultrasound dataset. The statistical t $t$ -test results also proved that these improvements of our method have statistical significance.
CONCLUSIONS: In this study, the promising results based on diverse evaluation metrics have demonstrated that our model is simple and effective in deformable image registration (DIR). The generalization ability of the model was also verified through experiments on liver CT images and cardiac ultrasound images.

Infrared (IR) small target detection, especially in a complex background, continues to present challenges in the low false alarm rate and high robustness. In this paper, a background subtraction local contrast measure (BSLCM) and Gaussian structural similarity (GSS) integrated structural model is proposed to detect IR small target. In the proposed model, BSLCM is used to suppress the complex background and enhance the target. GSS calculation is conducted to eliminate the high-brightened background residual and noise further. To evaluate the performance of the proposed method, real IR sequences and seven state-of-the-art (SOTA) methods were adopted. The results demonstrated that the BSLCM can suppress all types of strong background clutter and enhance the true target effectively.

Delay tolerant networks (DTNs), are characterized by their difficulty in establishing end-to-end paths and and large message propagation delays. To control network overhead costs, reduce message delays, and improve delivery rates in DTNs, it is essential to not only delete messages that have reached their destination but also to more precisely determine appropriate relay nodes. Based on the above goals, this paper constructs a multi-copy relay node selection router algorithm based on Q-lambda reinforcement learning with reference to the idea of community division (QLCR). In community division, if a node has the highestdegree, it is considered the core node, and nodes with similar interests and structural properties are divided into a community. Node degree refers to the number of nodes associated with the node, indicating its importance in the network. Structural similarity determines the distance between nodes. The selection of relay nodes considers node degree, interests, and structural similarity. The Q-lambda reinforcement learning algorithm enables each node to learn from the entire network, setting corresponding reward values based on encountered nodes meeting the specified conditions. Through iterative processes, the node with the most cumulative reward value is chosen as the final relay node. Experimental results demonstrate that the proposed algorithm achieves a high delivery rate while maintaining low network overhead and delay.

A large number of studies have shown that circular RNA (circRNA) affects biological processes by competitively binding miRNA, providing a new perspective for the diagnosis, and treatment of human diseases. Therefore, exploring the potential circRNA-miRNA interactions (CMIs) is an important and urgent task at present. Although some computational methods have been tried, their performance is limited by the incompleteness of feature extraction in sparse networks and the low computational efficiency of lengthy data.
In this paper, we proposed JSNDCMI, which combines the multi-structure feature extraction framework and Denoising Autoencoder (DAE) to meet the challenge of CMI prediction in sparse networks. In detail, JSNDCMI integrates functional similarity and local topological structure similarity in the CMI network through the multi-structure feature extraction framework, then forces the neural network to learn the robust representation of features through DAE and finally uses the Gradient Boosting Decision Tree classifier to predict the potential CMIs. JSNDCMI produces the best performance in the 5-fold cross-validation of all data sets. In the case study, seven of the top 10 CMIs with the highest score were verified in PubMed.
The data and source code can be found at https://github.com/1axin/JSNDCMI.

UNASSIGNED: The fusion of PET metabolic images and CT anatomical images can simultaneously display the metabolic activity and anatomical position, which plays an indispensable role in the staging diagnosis and accurate positioning of lung cancer.
UNASSIGNED: In order to improve the information of PET-CT fusion image, this article proposes a PET-CT fusion method via Siamese Pyramid Fusion Network (SPFN). In this method, feature pyramid transformation is introduced to the siamese convolution neural network to extract multi-scale information of the image. In the design of the objective function, this article considers the nature of image fusion problem, utilizes the image structure similarity as the objective function and introduces L1 regularization to improve the quality of the image.
UNASSIGNED: The effectiveness of the proposed method is verified by more than 700 pairs of PET-CT images and elaborate experimental design. The visual fidelity after fusion reaches 0.350, the information entropy reaches 0.076.
UNASSIGNED: The quantitative and qualitative results proved that the proposed PET-CT fusion method has some advantages. In addition, the results show that PET-CT fusion image can improve the ability of staging diagnosis compared with single modal image.

Bacterial infection is one of the most important factors affecting the human life span. Elderly people are more harmed by bacterial infections due to their deficits in immunity. Because of the lack of new antibiotics in recent years, bacterial resistance has increasingly become a serious problem globally. In this study, an antibacterial compound predictor was constructed using the support vector machines and random forest methods and the data of the active and inactive antibacterial compounds from the ChEMBL database. The results showed that both models have excellent prediction performance (mean accuracy >0.9 and mean AUC >0.9 for the two models). We used the predictor to screen potential antibacterial compounds from FDA-approved drugs in the DrugBank database. The screening results showed that 1087 small-molecule drugs have potential antibacterial activity and 154 of them are FDA-approved antibacterial drugs, which accounts for 76.2% of the approved antibacterial drugs collected in this study. Through molecular fingerprint similarity analysis and common substructure analysis, we screened 8 predicted antibacterial small-molecule compounds with novel structures compared with known antibacterial drugs, and 5 of them are widely used in the treatment of various tumors. This study provides a new insight for predicting antibacterial compounds by using approved drugs, the predicted compounds might be used to treat bacterial infections and extend lifespan.

Protein design has become a powerful method to expand the number of natural proteins and design customized proteins according to demands. Domain-based protein design spares the need to create novel elements from scratch, which makes it a more efficient strategy than scratch-based protein design in designing multi-domain proteins, protein complexes and biomaterials. As the surface shape plays a central role in domain-domain and protein-protein interactions, a global map of the surface shapes of all domains should be very beneficial for domain-based protein design. Therefore, in this study, we characterized the surface shapes of protein domains, collected from CATH and SCOP databases, with their 3D-Zernike descriptors (3DZDs). Then similarities of domain shape features were identified, and all domains were classified accordingly. The preferences of the combinations of domains between different clusters were analyzed in natural proteins from the Protein Data Bank. A user-friendly website, termed CPD3DS, was also developed for storage, retrieval, analyses and visualization of our results. This work not only provides an overall view of protein domain shapes by showing their variety and similarities, but also opens up a new avenue to understand the properties of protein structural domains, and design principles of protein architectures.

A two-step emulsification prior to complex coacervation was employed to develop a co-encapsulation technology of hydrophilic and hydrophobic components for nutrition enhancement. Processing parameters of mononuclear ellipse-like microcapsules using gelatin and sodium carboxymethyl cellulose as wall materials were evaluated. The particle size and morphology of microcapsules and the encapsulation efficiency of L-ascorbic acid were significantly affected by the water-oil phase ratio and total biopolymer concentration. The L-ascorbic acid and quercetin co-encapsulated microcapsules with an average size of 65.26 µm showed good physical and chemical stability. The encapsulation efficiencies of L-ascorbic acid and quercetin were 69.91% and 88.21%, respectively. To predict the potential of functional lipids as hydrophobic carriers, microcapsules using soybean oil, olive oil, fish oil, and conjugated linoleic acid as interlayer oils were developed. The encapsulation efficiencies of hydrophobic compounds carried by different oils were similarly high (88.21-93.08%), whereas, hydrophilic ones carried by conjugated linoleic acid had the lowest encapsulation efficiency (32.54%). The interface tension results indicated that the interfacial stability was impaired by a competitive relation between conjugated linoleic acid and hydrophobic emulsifier at the interface, due to their structural similarity. These results provided the guidance for improving the quality of interlayer oils from microcapsules.

Convolutional neural networks have recently been used for multi-focus image fusion. However, some existing methods have resorted to adding Gaussian blur to focused images, to simulate defocus, thereby generating data (with ground-truth) for supervised learning. Moreover, they classify pixels as \'focused\' or \'defocused\', and use the classified results to construct the fusion weight maps. This then necessitates a series of post-processing steps. In this paper, we present an end-to-end learning approach for directly predicting the fully focused output image from multi-focus input image pairs. The suggested approach uses a CNN architecture trained to perform fusion, without the need for ground truth fused images. The CNN exploits the image structural similarity (SSIM) to calculate the loss, a metric that is widely accepted for fused image quality evaluation. What is more, we also use the standard deviation of a local window of the image to automatically estimate the importance of the source images in the final fused image when designing the loss function. Our network can accept images of variable sizes and hence, we are able to utilize real benchmark datasets, instead of simulated ones, to train our network. The model is a feed-forward, fully convolutional neural network that can process images of variable sizes during test time. Extensive evaluation on benchmark datasets show that our method outperforms, or is comparable with, existing state-of-the-art techniques on both objective and subjective benchmarks.