Improving Hybrid CTC/Attention Architecture for Agglutinative Language Speech Recognition.

Unlike the traditional model, the end-to-end (E2E) ASR model does not require speech information such as a pronunciation dictionary, and its system is built through a single neural network and obtains performance comparable to that of traditional methods. However, the model requires massive amounts of training data. Recently, hybrid CTC/attention ASR systems have become more popular and have achieved good performance even under low-resource conditions, but they are rarely used in Central Asian languages such as Turkish and Uzbek. We extend the dataset by adding noise to the original audio and using speed perturbation. To develop the performance of an E2E agglutinative language speech recognition system, we propose a new feature extractor, MSPC, which uses different sizes of convolution kernels to extract and fuse features of different scales. The experimental results show that this structure is superior to VGGnet. In addition to this, the attention module is improved. By using the CTC objective function in training and the BERT model to initialize the language model in the decoding stage, the proposed method accelerates the convergence of the model and improves the accuracy of speech recognition. Compared with the baseline model, the character error rate (CER) and word error rate (WER) on the LibriSpeech test-other dataset increases by 2.42% and 2.96%, respectively. We apply the model structure to the Common Voice-Turkish (35 h) and Uzbek (78 h) datasets, and the WER is reduced by 7.07% and 7.08%, respectively. The results show that our method is close to the advanced E2E systems.

Real-Time Stylized Humanoid Behavior Control through Interaction and Synchronization.

Restricted by the diversity and complexity of human behaviors, simulating a character to achieve human-level perception and motion control is still an active as well as a challenging area. We present a style-based teleoperation framework with the help of human perceptions and analyses to understand the tasks being handled and the unknown environment to control the character. In this framework, the motion optimization and body controller with center-of-mass and root virtual control (CR-VC) method are designed to achieve motion synchronization and style mimicking while maintaining the balance of the character. The motion optimization synthesizes the human high-level style features with the balance strategy to create a feasible, stylized, and stable pose for the character. The CR-VC method including the model-based torque compensation synchronizes the motion rhythm of the human and character. Without any inverse dynamics knowledge or offline preprocessing, our framework is generalized to various scenarios and robust to human behavior changes in real-time. We demonstrate the effectiveness of this framework through the teleoperation experiments with different tasks, motion styles, and operators. This study is a step toward building a human-robot interaction that uses humans to help characters understand and achieve the tasks.

Large Foundation Model for Cancer Segmentation.

Recently, large language models such as ChatGPT have made huge strides in understanding and generating human-like text and have demonstrated considerable success in natural language processing. These foundation models also perform well in computer vision. However, there is a growing need to use these technologies for specific medical tasks, especially for identifying cancer in images. This paper looks at how these foundation models, such as the segment anything model, could be used for cancer segmentation, discussing the potential benefits and challenges of applying large foundation models to help with cancer diagnoses.

UKSSL: Underlying Knowledge Based Semi-Supervised Learning for Medical Image Classification.

Goal: Deep learning techniques have made significant progress in medical image analysis. However, obtaining ground truth labels for unlabeled medical images is challenging as they often outnumber labeled images. Thus, training a high-performance model with limited labeled data has become a crucial challenge. Methods: This study introduces an underlying knowledge-based semi-supervised framework called UKSSL, consisting of two components: MedCLR extracts feature representations from the unlabeled dataset; UKMLP utilizes the representation and fine-tunes it with the limited labeled dataset to classify the medical images. Results: UKSSL evaluates on the LC25000 and BCCD datasets, using only 50% labeled data. It gets precision, recall, F1-score, and accuracy of 98.9% on LC25000 and 94.3%, 94.5%, 94.3%, and 94.1% on BCCD, respectively. These results outperform other supervised-learning methods using 100% labeled data. Conclusions: The UKSSL can efficiently extract underlying knowledge from the unlabeled dataset and perform better using limited labeled medical images.

Exploring simple triplet representation learning.

Fully supervised learning methods necessitate a substantial volume of labelled training instances, a process that is typically both labour-intensive and costly. In the realm of medical image analysis, this issue is further amplified, as annotated medical images are considerably more scarce than their unlabelled counterparts. Consequently, leveraging unlabelled images to extract meaningful underlying knowledge presents a formidable challenge in medical image analysis. This paper introduces a simple triple-view unsupervised representation learning model (SimTrip) combined with a triple-view architecture and loss function, aiming to learn meaningful inherent knowledge efficiently from unlabelled data with small batch size. With the meaningful representation extracted from unlabelled data, our model demonstrates exemplary performance across two medical image datasets. It achieves this using only partial labels and outperforms other state-of-the-art methods. The method we present herein offers a novel paradigm for unsupervised representation learning, establishing a baseline that is poised to inspire the development of more intricate SimTrip-based methods across a spectrum of computer vision applications. Code and user guide are released at https://github.com/JerryRollingUp/SimTripSystem, the system also runs at http://43.131.9.159:5000/.

Mechanistic insights into the role of USP14 in adipose tissue macrophage recruitment and insulin resistance in obesity.

Diet-induced obesity can cause metabolic syndromes. The critical link in disease progression is adipose tissue macrophage (ATM) recruitment, which drives low-level inflammation, triggering adipocyte dysfunction. It is unclear whether ubiquitin-specific proteinase 14 (USP14) affects metabolic disorders by mediating adipose tissue inflammation. In the present study, we showed that USP14 is highly expressed in ATMs of obese human patients and diet-induced obese mice. Mouse USP14 overexpression aggravated obesity-related insulin resistance by increasing the levels of pro-inflammatory ATMs, leading to adipose tissue inflammation, excessive lipid accumulation, and hepatic steatosis. In contrast, USP14 knockdown in adipose tissues alleviated the phenotypes induced by a high-fat diet. Co-culture experiments showed that USP14 deficiency in macrophages led to decreased adipocyte lipid deposition and enhanced insulin sensitivity, suggesting that USP14 plays an important role in ATMs. Mechanistically, USP14 interacted with TNF receptor-associated 6, preventing K48-linked ubiquitination as well as proteasome degradation, leading to increased pro-inflammatory polarization of macrophages. In contrast, the pharmacological inhibition of USP14 significantly ameliorated diet-induced hyperlipidemia and insulin resistance in mice. Our results demonstrated that macrophage USP14 restriction constitutes a key constraint on the pro-inflammatory M1 phenotype, thereby inhibiting obesity-related metabolic diseases.

DF-dRVFL: A novel deep feature based classifier for breast mass classification.

Amongst all types of cancer, breast cancer has become one of the most common cancers in the UK threatening millions of people's health. Early detection of breast cancer plays a key role in timely treatment for morbidity reduction. Compared to biopsy, which takes tissues from the lesion for further analysis, image-based methods are less time-consuming and pain-free though they are hampered by lower accuracy due to high false positivity rates. Nevertheless, mammography has become a standard screening method due to its high efficiency and low cost with promising performance. Breast mass, as the most palpable symptom of breast cancer, has received wide attention from the community. As a result, the past decades have witnessed the speeding development of computer-aided systems that are aimed at providing radiologists with useful tools for breast mass analysis based on mammograms. However, the main issues of these systems include low accuracy and require enough computational power on a large scale of datasets. To solve these issues, we developed a novel breast mass classification system called DF-dRVFL. On the public dataset DDSM with more than 3500 images, our best model based on deep random vector functional link network showed promising results through five-cross validation with an averaged AUC of 0.93 and an average accuracy of 81.71%. Compared to sole deep learning based methods, average accuracy has increased by 0.38. Compared with the state-of-the-art methods, our method showed better performance considering the number of images for evaluation and the overall accuracy.

NDUFA9 and its crotonylation modification promote browning of white adipocytes by activating mitochondrial function in mice.

Protein crotonylation plays a role in regulating cellular metabolism, gene expression, and other biological processes. NDUFA9 (NADH dehydrogenase [ubiquinone] 1 alpha subcomplex subunit 9) is closely associated with the activity and function of mitochondrial respiratory chain complex I. Mitochondrial function and respiratory chain are closely related to browning of white adipocytes, it's speculated that NDUFA9 and its crotonylation are associated with browning of white adipocytes. Firstly, the effect of NDUFA9 on white adipose tissue was verified in white fat browning model mice, and it was found that NDUFA9 promoted mitochondrial respiration, thermogenesis, and browning of white adipose tissue. Secondly, in cellular studies, it was discovered that NDUFA9 facilitated browning of white adipocytes by enhancing mitochondrial function, mitochondrial complex I activity, ATP synthesis, and mitochondrial respiration. Again, the level of NDUFA9 crotonylation was increased by treating cells with vorinostat (SAHA)+sodium crotonate (NaCr) and overexpressing NDUFA9, it was found that NDUFA9 crotonylation promoted browning of white adipocytes. Meanwhile, the acetylation level of NDUFA9 was increased by treating cells with SAHA+sodium acetate (NaAc) and overexpressing NDUFA9, the assay revealed that NDUFA9 acetylation inhibited white adipocytes browning. Finally, combined with the competitive relationship between acetylation and crotonylation, it was also demonstrated that NDUFA9 crotonylation promoted browning of white adipocytes. Above results indicate that NDUFA9 and its crotonylation modification promote mitochondrial function, which in turn promotes browning of white adipocytes. This study establishes a theoretical foundation for the management and intervention of obesity, which is crucial in addressing obesity and related medical conditions in the future.

Botany, phytochemistry, pharmacology, and applications of Pandanus amaryllifolius Roxb.: A review.

Pandan (Pandanus amaryllifolius Roxb.), a member of the Pandanaceae family, has been consumed as food and medicine since ancient times. The current paper provides an overview of the botanical profile, phytochemistry, pharmacology, and applications of P. amaryllifolius. Information regarding P. amaryllifolius was collected from online sources (using PubMed, Science Direct, Google Scholar, Web of Science, ACS, and CNKI) as well as traditional textbooks. Over 100 compounds have been identified, including its characteristic components 2-Acetyl-1-pyrroline and Pandanus alkaloids. Several therapeutic uses of P. amaryllifolius, such as antioxidant, hypoglycemic, antimicrobial, and antitumor activities, have been demonstrated in modern pharmacological studies. Additionally, it could be applied in various fields, including food, energy, material, and the environment. Continued research on P. amaryllifolius can contribute to the development of new drugs and therapies for various diseases. And further studies are needed to improve its utilization.

Integrated ATAC-seq and RNA-seq Analysis of In Vitro Cultured Skeletal Muscle Satellite Cells to Understand Changes in Cell Proliferation.

Skeletal muscle satellite cells, the resident stem cells in pig skeletal muscle, undergo proliferation and differentiation to enable muscle tissue repair. The proliferative and differentiative abilities of these cells gradually decrease during in vitro cultivation as the cell passage number increases. Despite extensive research, the precise molecular mechanisms that regulate this process are not fully understood. To bridge this knowledge gap, we conducted transcriptomic analysis of skeletal muscle satellite cells during in vitro cultivation to quantify passage number-dependent changes in the expression of genes associated with proliferation. Additionally, we explored the relationships between gene transcriptional activity and chromatin accessibility using transposase-accessible chromatin sequencing. This revealed the closure of numerous open chromatin regions, which were primarily located in intergenic regions, as the cell passage number increased. Integrated analysis of the transcriptomic and epigenomic data demonstrated a weak correlation between gene transcriptional activity and chromatin openness in expressed genic regions; although some genes (e.g., GNB4 and FGD5) showed consistent relationships between gene expression and chromatin openness, a substantial number of differentially expressed genes had no clear association with chromatin openness in expressed genic regions. The p53-p21-RB signaling pathway may play a critical regulatory role in cell proliferation processes. The combined transcriptomic and epigenomic approach taken here provided key insights into changes in gene expression and chromatin openness during in vitro cultivation of skeletal muscle satellite cells. These findings enhance our understanding of the intricate mechanisms underlying the decline in cellular proliferation capacity in cultured cells.

DLBCNet: A Deep Learning Network for Classifying Blood Cells.

Background: Blood is responsible for delivering nutrients to various organs, which store important health information about the human body. Therefore, the diagnosis of blood can indirectly help doctors judge a person's physical state. Recently, researchers have applied deep learning (DL) to the automatic analysis of blood cells. However, there are still some deficiencies in these models. Methods: To cope with these issues, we propose a novel network for the multi-classification of blood cells, which is called DLBCNet. A new specifical model for blood cells (BCGAN) is designed to generate synthetic images. The pre-trained ResNet50 is implemented as the backbone model, which serves as the feature extractor. The extracted features are fed to the proposed ETRN to improve the multi-classification performance of blood cells. Results: The average accuracy, average sensitivity, average precision, average specificity, and average f1-score of the proposed model are 95.05%, 93.25%, 97.75%, 93.72%, and 95.38%, accordingly. Conclusions: The performance of the proposed model surpasses other state-of-the-art methods in reported classification results.

Deoxynivalenol induces endoplasmic reticulum stress-associated apoptosis via the IRE1/JNK/CHOP pathway in porcine alveolar macrophage 3D4/21 cells.

The interplay between cell apoptosis and endoplasmic reticulum (ER) stress has garnered increasing attention. Nevertheless, the precise involvement of the unfolded protein response (UPR) signaling in the apoptosis of porcine macrophage cells induced by Deoxynivalenol (DON) remains enigmatic. In this study, we revealed that exposure to 2 µM DON resulted in a substantial decline in cell viability, concomitant with the initiation of cell apoptosis and the halting of the G1 phase cell cycle in the porcine alveolar macrophage line 3D4/21. Transcriptomic analysis of DON-exposed cells showed distinct expression patterns in 3104 genes, with notable upregulation of ER stress-related genes, including IRE1, CHOP, XBP1 and JNK. Our subsequent validation via qPCR and Western blot analyses confirmed the attenuation of GRP78 and BCL-2, coupled with the upregulation of IRE1, CHOP, JNK, p-JNK, and Bax in DON-induced cells, indicating the instigation of ER stress-associated apoptosis by DON. The addition of 5 mM 4-phenylbutyric acid (4-PBA), an ER stress inhibitor, decreased levels of CHOP, IRE1, JNK, p-JNK, and Bax, while increasing levels of GRP78 and Bcl-2, suggesting that 4-PBA alleviated DON-induced ER stress and apoptosis. Overall, our findings provide new insights into DON-induced ER stress via the IRE1/JNK/CHOP pathway, leading to subsequent cellular apoptosis.

A Hybrid Framework for Lung Cancer Classification.

Cancer is the second leading cause of death worldwide, and the death rate of lung cancer is much higher than other types of cancers. In recent years, numerous novel computer-aided diagnostic techniques with deep learning have been designed to detect lung cancer in early stages. However, deep learning models are easy to overfit, and the overfitting problem always causes lower performance. To solve this problem of lung cancer classification tasks, we proposed a hybrid framework called LCGANT. Specifically, our framework contains two main parts. The first part is a lung cancer deep convolutional GAN (LCGAN) to generate synthetic lung cancer images. The second part is a regularization enhanced transfer learning model called VGG-DF to classify lung cancer images into three classes. Our framework achieves a result of 99.84% ± 0.156% (accuracy), 99.84% ± 0.153% (precision), 99.84% ± 0.156% (sensitivity), and 99.84% ± 0.156% (F1-score). The result reaches the highest performance of the dataset for the lung cancer classification task. The proposed framework resolves the overfitting problem for lung cancer classification tasks, and it achieves better performance than other state-of-the-art methods.

LCDAE: Data Augmented Ensemble Framework for Lung Cancer Classification.

Objective: The only possible solution to increase the patients' fatality rate is lung cancer early-stage detection. Recently, deep learning techniques became the most promising methods in medical image analysis compared with other numerous computer-aided diagnostic techniques. However, deep learning models always get lower performance when the model is overfitting. Methods: We present a Lung Cancer Data Augmented Ensemble (LCDAE) framework to solve the overfitting and lower performance problems in the lung cancer classification tasks. The LCDAE has 3 parts: The Lung Cancer Deep Convolutional GAN, which can synthesize images of lung cancer; A Data Augmented Ensemble model (DA-ENM), which ensembled 6 fine-tuned transfer learning models for training, testing, and validating on a lung cancer dataset; The third part is a Hybrid Data Augmentation (HDA) which combines all the data augmentation techniques in the LCDAE. Results: By comparing with existing state-of-the-art methods, the LCDAE obtains the best accuracy of 99.99%, the precision of 99.99%, and the F1-score of 99.99%. Conclusion: Our proposed LCDAE can overcome the overfitting issue for the lung cancer classification tasks by applying different data augmentation techniques, our method also has the best performance compared to state-of-the-art approaches.

Determination of prostate cancer marker Zn2+ with a highly selective surface-enhanced Raman scattering probe on liquid-liquid self-assembled Au nanoarrays.

As the concentration of Zn2+ in patients with prostate cancer is much less than that in healthy persons, Zn2+ concentration can be used as a marker to expediently screen prostate cancer. In this study, a sensitive and highly selective surface-enhanced Raman scattering (SERS) method to detect Zn2+ concentration in human prostatic fluids by utilizing water-insoluble 2-carboxyl-2'-hydroxyl-5'-sulfoformazylbenze (Zincon) as a SERS probe based on self-assembled Au nanoarrays at a liquid-liquid interface between n-hexane and Au colloids was proposed. Zincon showed remarkably different SERS bands before and after coordinating Zn2+ in the controlled conditions (70 µL of ethanol, 500 µL of n-hexane, pH value of 7.1 and 10 s of vortex mixing time), which can be used in quantifying Zn2+ with characteristic peaks. The proposed SERS method presented a good linear relationship ranging from 0.5 to 10 µmol/L and a satisfactory detection limit of 0.1 µmol/L as well as low interference with other metal ions. Moreover, the detection results are close to those of the conventional standard atomic absorption spectroscopy (AAS) method.

Kinetic parameters estimation of pinus sylvestris pyrolysis by Kissinger-Kai method coupled with Particle Swarm Optimization and global sensitivity analysis.

Pyrolysis of pinewood (pinus sylvestris) was investigated based on thermogravimetric analysis. A new method was put forward to estimate its kinetic parameters by coupling model-free and model-fitting models. Kissinger-Kai method updated from Kissinger method was used as the representative of model-free method. Particle Swarm Optimization heuristic algorithm, as the typical model-fitting method, was coupled with three-component parallel reaction mechanism to search the optimized values, wherein its search ranges of kinetic parameters were referred to the original calculated values by Kissinger-Kai method. Furthermore, to explore the influence of separate kinetic parameter on the final predicted thermogravimetric results, global sensitivity analysis about these parameters was conducted by comparison of Spearman rank correlation coefficient based on Latin Hypercube Sampling and rank transformation. It was found that the top three parameters affecting the predicted results were activation energy of lignin, reaction order of cellulose and pre-exponential factor of lignin.