KHSRP Stabilizes m6A-Modified Transcripts to Activate FAK Signaling and Promote Pancreatic Ductal Adenocarcinoma Progression.

N6-methyladenosine (m6A) is the most prevalent RNA modification and is associated with various biological processes. Proteins that function as readers and writers of m6A modifications have been shown to play critical roles in human malignancies. Here, we identified KH-type splicing regulatory protein (KHSRP) as an m6A binding protein that contributes to the progression of pancreatic ductal adenocarcinoma (PDAC). High KHSRP levels were detected in PDAC and predicted poor patient survival. KHSRP deficiency suppressed PDAC growth and metastasis in vivo. Mechanistically, KHSRP recognized and stabilized FAK pathway mRNAs, including MET, ITGAV and ITGB1, in an m6A-dependent manner, which led to activation of downstream FAK signaling that promoted PDAC progression. Targeting KHSRP with a PROTAC showed promising tumor suppressive effects in mouse models, leading to prolonged survival. Together, these findings indicate that KHSRP mediates FAK pathway activation in an m6A-dependent manner to support PDAC growth and metastasis, highlighting the potential of KHSRP as a therapeutic target in pancreatic cancer.

Synthesis of higher-B0 CEST Z-spectra from lower-B0 data via deep learning and singular value decomposition.

Chemical exchange saturation transfer (CEST) MRI at 3 T suffers from low specificity due to overlapping CEST effects from multiple metabolites, while higher field strengths (B0) allow for better separation of Z-spectral "peaks," aiding signal interpretation and quantification. However, data acquisition at higher B0 is restricted by equipment access, field inhomogeneity and safety issues. Herein, we aim to synthesize higher-B0 Z-spectra from readily available data acquired with 3 T clinical scanners using a deep learning framework. Trained with simulation data using models based on Bloch-McConnell equations, this framework comprised two deep neural networks (DNNs) and a singular value decomposition (SVD) module. The first DNN identified B0 shifts in Z-spectra and aligned them to correct frequencies. After B0 correction, the lower-B0 Z-spectra were streamlined to the second DNN, casting into the key feature representations of higher-B0 Z-spectra, obtained through SVD truncation. Finally, the complete higher-B0 Z-spectra were recovered from inverse SVD, given the low-rank property of Z-spectra. This study constructed and validated two models, a phosphocreatine (PCr) model and a pseudo-in-vivo one. Each experimental dataset, including PCr phantoms, egg white phantoms, and in vivo rat brains, was sequentially acquired on a 3 T human and a 9.4 T animal scanner. Results demonstrated that the synthetic 9.4 T Z-spectra were almost identical to the experimental ground truth, showing low RMSE (0.11% ± 0.0013% for seven PCr tubes, 1.8% ± 0.2% for three egg white tubes, and 0.79% ± 0.54% for three rat slices) and high R2 (>0.99). The synthesized amide and NOE contrast maps, calculated using the Lorentzian difference, were also well matched with the experiments. Additionally, the synthesis model exhibited robustness to B0 inhomogeneities, noise, and other acquisition imperfections. In conclusion, the proposed framework enables synthesis of higher-B0 Z-spectra from lower-B0 ones, which may facilitate CEST MRI quantification and applications.

Improvement of model simulation for summer PM2.5 and O3 through coupling with two new potential HONO sources in the North China Plain.

A large fraction of fine particulate matter (PM2.5) and ozone (O3) in the troposphere originates from secondary formation through photochemical processes, which remarkably contributes to the deterioration of regional air quality in China. The photochemical reactions initiated by hydroxyl radicals (OH) play vital roles in secondary PM2.5 and O3 formation. In contrast, the OH levels in polluted areas are underestimated by current chemical transport models (CTMs) because of the strongly unknown daytime sources of tropospheric nitric acid (HONO), which has been recognized as the dominant source of primary OH in polluted areas of China. In this study, the atmospheric HONO levels at two urban sites were found to be significantly underestimated by the WRF-Chem model based on available information on HONO sources. The HONO levels could be well reproduced by the WRF-Chem model after incorporating two new potential HONO sources from the photochemical reactions of NOx, as proposed in our previous study based on chamber experiment results. Comparing the simulations with available information of HONO sources, the simulated levels of atmospheric OH, secondary inorganic and organic aerosols (SIA and SOA), PM2.5 and daily maximum 8-h average (MDA8) O3 were evidently elevated or were closer to the observations over the North China Plain (NCP), with elevation percentages of 0.48-20.1 %, and a decrement percentage of -5.79 % for pNO3-. Additionally, the compensating errors in modeling PM2.5 and the gap in MDA8 O3 levels between observation and simulation in 2 + 26 cities became evidently smaller. The results of this study indicated that the empirical parameterization of two new potential HONO sources through photochemical reactions of NOx improved the model performance in modeling PM2.5 and O3 by narrowing the gap in daytime HONO levels between simulation and observation, although their detailed chemical mechanisms are still unknown and should be further investigated and explicitly parameterized.

Light&fast generative adversarial network for high-fidelity CT image synthesis of liver tumor.

BACKGROUND AND OBJECTIVE: Hepatocellular carcinoma is a common disease with high mortality. Through deep learning methods to analyze HCC CT, the screening classification and prognosis model of HCC can be established, which further promotes the development of computer-aided diagnosis and treatment in the treatment of HCC. However, there are significant challenges in the actual establishment of HCC auxiliary diagnosis model due to data imbalance, especially for rare subtypes of HCC and underrepresented demographic groups. This study proposes a GAN model aimed at overcoming these obstacles and improving the accuracy of HCC auxiliary diagnosis. METHODS: In order to generate liver and tumor images close to the real distribution. Firstly, we construct a new gradient transfer sampling module to improve the lack of texture details and excessive gradient transfer parameters of the deep model; Secondly, we construct an attention module with spatial and cross-channel feature extraction ability to improve the discriminator's ability to distinguish images; Finally, we design a new loss function for liver tumor imaging features to constrain the model to approach the real tumor features in iterations. RESULTS: In qualitative analysis, the images synthetic by our method closely resemble the real images in liver parenchyma, blood vessels, tumors, and other parts. In quantitative analysis, the optimal results of FID, PSNR, and SSIM are 75.73, 22.77, and 0.74, respectively. Furthermore, our experiments establish classification models for imbalanced data and enhanced data, resulting in an increase in accuracy rate by 21%-34%, an increase in AUC by 0.29 - 0.33, and an increase in specificity to 0.89. CONCLUSION: Our solution provides a variety of training data sources with low cost and high efficiency for the establishment of classification or prognostic models for imbalanced data.

The complex world of kefir: Structural insights and symbiotic relationships.

Kefir milk, known for its high nutritional value and health benefits, is traditionally produced by fermenting milk with kefir grains. These grains are a complex symbiotic community of lactic acid bacteria, acetic acid bacteria, yeasts, and other microorganisms. However, the intricate coexistence mechanisms within these microbial colonies remain a mystery, posing challenges in predicting their biological and functional traits. This uncertainty often leads to variability in kefir milk's quality and safety. This review delves into the unique structural characteristics of kefir grains, particularly their distinctive hollow structure. We propose hypotheses on their formation, which appears to be influenced by the aggregation behaviors of the community members and their alliances. In kefir milk, a systematic colonization process is driven by metabolite release, orchestrating the spatiotemporal rearrangement of ecological niches. We place special emphasis on the dynamic spatiotemporal changes within the kefir microbial community. Spatially, we observe variations in species morphology and distribution across different locations within the grain structure. Temporally, the review highlights the succession patterns of the microbial community, shedding light on their evolving interactions.Furthermore, we explore the ecological mechanisms underpinning the formation of a stable community composition. The interplay of cooperative and competitive species within these microorganisms ensures a dynamic balance, contributing to the community's richness and stability. In kefir community, competitive species foster diversity and stability, whereas cooperative species bolster mutualistic symbiosis. By deepening our understanding of the behaviors of these complex microbial communities, we can pave the way for future advancements in the development and diversification of starter cultures for food fermentation processes.

Analysis of Distribution and Drug Susceptibility Test Results of Pathogenic Bacteria in Diabetic Foot Ulcers.

INTRODUCTION: This study aimed to determine the pathogen distribution and drug susceptibility of diabetic foot wound secretions in a tertiary hospital in a coastal area of southeastern China to guide clinical antibiotic selection. METHODS: A retrospective analysis was conducted on 212 patients with diabetic foot hospitalized at Xiamen Third Hospital from 2018 to 2023, and foot wound secretions were collected for microbial culture and drug susceptibility testing. RESULTS: Among 212 cases of patients with diabetic foot wound secretions, 163 cases (76.9%) were cultured with pathogenic bacteria, and a total of 207 strains of pathogenic bacteria were cultured, including 75 strains (36.23%) of Gram-positive (G+) bacteria, 118 strains of Gram-negative (G-) bacteria (57.00%), 14 strains of fungi (6.76%), 120 cases of single microorganism infection (73.62%), 43 cases of mixed infection (26.38%), and 15 strains of multidrug-resistant bacteria (7.25%). The top three pathogenic bacteria were Staphylococcus aureus, Klebsiella pneumoniae, and Pseudomonas aeruginosa. G+ bacteria were dominated by S. aureus. Drug susceptibility results showed that G+ bacteria were highly susceptible to vancomycin, linezolid, tigecycline, quinupristin/dalfopristin, rifampicin, and furotoxin, and somewhat resistant to penicillin, erythromycin, clindamycin, and cefoxitin. Among G- bacterial infections, Klebsiella pneumoniae, Pseudomonas aeruginosa, Escherichia coli, and Proteus were the major species. Drug susceptibility testing indicated that carbapenems such as imipenem and ertapenem were the most effective antibacterial drugs against G- strains, followed by amikacin, piperacillin, and tazabactams to which these bacteria were also relatively sensitive, while resistance to penicillins and first-generation cephalosporins increased significantly. We isolated one strain of pathogenic bacteria from a Wagner grade 1 ulcer, which was G+ bacteria. In Wagner grade 2 ulcers, the distribution of pathogenic bacteria was mainly G+ bacteria. In Wagner grade 3 and 4 ulcers, the distribution of pathogenic bacteria was mainly G- bacteria, and the increased rate of mixed infection was mainly due to mixed infection of G+ and G-. Two strains of pathogenic bacteria were isolated at Wagner grade 5, which were mixed infections of G+ and G-. CONCLUSIONS: Pathogenic bacteria in diabetic foot wounds are predominantly G- bacteria, followed by G+ bacteria. As the Wagner ulcer grade increases, the distribution of pathogenic bacteria changes from G+ bacteria to G- bacteria, and the mixed infection rate increases. G+ bacteria are highly susceptible to vancomycin, linezolid, tigecycline, quinupristin/dalfopristin, rifampicin, and furotoxin, and somewhat resistant to penicillin, erythromycin, clindamycin, and cefoxitin. G- bacteria are more sensitive to the antimicrobial drugs ertapenem, imipenem, amikacin, piperacillin tazobactam, and have high resistance to penicillin and first-generation cephalosporins.

QDPR deficiency drives immune suppression in pancreatic cancer.

The relevance of biopterin metabolism in resistance to immune checkpoint blockade (ICB) therapy remains unknown. We demonstrate that the deficiency of quinoid dihydropteridine reductase (QDPR), a critical enzyme regulating biopterin metabolism, causes metabolite dihydrobiopterin (BH2) accumulation and decreases the ratio of tetrahydrobiopterin (BH4) to BH2 in pancreatic ductal adenocarcinomas (PDACs). The reduced BH4/BH2 ratio leads to an increase in reactive oxygen species (ROS) generation and a decrease in the distribution of H3K27me3 at CXCL1 promoter. Consequently, myeloid-derived suppressor cells are recruited to tumor microenvironment via CXCR2 causing resistance to ICB therapy. We discovered that BH4 supplementation is capable to restore the BH4/BH2 ratio, enhance anti-tumor immunity, and overcome ICB resistance in QDPR-deficient PDACs. Tumors with lower QDPR expression show decreased responsiveness to ICB therapy. These findings offer a novel strategy for selecting patient and combining therapies to improve the effectiveness of ICB therapy in PDAC.

Group sparse-based Taylor expansion method for liver pharmacokinetic parameters imaging of dynamic fluorescence molecular tomography.

Objective.Pharmacokinetic parametric images obtained through dynamic fluorescence molecular tomography (DFMT) has ability of capturing dynamic changes in fluorescence concentration, thereby providing three-dimensional metabolic information for applications in biological research and drug development. However, data processing of DFMT is time-consuming, involves a vast amount of data, and the problem itself is ill-posed, which significantly limits the application of pharmacokinetic parametric images reconstruction. In this study, group sparse-based Taylor expansion method is proposed to address these problems.Approach.Firstly, Taylor expansion framework is introduced to reduce time and computational cost. Secondly, group sparsity based on structural prior is introduced to improve reconstruction accuracy. Thirdly, alternating iterative solution based on accelerated gradient descent algorithm is introduced to solve the problem.Main results.Numerical simulation andin vivoexperimental results demonstrate that, in comparison to existing methods, the proposed approach significantly enhances reconstruction speed without a degradation of quality, particularly when confronted with background fluorescence interference from other organs.Significance.Our research greatly reduces time and computational cost, providing strong support for real-time monitoring of liver metabolism.

A Deep Learning Model for Predicting Molecular Subtype of Breast Cancer by Fusing Multiple Sequences of DCE-MRI From Two Institutes.

RATIONALE AND OBJECTIVES: To evaluate the performance of deep learning (DL) in predicting different breast cancer molecular subtypes using DCE-MRI from two institutes. MATERIALS AND METHODS: This retrospective study included 366 breast cancer patients from two institutes, divided into training (n = 292), validation (n = 49) and testing (n = 25) sets. We first transformed the public DCE-MRI appearance to ours to alleviate small-data-size and class-imbalance issues. Second, we developed a multi-branch convolutional-neural-network (MBCNN) to perform molecular subtype prediction. Third, we assessed the MBCNN with different regions of interest (ROIs) and fusion strategies, and compared it to previous DL models. Area under the curve (AUC) and accuracy (ACC) were used to assess different models. Delong-test was used for the comparison of different groups. RESULTS: MBCNN achieved the optimal performance under intermediate fusion and ROI size of 80 pixels with appearance transformation. It outperformed CNN and convolutional long-short-term-memory (CLSTM) in predicting luminal B, HER2-enriched and TN subtypes, but without demonstrating statistical significance except against CNN in TN subtypes, with testing AUCs of 0.8182 vs. [0.7208, 0.7922] (p=0.44, 0.80), 0.8500 vs. [0.7300, 0.8200] (p=0.36, 0.70) and 0.8900 vs. [0.7600, 0.8300] (p=0.03, 0.63), respectively. When predicting luminal A, MBCNN outperformed CNN with AUCs of 0.8571 vs. 0.7619 (p=0.08) without achieving statistical significance, and is comparable to CLSTM. For four-subtype prediction, MBCNN achieved an ACC of 0.64, better than CNN and CLSTM models with ACCs of 0.48 and 0.52, respectively. CONCLUSION: Developed DL model with the feature extraction and fusion of DCE-MRI from two institutes enabled preoperative prediction of breast cancer molecular subtypes with high diagnostic performance.

A Novel Prolene Suture Pull-out Technique Combined with Anchor Technique for the Treatment of Tendinous Mallet Finger.

Background: Mallet finger deformity is a prevalent disability that causes discomfort and inconvenience to the patients. Despite the existence of various surgical approaches, surgical management remains a controversial subject. Methods: We retrospectively analyzed the clinical data of 26 patients with isolated tendinous mallet fingers who were admitted between January 2021 and June 2022. Among them, there were 18 men and eight women, aged between 20 and 56 years, with an average age of 38 years. The causes of injury were cutting injuries (15 cases), sports impact injuries (nine cases), and sprains (two cases). The time interval between injury and surgery ranged from 2 hours to 48 days, with an average of 12 days. During the surgical procedure, the distal interphalangeal joint was fixed in a mild dorsiflexion position using Kirschner wire. Absorbable anchors were used to assist in the reconstruction of the insertion point of the finger extensor tendon. Additionally, a 4-0 Prolene suture was used for reinforcement. Results: All 26 patients were followed up for a period ranging from 6 to 24 months, with an average follow-up duration of 9 months. The function of distal interphalangeal joint was preserved. According to the Crawford functional evaluation criteria, the function of the affected fingers was excellent in 15 cases, good in eight cases, fair in three cases, and poor in no cases. Conclusions: A novel Prolene suture pull-out technique is an effective approach to repair tendon mallet finger and reconstruct the tendon-bone anatomical unit. This treatment option provides favorable outcomes, with high rates of excellent and good functional results.

Nonreciprocal P T-symmetric magnon laser in spinning cavity optomagnonics.

We propose a scheme to achieve nonreciprocal parity-time (P T)-symmetric magnon laser in a P T-symmetric cavity optomagnonical system. The system consists of active and passive optical spinning resonators. We demonstrate that the Fizeau light-dragging effect induced by the spinning of a resonator results in significant variations in magnon gain and stimulated emitted magnon numbers for different driving directions. We find that utilizing the Fizeau light-dragging effect allows the system to operate at ultra-low thresholds even without reaching gain-loss balance. A one-way magnon laser can also be realized across a range of parameters. High tunability of the magnon laser is achieved by changing the spinning speed of the resonators and driving direction. Our work provides a new way to explore various nonreciprocal effects in non-Hermitian magnonic systems, which may be applied to manipulate photons and magnons in multi-body non-Hermitian coupled systems.

FSMN-Net: a free space matching network based on manifold convolution for optical molecular tomography.

Optical molecular tomography (OMT) can monitor glioblastomas in small animals non-invasively. Although deep learning (DL) methods have made remarkable achievements in this field, improving its generalization against diverse reconstruction systems remains a formidable challenge. In this Letter, a free space matching network (FSMN-Net) was presented to overcome the parameter mismatch problem in different reconstruction systems. Specifically, a novel, to the best of our knowledge, manifold convolution operator was designed by considering the mathematical model of OMT as a space matching process. Based on the dynamic domain expansion concept, an end-to-end fully convolutional codec further integrates this operator to realize robust reconstruction with voxel-level accuracy. The results of numerical simulations and in vivo experiments demonstrate that the FSMN-Net can stably generate high-resolution reconstruction volumetric images under different reconstruction systems.

Comprehension of Ditransitive Constructions in Mandarin-Speaking Children with Developmental Language Disorder and Children with Autism Spectrum Disorder Plus Language Impairment.

This study examined and compared the comprehension of Mandarin ditransitive constructions in children with developmental language disorder (DLD) and children with autism spectrum disorder plus language impairment (ALI). Eighteen children with DLD, 17 children with ALI, and 27 age-matched typically developing (TDA) children, participated in a sentence-picture matching task on four patterns of Mandarin ditransitive constructions. Both children with DLD and children with ALI received significantly lower accuracy than TDA children in general and their most common errors were thematic role reversals. However, while children with ALI evinced a generalized deficit in all four patterns, only the comprehension of S1 (Subj. + Vgei + IO + DO) and S3 (Subj. + gei + IO + V + DO) was affected in children with DLD, with that of S2 (Subj. + V + DO + gei + IO) and S4 (Subj. + V + IO + DO) preserved in this population. Additionally, thematic role reversal errors were more dominant in children with DLD than in children with ALI who also committed a relatively higher proportion of Wrong Theme and No Recipient errors. It is concluded that the primary deficit of children with DLD lies in representing dependent relationships between the arguments and the verb as involved in thematic role assignment, but this is less critical in children with ALI, with their performance on the comprehension task possibly also related to other factors associated with the condition. To enhance the development of ditransitive constructions, intervention efforts for children with DLD and children with ALI could focus on strengthening the connection between each argument and its thematic role.

Enhanced model iteration algorithm with graph neural network for diffuse optical tomography.

Diffuse optical tomography (DOT) employs near-infrared light to reveal the optical parameters of biological tissues. Due to the strong scattering of photons in tissues and the limited surface measurements, DOT reconstruction is severely ill-posed. The Levenberg-Marquardt (LM) is a popular iteration method for DOT, however, it is computationally expensive and its reconstruction accuracy needs improvement. In this study, we propose a neural model based iteration algorithm which combines the graph neural network with Levenberg-Marquardt (GNNLM), which utilizes a graph data structure to represent the finite element mesh. In order to verify the performance of the graph neural network, two GNN variants, namely graph convolutional neural network (GCN) and graph attention neural network (GAT) were employed in the experiments. The results showed that GCNLM performs best in the simulation experiments within the training data distribution. However, GATLM exhibits superior performance in the simulation experiments outside the training data distribution and real experiments with breast-like phantoms. It demonstrated that the GATLM trained with simulation data can generalize well to situations outside the training data distribution without transfer training. This offers the possibility to provide more accurate absorption coefficient distributions in clinical practice.

Porous FeNi Prussian blue cubes derived carbon-based phosphides as superior sulfur hosts for high-performance lithium-sulfur batteries.

In contrast to lithium-ion batteries, lithium-sulfur batteries have higher theoretical energy density and lower cost, so they would become competitive in the practical application. However, the shuttle effect of polysulfides and slow oxidation-reduction kinetics can degrade their electrochemical performance and cycle life. In this work, we have first developed the porous FeNi Prussian blue cubes as precursors. The calcination in different atmospheres was employed to make precursors convert into common pyrolysis products or novel carbon-based phosphides, and sulfides, labeled as FeNiP/A-C, FeNiP/A-P, and FeNiP/A-S. When these products serve as host materials in the sulfur cathode, the electrochemical performance of lithium-sulfur batteries is in the order of S@FeNiP/A-P > S@FeNiP/A-S > S@FeNiP/A-C. Specifically, the initial discharge capacity of S@FeNiP/A-P can reach 679.1 mAh g-1at 1 C, and the capacity would maintain 594.6 mAh g-1after 300 cycles. That is because the combination of carbon-based porous structure and numerous well-dispersed Ni2P/Fe2P active sites contribute FeNiP/A-P to obtain larger lithium-ion diffusion, lower resistance, stronger chemisorption, and more excellent catalytic effect than other samples. This work may deliver that metal-organic framework-derived carbon-based phosphides are more suitable to serve as sulfur hosts than carbon-based sulfides or common pyrolysis products for enhancing Li-S batteries' performance.

Two-step reconstruction framework of fluorescence molecular tomography based on energy statistical probability.

Fluorescence molecular tomography (FMT), as a promising technique for early tumor detection, can non-invasively visualize the distribution of fluorescent marker probe three-dimensionally. However, FMT reconstruction is a severely ill-posed problem, which remains an obstacle to wider application of FMT. In this paper, a two-step reconstruction framework was proposed for FMT based on the energy statistical probability. First, the tissue structural information obtained from computed tomography (CT) is employed to associate the tissue optical parameters for rough solution in the global region. Then, according to the global-region reconstruction results, the probability that the target belongs to each region can be calculated. The region with the highest probability is delineated as region of interest to realize accurate and fast source reconstruction. Numerical simulations and in vivo experiments were carried out to evaluate the effectiveness of the proposed framework. The encouraging results demonstrate the significant effectiveness and potential of our method for practical FMT applications.

Dynamic fluorescence molecular tomography metabolic parameters solution based on problem decomposition and prior refactor.

Dynamic fluorescence molecular tomography (DFMT), as a noninvasive optical imaging method, can quantify metabolic parameters of living animal organs and assist in the diagnosis of metabolic diseases. However, existing DFMT methods do not have a high capacity to reconstruct abnormal metabolic regions, and require additional prior information and complicated solution methods. This paper introduces a problem decomposition and prior refactor (PDPR) method. The PDPR decomposes the metabolic parameters into two kinds of problems depending on their temporal coupling, which are solved using regularization and parameter fitting. Moreover, PDPR introduces the idea of divide-and-conquer to refactor prior information to ensure discrimination between metabolic abnormal regions and normal tissues. Experimental results show that PDPR is capable of separating abnormal metabolic regions of the liver and has the potential to quantify metabolic parameters and diagnose liver metabolic diseases in small animals.

CORONet: A Cross-Sequence Joint Representation and Hypergraph Convolutional Network for Classifying Molecular Subtypes of Breast Cancer Using Incomplete DCE-MRI.

Breast cancer, the predominant malignancy among women, is characterized by significant heterogeneity, leading to the emergence of distinct molecular subtypes. Accurate differentiation of these molecular subtypes holds paramount clinical significance, owing to substantial variations in prognosis, therapeutic strategies, and survival outcomes. In this study, we propose a cross-sequence joint representation and hypergraph convolution network (CORONet) for classifying molecular subtypes of breast cancer using incomplete DCE-MRI. Specifically, we first build a cross-sequence joint representation (COR) module to integrate image imputation and feature representation into a unified framework, encouraging effective feature extraction for subsequent classification. Then, we fuse multiple COR features and applied feature selection to reduce the redundant information between sequences. Finally, we deploy hypergraph structures to model high-order correlation among different subjects and extracted high-level semantic features by hypergraph convolutions for molecular subtyping. Extensive experiments on incomplete DCE-MRIs of 395 patients from the TCIA repository showed a significant improvement of our CORONet over state of the arts, with the area under the curve (AUC) of 0.891 and 0.903 for luminal and triple-negative (TN) subtype prediction, respectively. Similar advantages of CORONet were also confirmed in partial complete DCE-MRIs of 144 patients, achieving an AUC of 0.858 and 0.832 for predicting luminal and TN subtypes of breast cancer, respectively. Nevertheless, both of these values were lower compared to the scenario where DCE-MRIs from all 395 patients were utilized. Our study contributes to the precise molecular subtyping using incomplete multi-sequence DCE-MRI, thereby offering promising prospects for future risk stratification of breast cancer patients.

Infrared Light-Emitting Diodes Based on Chirality-Sorted Carbon Nanotube Films.

An important goal in carbon nanotube optoelectronics is to achieve a high-performance near-infrared light source. But there are still many challenges such as the purity of single-walled carbon nanotube (SWCNT) chirality, nonradiative defects, thin-film quality, and device structure design. Here, we realize infrared light-emitting diodes (LEDs) based on chirality-sorted (10, 5) SWCNT network films, which operate at a low bias voltage and emit at a telecom O band of 1290 nm. Asymmetric palladium (Pd) and hafnium (Hf) contacts are used as electrodes for hole and electron injection, respectively. However, the large Schottky barrier at the interface of the SWCNTs and the Hf electrode, primarily resulting from the polymer wrapped on the nanotube surface during the sorting process, leads to inefficient electron injection and thus a low electroluminescence efficiency. We find that the efficiency of electron injection can be improved by the local doping of the nanotubes with dielectric layers of YOX-HfO2, which reduces the Schottky barrier at the SWCNT/Hf interface. Accordingly, the (10, 5) SWCNT film-based LED achieves an external quantum efficiency of larger than 0.05% without any optical coupling structure. With further improvement, we expect that such an infrared light source will have great application potential in the carbon nanotube monolithic optoelectronic integrated system and on-chip optical interconnection, especially in the field of short-distance optical fiber communications and data center.