Parotid Gland Segmentation Using Purely Transformer-Based U-Shaped Network and Multimodal MRI.

Parotid gland tumors account for approximately 2% to 10% of head and neck tumors. Segmentation of parotid glands and tumors on magnetic resonance images is essential in accurately diagnosing and selecting appropriate surgical plans. However, segmentation of parotid glands is particularly challenging due to their variable shape and low contrast with surrounding structures. Recently, deep learning has developed rapidly, and Transformer-based networks have performed well on many computer vision tasks. However, Transformer-based networks have yet to be well used in parotid gland segmentation tasks. We collected a multi-center multimodal parotid gland MRI dataset and implemented parotid gland segmentation using a purely Transformer-based U-shaped segmentation network. We used both absolute and relative positional encoding to improve parotid gland segmentation and achieved multimodal information fusion without increasing the network computation. In addition, our novel training approach reduces the clinician's labeling workload by nearly half. Our method achieved good segmentation of both parotid glands and tumors. On the test set, our model achieved a Dice-Similarity Coefficient of 86.99%, Pixel Accuracy of 99.19%, Mean Intersection over Union of 81.79%, and Hausdorff Distance of 3.87. The purely Transformer-based U-shaped segmentation network we used outperforms other convolutional neural networks. In addition, our method can effectively fuse the information from multi-center multimodal MRI dataset, thus improving the parotid gland segmentation.

Swin MoCo: Improving parotid gland MRI segmentation using contrastive learning.

BACKGROUND: Segmentation of the parotid glands and tumors by MR images is essential for treating parotid gland tumors. However, segmentation of the parotid glands is particularly challenging due to their variable shape and low contrast with surrounding structures. PURPOSE: The lack of large and well-annotated datasets limits the development of deep learning in medical images. As an unsupervised learning method, contrastive learning has seen rapid development in recent years. It can better use unlabeled images and is hopeful to improve parotid gland segmentation. METHODS: We propose Swin MoCo, a momentum contrastive learning network with Swin Transformer as its backbone. The ImageNet supervised model is used as the initial weights of Swin MoCo, thus improving the training effects on small medical image datasets. RESULTS: Swin MoCo trained with transfer learning improves parotid gland segmentation to 89.78% DSC, 85.18% mIoU, 3.60 HD, and 90.08% mAcc. On the Synapse multi-organ computed tomography (CT) dataset, using Swin MoCo as the pre-trained model of Swin-Unet yields 79.66% DSC and 12.73 HD, which outperforms the best result of Swin-Unet on the Synapse dataset. CONCLUSIONS: The above improvements require only 4 h of training on a single NVIDIA Tesla V100, which is computationally cheap. Swin MoCo provides new approaches to improve the performance of tasks on small datasets. The code is publicly available at https://github.com/Zian-Xu/Swin-MoCo.

Synthesis of Metal-Nitrogen-Carbon Electrocatalysts with Atomically Regulated Nitrogen-Doped Polycyclic Aromatic Hydrocarbons.

Tuning the active site structure of metal-nitrogen-carbon electrocatalysts has recently attracted increasing interest. Herein, we report a bottom-up synthesis strategy in which atomically regulated N-doped polycyclic aromatic hydrocarbons (N-PAHs) of NxC42-x (x = 1, 2, 3, 4) were used as ligands to allow tuning of the active site's structures of M-Nx and establish correlations between the structures and electrocatalytic properties. Based on the synthesis process, detailed characterization, and DFT calculation results, active structures of Nx-Fe1-Nx in Fe1-Nx/RGO catalysts were constructed. The results demonstrated that the extra uncoordinated N atoms around the Fe1-N4 moieties disrupted the π-conjugated NxC42-x ligands, which led to more localized electronic state in the Fe1-N4 moieties and superior catalytic performance. Especially, the Fe1-N4/RGO exhibited optimized performance for ORR with E1/2 increasing by 80 mV and Jk at 0.85 V improved 18 times (compared with Fe1-N1/RGO). This synthesis strategy utilizing N-PAHs holds significant promise for enhancing the controllability of metal-nitrogen-carbon electrocatalyst preparation.

Real-world cost- effectiveness analysis: Tumor Treating Fields for newly diagnosed glioblastoma in China.

BACKGROUND: Glioblastoma (GBM) stands as the most aggressive and prevalent primary brain malignancy. Tumor Treating Fields (TTFields), an innovative therapy complementing chemotherapy for GBM treatment, which can significantly enhance overall survival, disease progression-free survival, and patient's quality of life. However, there is a dearth of health economics evaluation on TTFields therapy both domestically and internationally. OBJECTIVE: The study aims to assess the cost-effectiveness of TTFields + temozolomide (TMZ) in comparison to TMZ alone for newly diagnosed GBM patients. The intent is to provide robust economic evidence to serve as a foundation for policymaking and decision-making processes in GBM treatment. METHODS: We estimated outcomes for newly diagnosed GBM patients over a lifetime horizon using a partitioned survival model with three states: Progression-Free Survival, Progression Disease, and Death. The survival model was derived from a real-world study in China, with long-term survival data drawn from GBM epidemiology literature. Adverse event rates were sourced from the EF-14 trial data. Cost data, validated by expert consultation, was obtained from public literature and databases. Utility values were extracted from published literature. Using Microsoft Excel, we calculated expected costs and quality-adjusted life years (QALYs) over 15 years from a health system perspective. The willingness-to-pay threshold was set at three times the Chinese per capita Gross Domestic Product (GDP) in 2022, amounting to CN¥242,928 (US$37,655) /QALY. A 5% discount rate was applied to costs and utilities. Results underwent analysis through single factor and probability sensitivity analyses. RESULTS: TTFields + TMZ demonstrated a mean increase in cost by CN¥389,326 (US$57,859) and an increase of 2.46 QALYs compared to TMZ alone. The incremental cost-effectiveness ratio (ICER) was CN¥157,979 (US$23,474) per QALY gained. The model exhibited heightened sensitivity to changes in the discount rate. Probability sensitivity analysis indicates that, under the existing threshold, the probability of TTFields + TMZ being economical is 95.60%. CONCLUSIONS: This cost-effectiveness analysis affirms that incorporating TTFields into TMZ treatment proves to be cost-effective, given a threshold three times the Chinese per capita GDP.

Topological minibands and interaction driven quantum anomalous Hall state in topological insulator based moiré heterostructures.

The presence of topological flat minibands in moiré materials provides an opportunity to explore the interplay between topology and correlation. In this work, we study moiré minibands in topological insulator films with two hybridized surface states under a moiré superlattice potential created by two-dimensional insulating materials. We show the lowest conduction (highest valence) Kramers' pair of minibands can be Z 2 non-trivial when the minima (maxima) of moiré potential approximately form a hexagonal lattice with six-fold rotation symmetry. Coulomb interaction can drive the non-trivial Kramers' minibands into the quantum anomalous Hall state when they are half-filled, which is further stabilized by applying external gate voltages to break inversion. We propose the monolayer Sb2 on top of Sb2Te3 films as a candidate based on first principles calculations. Our work demonstrates the topological insulator based moiré heterostructure as a potential platform for studying interacting topological phases.

Robustness of Trion State in Gated Monolayer MoSe2 under Pressure.

Quasiparticles consisting of correlated electron(s) and hole(s), such as excitons and trions, play important roles in the optical phenomena of van der Waals semiconductors and serve as unique platforms for studies of many-body physics. Herein, we report a gate-tunable exciton-to-trion transition in pressurized monolayer MoSe2, in which the electronic band structures are modulated continuously within a diamond anvil cell. The emission energies of both the exciton and trion undergo large blueshifts over 90 meV with increasing pressure. Surprisingly, the trion binding energy remains constant at 30 meV, regardless of the applied pressure. Combining ab initio density functional theory calculations and quantum Monte Carlo simulations, we find that the remarkable robustness of the trion binding energy originates from the spatially diffused nature of the trion wave function and the weak correlation between its constituent electron-hole pairs. Our findings shed light on the optical properties of correlated excitonic quasiparticles in low-dimensional materials.

An anisotropic van der Waals dielectric for symmetry engineering in functionalized heterointerfaces.

Van der Waals dielectrics are fundamental materials for condensed matter physics and advanced electronic applications. Most dielectrics host isotropic structures in crystalline or amorphous forms, and only a few studies have considered the role of anisotropic crystal symmetry in dielectrics as a delicate way to tune electronic properties of channel materials. Here, we demonstrate a layered anisotropic dielectric, SiP2, with non-symmorphic twofold-rotational C2 symmetry as a gate medium which can break the original threefold-rotational C3 symmetry of MoS2 to achieve unexpected linearly-polarized photoluminescence and anisotropic second harmonic generation at SiP2/MoS2 interfaces. In contrast to the isotropic behavior of pristine MoS2, a large conductance anisotropy with an anisotropy index up to 1000 can be achieved and modulated in SiP2-gated MoS2 transistors. Theoretical calculations reveal that the anisotropic moiré potential at such interfaces is responsible for the giant anisotropic conductance and optical response. Our results provide a strategy for generating exotic functionalities at dielectric/semiconductor interfaces via symmetry engineering.

Swin MAE: Masked autoencoders for small datasets.

The development of deep learning models in medical image analysis is majorly limited by the lack of large-sized and well-annotated datasets. Unsupervised learning does not require labels and is more suitable for solving medical image analysis problems. However, most unsupervised learning methods must be applied to large datasets. To make unsupervised learning applicable to small datasets, we proposed Swin MAE, a masked autoencoder with Swin Transformer as its backbone. Even on a dataset of only a few thousand medical images, Swin MAE can still learn useful semantic features purely from images without using any pre-trained models. It can equal or even slightly outperform the supervised model obtained by Swin Transformer trained on ImageNet in the transfer learning results of downstream tasks. Compared to MAE, Swin MAE brought a performance improvement of twice and five times for downstream tasks on BTCV and our parotid dataset, respectively. The code is publicly available at https://github.com/Zian-Xu/Swin-MAE.

Dabigatran etexilate is efficacious in consumptive coagulopathy and pain associated with venous malformations.

OBJECTIVE: Consumptive coagulopathy treatment and pain management are crucial for patients with venous malformations (VMs). Dabigatran etexilate, a non-vitamin K antagonist oral anticoagulant, has known advantages compared with low-molecular-weight heparin and vitamin K antagonists, including oral administration, a more consistent pharmacokinetics/pharmacodynamics profile, a better safety profile, and no need for coagulation surveillance. In the present study, we tested the efficacy and safety of dabigatran etexilate for consumptive coagulopathy treatment and pain management for patients with VMs. METHODS: To investigate the efficacy and safety of dabigatran etexilate in treating localized intravascular coagulation (LIC) associated with VM, we retrospectively collected data for 19 outpatients with VM and LIC, who had been treated with dabigatran etexilate from September 2019 to June 2021. The patients provided oral informed consent and underwent biologic blood testing, routine examinations, and determination of coagulation function before and after treatment. The dosage of dabigatran etexilate was 110 mg twice daily for adults and 55 mg twice daily for children. RESULTS: All 19 patients had benefited from dabigatran etexilate treatment with coagulation improvement and pain relief. Pain had improved in all 16 evaluable patients. The fibrinogen and D-dimer levels had improved in 18 of 19 patients. The fibrin degradation product level had improved in 10 of 14 patients. None of patients reported lesion regression, appearance changes, or improvement in mobility. No significant differences were found in the D-dimer, fibrinogen, and fibrin degradation product levels between the short-term (<10 days) and long-term (≥10 days) use of the medication. Dabigatran etexilate was well tolerated by all patients. No bleeding event had occurred during follow-up. CONCLUSIONS: The results of our study have confirmed the efficacy and safety of dabigatran etexilate in treating pain and LIC in patients with VMs. Dabigatran etexilate is a suitable choice preoperatively to modify coagulation function and pain in patients with VMs.

Evolution of Electronic Structure in Pristine and Rb-Reconstructed Surfaces of Kagome Metal RbV3Sb5.

We report on in situ low-temperature (4 K) scanning tunneling microscope measurements of atomic and electronic structures of the cleaved surfaces of an alkali-based kagome metal RbV3Sb5 single crystals. We find that the dominant pristine surface exhibits Rb-1×1 structure, in which a unique unidirectional √3a0 charge order is discovered. As the sample temperature slightly rises, Rb-√3×1 and Rb-√3×√3 reconstructions form due to desorption of surface Rb atoms. Our conductance mapping results demonstrate that Rb desorption not only gives rise to hole doping but also reconstructs the electronic band structures. Surprisingly, we find a ubiquitous gap opening near the Fermi level in tunneling spectra on all the surfaces despite their large differences of hole-carrier concentration, indicating an orbital-selective band reconstruction in RbV3Sb5. The Rb desorption induced electronic reconstructions are further confirmed by our density functional theory calculations.

Construction and Application of Bridge Expansion and Contraction Installation Replacement Decision System Based on the Analytic Hierarchy Process.

Bridge expansion and contraction installation (BECI) has proved to be an indispensable component of bridge structures due to its stability, comfort, and durability benefits. At present, conventional replacement technologies for modular-type, comb plate-type, and seamless-type BECIs are widely applied worldwide. However, it is unfortunate that there remains no systematic research on quantitative assessment approaches for evaluating the overall technical status and selecting optimal replacement methods for existing BECIs. Therefore, considering the installation performance according to functional index evaluations and the economic cost based on life-cycle value assessment (LCVA), a standardized quantitative assessment approach is proposed for optimal replacement method selection in this article. Simultaneously, the other new quantitative assessment method is developed for evaluating the overall technical status of BECIs, which provides a basis for the necessity of replacement. A BECI replacement decision system is constructed, and a corresponding case study illustrates that the proposed system based on the analytic hierarchy process (AHP) in this article proves to be reasonable and feasible. The results reveal that the selected replacement method with both a higher function coefficient and a lower economic coefficient can not only fulfil the performance requirements but also pursue a cost reduction, which leads to a considerable value increment. This system can effectively assist bridge managers in making appropriate operation and maintenance (O and M) decisions in actual engineering projects.

Total synthesis of polyene natural product dihydroxerulin by mild organocatalyzed dehydrogenation of alcohols.

Polyene synthesis: An efficient approach to the total synthesis of polyene natural product dihydroxrulin (1) is described. A novel, mild, direct organocatalytic IBX-mediated dehydrogenation process of simple alcohols to enals has been developed, which serves as a key step in the synthesis (see scheme).