Utilizing deep learning for automated detection of oral lesions: A multicenter study.

OBJECTIVES: We aim to develop a YOLOX-based convolutional neural network model for the precise detection of multiple oral lesions, including OLP, OLK, and OSCC, in patient photos. MATERIALS AND METHODS: We collected 1419 photos for model development and evaluation, conducting both a comparative analysis to gauge the model's capabilities and a multicenter evaluation to assess its diagnostic aid, where 24 participants from 14 centers across the nation were invited. We further integrated this model into a mobile application for rapid and accurate diagnostics. RESULTS: In the comparative analysis, our model overperformed the senior group (comprising three most experienced experts with more than 10 years of experience) in macro-average recall (85 % vs 77.5 %), precision (87.02 % vs 80.29 %), and specificity (95 % vs 92.5 %). In the multicenter model-assisted diagnosis evaluation, the dental, general, and community hospital groups showed significant improvement when aided by the model, reaching a level comparable to the senior group, with all macro-average metrics closely aligning or even surpassing with those of the latter (recall of 78.67 %, 74.72 %, 83.54 % vs 77.5 %, precision of 80.56 %, 76.42 %, 85.15 % vs 80.29 %, specificity of 92.89 %, 91.57 %, 94.51 % vs 92.5 %). CONCLUSION: Our model exhibited a high proficiency in detection of oral lesions, surpassing the performance of highly experienced specialists. The model can also help specialists and general dentists from dental and community hospitals in diagnosing oral lesions, reaching the level of highly experienced specialists. Moreover, our model's integration into a mobile application facilitated swift and precise diagnostic procedures.

Evaluation of a convolutional neural network to identify scaphoid fractures on radiographs.

This study aimed to develop and evaluate a convolutional neural network for identifying scaphoid fractures on radiographs. A dataset of 1918 wrist radiographs (600 patients) was taken from an orthopaedic referral centre between 2010 to 2020. A YOLOv3 and a MobileNetV3 convolutional neural network were trained for scaphoid detection and fracture classification, respectively. The diagnostic performance of the convolutional neural network was compared with the majority decision of four hand surgeons. The convolutional neural network achieved a sensitivity of 82% and specificity of 94%, with an area under the receiver operating characteristic of 92%, whereas the surgeons achieved a sensitivity of 76% and specificity of 96%. The comparison indicated that the convolutional neural network's performance was similar to the majority vote of surgeons. It further revealed that convolutional neural network could be used in identifying scaphoid fractures on radiographs reliably, and has potential to achieve the expert-level performance.Level of evidence: III.

SISSO-assisted prediction and design of mechanical properties of porous graphene with a uniform nanopore array.

Mechanical properties of porous graphene can be effectively tuned by tailoring the nanopore arrangement. Knowledge of the relationship between the porous structure and overall mechanical properties is thus essential for the wide potential applications, and the existing challenge is to efficiently predict and design the mechanical properties of porous graphene due to the diverse nanopore arrangements. In this work, we report on how the SISSO (Sure Independence Screening and Sparsifying Operator) algorithm can be applied to build a bridge between the mechanical properties of porous graphene and the uniform nanopore array. We first construct a database using the strength and work of fracture calculated by large-scale molecular dynamics simulations. Then the SISSO algorithm is adopted to train a predictive model and automatically derive the optimal fitting formulae which explicitly describe the nonlinear structure-property relationships. These expressions not only enable the direct and accurate prediction of targeted properties, but also serve as a convenient and portable tool for inverse design of the porous structure. Compared with other forecasting methods including several popular machine learning algorithms, the SISSO algorithm shows its advantages in both accuracy and convenience.

Armchair Janus MoSSe Nanoribbon with Spontaneous Curling: A First-Principles Study.

Based on density functional theory, we theoretically investigate the electronic structures of free-standing armchair Janus MoSSe nanoribbons (A-MoSSeNR) with width up to 25.5 nm. The equilibrium structures of nanoribbons with spontaneous curling are obtained by energy minimization in molecular dynamics (MD). The curvature is 0.178 nm-1 regardless of nanoribbon width. Both finite element method and analytical solution based on continuum theory provide qualitatively consistent results for the curling behavior, reflecting that relaxation of intrinsic strain induced by the atomic asymmetry acts as the driving force. The non-edge bandgap of curled A-MoSSeNR reduces faster with the increase of width compared with planar nanoribbons. It can be observed that the real-space wave function at the non-edge VBM is localized in the central region of the curled nanoribbon. When the curvature is larger than 1.0 nm-1, both edge bandgap and non-edge bandgap shrink with the further increase of curvature. Moreover, we explore the spontaneous curling and consequent sewing process of nanoribbon to form nanotube (Z-MoSSeNT) by MD simulations. The spontaneously formed Z-MoSSeNT with 5.6 nm radius possesses the lowest energy. When radius is smaller than 0.9 nm, the bandgap of Z-MoSSeNT drops rapidly as the radius decreases. We expect the theoretical results can help build the foundation for novel nanoscale devices based on Janus TMD nanoribbons.

Lattice-distorted lithiation behavior of a square phase Janus MoSSe monolayer for electrode applications.

Janus transition metal dichalcogenides with unique physical properties have recently attracted increasing research interest for their energy and catalytic applications. In this paper, we investigate the lithiation behavior of a square phase Janus MoSSe monolayer (1S-MoSSe) using first-principles calculations. Computational results show that a single Li atom energetically prefers to adsorb on the central site of the octagonal ring (O site) and on the S-layer side of 1S-MoSSe. The predicted energy barriers for Li diffusion are surface dependent and in the range of 0.33 to 0.51 eV, indicating the acceptable Li migration kinetics on 1S-MoSSe in comparison with other 2D TMD materials. Further thermodynamic analysis demonstrates that Li adsorption on 1S-MoSSe is energetically stable up to a Li concentration of x = 1.0, above which the lithiation process becomes unstable with a negative charging potential. Phonon calculations also confirm that Li adsorption (0.25 ≤ x ≤ 0.75) results in the lattice distortion of 1S-MoSSe in order to suppress the structural instability of the lithiated monolayer 1S-Li x MoSSe with imaginary phonon frequencies. The less symmetric nature of 1S-MoSSe is believed to destabilize Li adsorption at much smaller x than 1H-MoSSe does, regardless of the higher dipole moment of 1S-MoSSe. This computational study provides a fundamental understanding of the electrochemical performance of 1S-MoSSe, as well as useful insight into the material design of Janus TMD anodes for Li-ion batteries.

Melt Electrospun Reduced Tungsten Oxide /Polylactic Acid Fiber Membranes as a Photothermal Material for Light-Driven Interfacial Water Evaporation.

The development of efficient photothermal materials is the most important issue in solar water evaporation. In this work, melt electrospun reduced tungsten oxide/polylactic acid (WO2.72/PLA) fiber membranes were successfully prepared with improved near-infrared (NIR) photothermal conversion properties owing to strong NIR photoabsorption by the metal oxide. WO2.72 powder nanoparticles were incorporated into PLA matrix by melt processing, following which the composites were extruded into wires using a single screw extruder. Subsequently, fiber membranes were prepared from the extruded wire of the WO2.72/PLA composite by melt electrospinning, which is a cost-effective technique that can produce fiber membranes without the addition of environmentally unfriendly chemicals. The melt electrospun WO2.72/PLA fiber membranes, floatable on water due to surface hydrophobicity, were systematically designed for, and applied to, vapor generation based on the interfacial concept of solar heating. With the photothermal WO2.72/PLA fiber membrane containing 7 wt % WO2.72 nanoparticles, the water evaporation efficiency was reached 81.39%, which is higher than that for the pure PLA fiber membrane and bulk water. Thus, this work contributes to the development of novel photothermal fiber membranes in order to enhance light-driven water evaporation performance for potential applications in the fields of water treatment and desalination.

Induction of Mitochondrial Dysfunction and Oxidative Damage by Antibiotic Drug Doxycycline Enhances the Responsiveness of Glioblastoma to Chemotherapy.

BACKGROUND Inducing mitochondrial dysfunction has been recently demonstrated to be an alternative therapeutic strategy for cancer treatment. Doxycycline is an antibiotic that has been shown to have anti-cancer activities in various cancers by way of targeting mitochondria. In this work, we examined whether doxycycline can be repurposed for glioblastoma treatment. MATERIAL AND METHODS The effects of doxycycline on the growth, survival, and mitochondrial metabolisms of glioblastoma were investigated. The efficacy of a combination of doxycycline with temozolomide was examined using xenograft mouse model in total number of 40 mice. RESULTS Doxycycline targeted glioblastoma cell lines, regardless of their origin, through inhibiting growth and inducing cell death, accompanied by a significant decrease in proliferating cell nuclear antigen (PCNA) and increase in cleaved caspase-3. In addition, doxycycline significantly sensitized glioblastoma cell response to temozolomide in vitro and in vivo. Mechanistically, doxycycline disrupted mitochondrial functions through decreasing mitochondrial membrane potential and mitochondrial respiration. Inducing mitochondrial dysfunctions by using doxycycline led to energy crisis, oxidative stress, and damage as shown by the decreased levels of ATP and the elevated levels of mitochondrial superoxide, intracellular ROS, 8-OHdG, protein carbonylation, and lipid peroxidation. An antioxidant N-acetyl-L-cysteine (NAC) significantly abolished the anti-proliferative and pro-apoptotic effects of doxycycline, demonstrating that doxycycline acts on glioblastoma via inducing oxidative stress. CONCLUSIONS In our study, we show that the antibiotic doxycycline is effective in targeting glioblastoma through inducing mitochondrial dysfunctions and oxidative stress. Our work also demonstrated the importance of mitochondrial metabolism in glioblastoma.