Browse Articles

An explainable deep learning model for diabetic foot ulcer classification using swin transformer and efficient multi-scale attention-driven network

Abstract Diabetic Foot Ulcer (DFU) is a severe complication of diabetes mellitus, resulting in significant health and socio-economic challenges for the diagnosed individual. Severe cases of DFU can lead to lower limb amputation in diabetic patients, making their diagnosis a complex and costly process that poses challenges for medical professionals. Manual identification of DFU is particularly difficult due to their diverse visual characteristics, leading to multiple cases going undiagnosed. To address this challenge, Deep Learning (DL) methods offer an efficient and automated approach to facilitate timely treatment and improve patient outcomes. This research proposes a novel feature fusion-based model that incorporates two parallel tracks for efficient feature extraction. The first track utilizes the Swin transformer, which captures long-range dependencies by employing shifted windows and self-attention mechanisms. The second track involves the Efficient Multi-Scale Attention-Driven Network (EMADN), which leverages Light-weight Multi-scale Deformable Shuffle (LMDS) and Global Dilated Attention (GDA) blocks to extract local features efficiently. These blocks dynamically adjust kernel sizes and leverage attention modules, enabling effective feature extraction. To the best of our knowledge, this is the first work reporting the findings of a dual track architecture for DFU classification, leveraging Swin transformer and EMADN networks. The obtained feature maps from both the networks are concatenated and subjected to shuffle attention for feature refinement at a reduced computational cost. The proposed work also incorporates Grad-CAM-based Explainable Artificial Intelligence (XAI) to visualize and interpret the decision making of the network. The proposed model demonstrated better performance on the DFUC-2021 dataset, surpassing existing works and pre-trained CNN architectures with an accuracy of 78.79% and a macro F1-score of 80%.

Seasonal recurrence and modular assembly of an Arctic pelagic marine microbiome

Abstract Deciphering how microbial communities are shaped by environmental variability is fundamental for understanding the structure and function of ocean ecosystems. While seasonal environmental gradients have been shown to structure the taxonomic dynamics of microbiomes over time, little is known about their impact on functional dynamics and the coupling between taxonomy and function. Here, we demonstrate annually recurrent, seasonal structuring of taxonomic and functional dynamics in a pelagic Arctic Ocean microbiome by combining autonomous samplers and in situ sensors with long-read metagenomics and SSU ribosomal metabarcoding. Specifically, we identified five temporal microbiome modules whose succession within each annual cycle represents a transition across different ecological states. For instance, Cand . Nitrosopumilus, Syndiniales, and the machinery to oxidise ammonia and reduce nitrite are signatures of early polar night, while late summer is characterised by Amylibacter and sulfur compound metabolism. Leveraging metatranscriptomes from Tara Oceans , we also demonstrate the consistency in functional dynamics across the wider Arctic Ocean during similar temporal periods. Furthermore, the structuring of genetic diversity within functions over time indicates that environmental selection pressure acts heterogeneously on microbiomes across seasons. By integrating taxonomic, functional and environmental information, our study provides fundamental insights into how microbiomes are structured under pronounced seasonal changes in understudied, yet rapidly changing polar marine ecosystems.

Sensory seeking and its influence on sustained attention performance in adult males with Autism Spectrum Condition

Getting the best of carbon bang for mangrove restoration buck

Crystal structure modulating performances for 213-nm GeO2 solar-blind photodetectors via DC reactive magnetron sputtering method

Structural basis for the ligand-dependent activation of heterodimeric AHR-ARNT complex

Impact of monthly air pollution and weather conditions on cardiorespiratory mortality in Portuguese Metropolitan Areas

Translaminar synchronous neuronal activity is required for columnar synaptic strengthening in the mouse neocortex

Dissociative experiences alter resting state functional connectivity after childhood abuse

Abstract Dissociative experiences commonly occur alongside adverse childhood experiences (ACE), yet research on their neurofunctional biomarkers has overlooked their unique association with dimensions of childhood abuse and neglect. We investigated interactions between dissociative experiences and childhood abuse, anticipating anti-correlations between the right-lateralized anterior middle frontal gyrus (raMFG) and the medial temporal lobe, as well as the temporal gyri. Examining resting-state functional connectivity in 91 participants with a history of ACE, we employed seed-to-voxel analyses seeding the raMFG. Multiple linear regression and post-hoc moderation/mediation models explored interactions and individual effects of dissociation and dimensions of ACE. The Dissociative Experiences Scale (DES) and Childhood Trauma Questionnaire (CTQ) quantified dissociation and dimensions of ACE. A DES by CTQ-A (childhood abuse) interaction predicted an anti-correlation between the raMFG and right hippocampus, moderated by CTQ-A. The CTQ revealed negative connectivity between the raMFG and right anterior cingulate cortex. CTQ-N (childhood neglect) indicated that both the right supplementary motor area and right insula related positively to the raMFG. Our findings underscore a distinct neural signature of childhood abuse-related dissociative experiences, potentially linked to dissociated memories.

Large-angle Lorentz Four-dimensional scanning transmission electron microscopy for simultaneous local magnetization, strain and structure mapping

Abstract Small adjustments in atomic configurations can significantly impact the magnetic properties of matter. Strain, for instance, can alter magnetic anisotropy and enable fine-tuning of magnetism. However, the effects of these changes on nanoscale magnetism remain largely unexplored. In particular, when strain fluctuates at the nanoscale, directly linking structural changes with magnetic behavior poses a substantial challenge. Here, we develop an approach, LA-Ltz-4D-STEM, to map structural information and magnetic fields simultaneously at the nanoscale. This approach opens avenues for an in-depth study of structure-property correlations of magnetic materials at the nanoscale. We applied LA-Ltz-4D-STEM to image strain, atomic packing, and magnetic fields simultaneously in a deformed amorphous ferromagnet with complex strain variations at the nanoscale. An anomalous magnetic configuration near shear bands, which reside in a magnetostatically high-energy state, was observed. By performing pixel-to-pixel correlation of the different physical quantities across a large field of view, a critical aspect for investigating industrial ferromagnetic materials, the magnetic moments were classified into two distinct groups: one influenced by magnetoelastic coupling and the other oriented by competition with magnetostatic energy.

Cost-effective adsorption of cationic dyes using ZnO nanorods supported by orange peel-derived carbon

Abstract Here, porous carbon (PC) and ZnO nanorods@PC (ZnO-NR@PC) composite derived from orange peel (OP) have been synthesized via a simple carbonization process. The prepared materials have been characterized by XRD, FT-IR, TEM, and BET analysis. The adsorptive properties of the prepared PC and ZnO-NR@PC composite have been investigated toward methylene blue (MB) and crystal violet (CV) cationic dyes from their aqueous solutions. The adsorption studies concluded that the maximum adsorption efficiency was achieved after 90 min in the basic conditions (pH = 10). Langmuir, Freundlich, Dubinin–Radushkevich (D-R), and Temkin non-linear isotherm models were applied to fit the experimental data. The adsorption of MB and CV dyes by the OP is fitted with the Freundlich model, and the adsorption of both dyes by the PC and the ZnO-NR@PC composite fitted with the Langmuir model. The estimated maximum adsorption capacity estimated from the adsorption of MB and CV by the ZnO-NR@PC composite was 74.45 and 74.89 mg/g, respectively. The calculated adsorption free energy from D-R and Temkin models indicates the adsorption of MB, and CV dye molecules by the OP, PC, and ZnO-NR@PC composite may be physical. The kinetic studies revealed the adsorption of MB and CV dyes onto the OP, PC and ZnO-NR@PC composite fitted with the pseudo-second-order model. On the otherhand, the thermodynamic studies confirmed the adsorption of MB, and CV dyes onto ZnO-NR@PC composite is an endothermic and spontaneous process. Furthermore, the prepared materials displayed high adsorption stability with an overall removal efficiency of about 90% after five cycles. The mechanism of MB and CV dyes by the ZnO-NR@PC composite is proposed to be controlled by electrostatic bonding, π-π interactions, and ion exchange. The results indicated the potential ability of OP-derived porous carbons as adsorbents for cationic dyes from aqueous media.

Structural insights into light-gating of potassium-selective channelrhodopsin

Author Correction: Normal tissue radioprotection by amifostine via Warburg-type effects

Proton-electron coupling and mixed conductivity in a hydrogen-bonded coordination polymer

Investigating risk factors for migraine in Syrian women: a cross-sectional case-control study

A self-powered soft triboelectric-electrohydrodynamic pump

Establishment of an animal model for monkeypox virus infection in dormice

Cloned airway basal progenitor cells to repair fibrotic lung through re-epithelialization

Isotemporal substitution of sedentary time with different physical activity intensities and sleep in obesity parameters across eight latin American countries

Diffusion-programmed catalysis in nanoporous material

Abstract In the realm of heterogeneous catalysis, the diffusion of reactants into catalytically active sites stands as a pivotal determinant influencing both turnover frequency and geometric selectivity in product formation. While accelerated diffusion of reactants can elevate reaction rates, it often entails a compromise in geometric selectivity. Porous catalysts, including metal-organic and covalent organic frameworks, confront formidable obstacles in regulating reactant diffusion rates. Consequently, the chemical functionality of the catalysts typically governs turnover frequency and geometric selectivity. This study presents an approach harnessing diffusion length to achieve improved selectivity and manipulation of reactant-active site residence time at active sites to augment reaction kinetics. Through the deployment of a thin film composed of a porous metal-organic framework catalyst, we illustrate how programming reactant diffusion within a cross-flow microfluidic catalytic reactor can concurrently amplify turnover frequency (exceeding 1000-fold) and enhance geometric selectivity ( ~ 2-fold) relative to conventional nano/microcrystals of catalyst in one-pot reactor. This diffusion-programed strategy represents a robust solution to surmount the constraints imposed by bulk nano/microcrystals of catalysts, marking advancement in the design of porous catalyst-driven organic reactions.