RESUMO
In this work, we report an n-type metal-oxide-semiconductor (nMOS) inverter using chemical vapor deposition (CVD)-grown monolayer WS2 field-effect transistors (FETs). Our large-area CVD-grown monolayer WS2 FETs exhibit outstanding electrical properties including a high on/off ratio, small subthreshold swing, and excellent drain-induced barrier lowering. These are achieved by n-type doping using AlOx/Al2O3 and a double-gate structure employing high-k dielectric HfO2. Due to the superior subthreshold characteristics, monolayer WS2 FETs show high transconductance and high output resistance in the subthreshold regime, resulting in significantly higher intrinsic gain compared to conventional Si MOSFETs. Therefore, we successfully realize subthreshold operating monolayer WS2 nMOS inverters with extremely high gains of 564 and 2056 at supply voltage (VDD) of 1 and 2 V, respectively, and low power consumption of â¼2.3 pW·µm-1 at VDD = 1 V. In addition, the monolayer WS2 nMOS inverter is further expanded to the demonstration of logic circuits such as AND, OR, NAND, NOR logic gates, and SRAM. These findings suggest the potential of monolayer WS2 for high-gain and low-power logic circuits and validate the practical application in large areas.
RESUMO
Despite promising advancements in deep learning in medical domains, challenges still remain owing to data scarcity, compounded by privacy concerns and data ownership disputes. Recent explorations of distributed-learning paradigms, particularly federated learning, have aimed to mitigate these challenges. However, these approaches are often encumbered by substantial communication and computational overhead, and potential vulnerabilities in privacy safeguards. Therefore, we propose a self-supervised masked sampling distillation technique called MS-DINO, tailored to the vision transformer architecture. This approach removes the need for incessant communication and strengthens privacy using a modified encryption mechanism inherent to the vision transformer while minimizing the computational burden on client-side devices. Rigorous evaluations across various tasks confirmed that our method outperforms existing self-supervised distributed learning strategies and fine-tuned baselines.
RESUMO
Metal-oxide semiconductors (MOSs) have emerged as pivotal components in technology related to biosensors and bioelectronics. Detecting biomarkers in sweat provides a glimpse into an individual's metabolism without the need for sample preparation or collection steps. The distinctive attributes of this biosensing technology position it as an appealing option for biomedical applications beyond the scope of diagnosis and healthcare monitoring. This review encapsulates ongoing developments of cutting-edge biosensors based on MOSs. Recent advances in MOS-based biosensors for human sweat analyses are reviewed. Also discussed is the progress in sweat-based biosensing technologies to detect and monitor diseases. Next, system integration of biosensors is demonstrated ultimately to ensure the accurate and reliable detection and analysis of target biomarkers beyond individual devices. Finally, the challenges and opportunities related to advanced biosensors and bioelectronics for biomedical applications are discussed.
Assuntos
Técnicas Biossensoriais , Metais , Óxidos , Semicondutores , Suor , Técnicas Biossensoriais/instrumentação , Técnicas Biossensoriais/métodos , Humanos , Suor/química , Metais/química , Óxidos/química , Desenho de Equipamento , Biomarcadores/análiseRESUMO
The panoramic characteristics of human-machine interfaces (HMIs) have prompted the needs to update the biotechnology community with the recent trends, developments, and future research direction toward next-generation bioelectronics. Bioinspired materials are promising for integrating various bioelectronic devices to realize HMIs. With the advancement of scientific biotechnology, state-of-the-art bioelectronic applications have been extensively investigated to improve the quality of life by developing and integrating bioinspired nanoplatforms in HMIs. This review highlights recent trends and developments in the field of biotechnology based on bioinspired nanoplatforms by demonstrating recently explored materials and cutting-edge device applications. Section 1 introduces the recent trends and developments of bioinspired nanomaterials for HMIs. Section 2 reviews various flexible, wearable, biocompatible, and biodegradable nanoplatforms for bioinspired applications. Section 3 furnishes recently explored substrates as carriers for advanced nanomaterials in developing HMIs. Section 4 addresses recently invented biomimetic neuroelectronic, nanointerfaces, biointerfaces, and nano/microfluidic wearable bioelectronic devices for various HMI applications, such as healthcare, biopotential monitoring, and body fluid monitoring. Section 5 outlines designing and engineering of bioinspired sensors for HMIs. Finally, the challenges and opportunities for next-generation bioinspired nanoplatforms in extending the potential on HMIs are discussed for a near-future scenario. We believe this review can stimulate the integration of bioinspired nanoplatforms into the HMIs in addition to wearable electronic skin and health-monitoring devices while addressing prevailing and future healthcare and material-related problems in biotechnologies.
Assuntos
Materiais Biomiméticos , Nanoestruturas , Dispositivos Eletrônicos Vestíveis , Humanos , Qualidade de Vida , BiotecnologiaRESUMO
PURPOSE: We investigated the proportions of patients eligible for accelerated partial breast irradiation (APBI) among those with pT1-2N0 breast cancer, based on the criteria set by the American Society for Radiation Oncology (ASTRO), the Groupe Européen de Curiethérapie and the European Society for Radiotherapy and Oncology (GEC-ESTRO), the American Brachytherapy Society (ABS), and the American Society of Breast Surgeons (ASBS). Additionally, we analyzed the rate of APBI utilization among eligible patients. MATERIALS AND METHODS: Patients diagnosed with pT1-2N0 breast cancer in 2019 were accrued in four tertiary medical centers in Korea. All patients had undergone breast conserving surgery followed by radiotherapy, either whole breast irradiation or APBI. To determine which guideline best predicts the use of APBI in Korea, the F1 score and Matthews Correlation Coefficient (MCC) were determined for each guideline. RESULTS: A total of 1,251 patients were analyzed, of whom 196 (15.7%) underwent APBI. The percentages of eligible patients identified by the ASTRO, GEC-ESTRO, ABS, and ASBS criteria were 13.7%, 21.0%, 50.5%, and 63.5%, respectively. APBI was used to treat 54.4%, 37.2%, 27.1%, and 23.7% of patients eligible by the ASTRO, GEC-ESTRO, ABS, and ASBS criteria, respectively. The ASTRO guideline exhibited the highest F1 score (0.76) and MCC (0.67), thus showing the best prediction of APBI utilization in Korea. CONCLUSION: The proportion of Korean breast cancer patients who are candidates for APBI is substantial. The actual rate of APBI utilization among eligible patients may suggest there is a room for risk-stratified optimization in offering radiation therapy.
Assuntos
Braquiterapia , Neoplasias da Mama , Radioterapia (Especialidade) , Humanos , Estados Unidos , Feminino , Neoplasias da Mama/radioterapia , Neoplasias da Mama/cirurgia , Mastectomia Segmentar , República da CoreiaRESUMO
BACKGROUND/AIM: Chronic kidney disease (CKD) is one of the most common causes of mortality in wild non-domestic felidae. The molecular mechanism regulating renal fibrosis in nephropathy is not fully understood especially in the felidae. This study aimed to elucidate senescence marker protein 30 (SMP30) expression patterns and its relationship with epithelial-mesenchymal transition (EMT) by immunostaining in two necropsied Siberian tigers (Panthera tigris altaica) with CKD. MATERIALS AND METHODS: Two kidney samples from male Siberian tigers were fixed and tissue sections were stained for histopathological assay. RESULTS: In CKD, renal tubular epithelial cells lost their tubular structures surrounded by severe interstitial fibrosis and were detached from the basement membrane. These damaged cells resembled the morphology of mesenchymal cells and showed much lower SMP30 expression compared with intact tubular epithelial cells. These cells also expressed vimentin, which is specifically expressed by mesenchymal cells, and through double staining, it was observed that vimentin was expressed in the tubular epithelial cells where SMP30 was not expressed. In addition, double-positive expression of pan-cytokeratin (pan-CK) and vimentin was found in damaged epithelial cells with mesenchymal features. CONCLUSION: We demonstrated possible evidence to understand the role of SMP30 as a new pivotal factor and the possibility of decreased SMP30 as a potential indicator of EMT at the end stage of CKD.
Assuntos
Felidae , Insuficiência Renal Crônica , Tigres , Animais , Masculino , Humanos , Vimentina , Rim , Insuficiência Renal Crônica/genética , Transição Epitelial-Mesenquimal/genética , FibroseRESUMO
The escalating demand for artificial intelligence (AI) systems that can monitor and supervise human errors and abnormalities in healthcare presents unique challenges. Recent advances in vision-language models reveal the challenges of monitoring AI by understanding both visual and textual concepts and their semantic correspondences. However, there has been limited success in the application of vision-language models in the medical domain. Current vision-language models and learning strategies for photographic images and captions call for a web-scale data corpus of image and text pairs which is not often feasible in the medical domain. To address this, we present a model named medical cross-attention vision-language model (Medical X-VL), which leverages key components to be tailored for the medical domain. The model is based on the following components: self-supervised unimodal models in medical domain and a fusion encoder to bridge them, momentum distillation, sentencewise contrastive learning for medical reports, and sentence similarity-adjusted hard negative mining. We experimentally demonstrated that our model enables various zero-shot tasks for monitoring AI, ranging from the zero-shot classification to zero-shot error correction. Our model outperformed current state-of-the-art models in two medical image datasets, suggesting a novel clinical application of our monitoring AI model to alleviate human errors. Our method demonstrates a more specialized capacity for fine-grained understanding, which presents a distinct advantage particularly applicable to the medical domain.
Assuntos
Inteligência Artificial , Radiologia , Humanos , Radiografia , Aprendizagem , IdiomaRESUMO
Automatic Speech Recognition (ASR) is a technology that converts spoken words into text, facilitating interaction between humans and machines. One of the most common applications of ASR is Speech-To-Text (STT) technology, which simplifies user workflows by transcribing spoken words into text. In the medical field, STT has the potential to significantly reduce the workload of clinicians who rely on typists to transcribe their voice recordings. However, developing an STT model for the medical domain is challenging due to the lack of sufficient speech and text datasets. To address this issue, we propose a medical-domain text correction method that modifies the output text of a general STT system using the Vision Language Pre-training (VLP) method. VLP combines textual and visual information to correct text based on image knowledge. Our extensive experiments demonstrate that the proposed method offers quantitatively and clinically significant improvements in STT performance in the medical field. We further show that multi-modal understanding of image and text information outperforms single-modal understanding using only text information.
Assuntos
Fala , Voz , Humanos , IdiomaRESUMO
Due to the short storage period, large quantities of platelet concentrate (PC) are expiring. The expired PC cannot be injected into a blood vessel, but the activity of bioactive molecules, especially growth factors, is still preserved. In this paper, we organized a process to obtain a growth factor-rich bioproduct for use as a supplement in human cell culture by optimizing freezing, thawing, and sterilization conditions. Each unit of PC displayed visual differences, diverse biochemical values, and growth factor concentrations. To minimize lot-to-lot variation, we pooled a minimum of 10 PC units. The concentrations of growth factors were maximized through five freeze-thaw cycles for 12 h at -80 °C for freezing and for 5 min at 36 °C for thawing. We used a cell strainer with 40 µm pores, followed by a 0.45 µm filter and a 0.22 µm filter sequentially to sterilize the bioproduct with minimizing loss. The obtained growth factors remained stable for 4-6 h at room temperature (23 °C), 24 h at 4 °C, and 12 months at -80 °C. Cellular responses to the growth factor-rich bioproduct were tested with primary human renal proximal tubule epithelial cells. The cells exhibited a significantly increased growth rate, compared to the fetal bovine serum (FBS)-treated control group. The cells maintained their characteristic cuboidal shape, and stem cells and renal progenitor cells also preserved their genetic characteristics during culture. Therefore, the growth factor-rich bioproduct isolated from expired PC through our process can be used as a medium supplement to replace FBS in human cell culture for clinical application.
RESUMO
Patients with acute coronary syndromes caused by plaque erosion might be managed conservatively without stenting. Currently, the diagnosis of plaque erosion requires an invasive imaging procedure. We sought to develop a deep learning (DL) model that enables an accurate diagnosis of plaque erosion using coronary computed tomography angiography (CTA). A total of 532 CTA scans from 395 patients were used to develop a DL model: 426 CTA scans from 316 patients for training and internal validation, and 106 separate scans from 79 patients for validation. Momentum Distillation-enhanced Composite Transformer Attention (MD-CTA), a novel DL model that can effectively process the entire set of CTA scans to diagnose plaque erosion, was developed. The novel DL model, compared to the convolution neural network, showed significantly improved AUC (0.899 [0.841-0.957] vs. 0.724 [0.622-0.826]), sensitivity (87.1 [70.2-96.4] vs. 71.0 [52.0-85.8]), and specificity (85.3 [75.3-92.4] vs. 68.0 [56.2-78.3]), respectively, for the patient-level prediction. Similar results were obtained at the slice-level prediction AUC (0.897 [0.890-0.904] vs. 0.757 [0.744-0.770]), sensitivity (82.2 [79.8-84.3] vs. 68.9 [66.2-71.6]), and specificity (80.1 [79.1-81.0] vs. 67.3 [66.3-68.4]), respectively. This newly developed DL model enables an accurate CT diagnosis of plaque erosion, which might enable cardiologists to provide tailored therapy without invasive procedures.Clinical Trial Registration: http://www.clinicaltrials.gov , NCT04523194.
Assuntos
Doença da Artéria Coronariana , Aprendizado Profundo , Placa Aterosclerótica , Humanos , Doença da Artéria Coronariana/diagnóstico por imagem , Angiografia Coronária/métodos , Placa Aterosclerótica/diagnóstico por imagem , Tomografia Computadorizada por Raios X , Vasos Coronários/diagnóstico por imagemRESUMO
Whole-brain imaging is important for understanding brain functions through deciphering tissue structures, neuronal circuits, and single-neuron tracing. Thus, many clearing methods have been developed to acquire whole-brain images or images of three-dimensional thick tissues. However, there are several limitations to imaging whole-brain volumes, including long image acquisition times, large volumes of data, and a long post-image process. Based on these limitations, many researchers are unsure about which light microscopy is most suitable for imaging thick tissues. Here, we compared fast-confocal microscopy with light-sheet fluorescence microscopy for whole-brain three-dimensional imaging, which can acquire images the fastest. To compare the two types of microscopies for large-volume imaging, we performed tissue clearing of a whole mouse brain, and changed the sample chamber and low- magnification objective lens and modified the sample holder of a light-sheet fluorescence microscope. We found out that light-sheet fluorescence microscopy using a 2.5× objective lens possesses several advantages, including saving time, large-volume image acquisitions, and high Z-resolution, over fast-confocal microscopy, which uses a 4× objective lens. Therefore, we suggest that light-sheet fluorescence microscopy is suitable for whole mouse brain imaging and for obtaining high-resolution three-dimensional images.
RESUMO
Background: The anti-platelet activity of the saponin fraction of Korean Red Ginseng has been widely studied. The saponin fraction consists of the panaxadiol fraction (PDF) and panaxatriol fraction (PTF); however, their anti-platelet activity is yet to be compared. Our study aimed to investigate the potency of anti-platelet activity of PDF and PTF and to elucidate how well they retain their anti-platelet activity via different administration routes. Methods: For ex vivo studies, Sprague-Dawley rats were orally administered 250 mg/kg PDF and PTF for 7 consecutive days before blood collection via cardiac puncture. Platelet aggregation was conducted after isolation of the washed platelets. For in vitro studies, washed platelets were obtained from Sprague-Dawley rats. Collagen and adenosine diphosphate (ADP) were used to induce platelet aggregation. Collagen was used as an agonist for assaying adenosine triphosphate release, thromboxane B2, serotonin, cyclic adenosine monophosphate, and cyclic guanosine monophosphate (cGMP) release. Results: When treated ex vivo, PDF not only inhibited ADP and collagen-induced platelet aggregation, but also upregulated cGMP levels and reduced platelet adhesion to fibronectin. Furthermore, it also inhibited Akt phosphorylation induced by collagen treatment. Panaxadiol fraction did not exert any anti-platelet activity in vitro, whereas PTF exhibited potent anti-platelet activity, inhibiting ADP, collagen, and thrombin-induced platelet aggregation, but significantly elevated levels of cGMP. Conclusion: Our study showed that in vitro and ex vivo PDF and PTF treatments exhibited different potency levels, indicating possible metabolic conversions of ginsenosides, which altered the content of ginsenosides capable of preventing platelet aggregation.
RESUMO
Waste plastics are degraded into microplastics (MPs), which are easily accumulated in the human body through digestive tracts, via the food chain. Alcohol is a widely consumed chemical throughout the world with the ability to alter the intestinal barrier. For this reason, this study was aimed to investigate exact relevance between alcohol consumption and organ distributions of MPs in an ethanol feeding animal model characterized by disrupted intestinal mucosal barriers. In this study, C57BL/6 mice were separated into control, control + MP, ethanol (EtOH), and EtOH + MP groups. Mice in the EtOH group ingested a Lieber-DeCarli diet containing EtOH. Mice in the MP groups ingested 0.1 mg/kg fluorophore polymerized polystyrene microplastics via oral gavage polystyrene MPs via oral gavage. The EtOH + MP group showed higher MP accumulation in the liver than the control + MP group. The same pattern was observed in the intestines, spleen, and brain. This pattern was more prominent in the intestines, with the EtOH + MP group showing the most severe damage due to EtOH ingestion. This result suggests that the intestinal mucosa disruption caused by EtOH ingestion exacerbates MP accumulation in the organs. Moreover, hepatic steatosis was more severe in the EtOH + MP group than in the EtOH group, suggesting the secondary manifestation mediated by MP accumulation. This study reports a novel MP accumulation pattern in the body by providing novel insights into alcohol-induced gut permeability and microplastics toxicity from the perspective of gut-liver axis.
RESUMO
OBJECTIVE: Detection of pneumoperitoneum using abdominal radiography, particularly in the supine position, is often challenging. This study aimed to develop and externally validate a deep learning model for the detection of pneumoperitoneum using supine and erect abdominal radiography. MATERIALS AND METHODS: A model that can utilize "pneumoperitoneum" and "non-pneumoperitoneum" classes was developed through knowledge distillation. To train the proposed model with limited training data and weak labels, it was trained using a recently proposed semi-supervised learning method called distillation for self-supervised and self-train learning (DISTL), which leverages the Vision Transformer. The proposed model was first pre-trained with chest radiographs to utilize common knowledge between modalities, fine-tuned, and self-trained on labeled and unlabeled abdominal radiographs. The proposed model was trained using data from supine and erect abdominal radiographs. In total, 191212 chest radiographs (CheXpert data) were used for pre-training, and 5518 labeled and 16671 unlabeled abdominal radiographs were used for fine-tuning and self-supervised learning, respectively. The proposed model was internally validated on 389 abdominal radiographs and externally validated on 475 and 798 abdominal radiographs from the two institutions. We evaluated the performance in diagnosing pneumoperitoneum using the area under the receiver operating characteristic curve (AUC) and compared it with that of radiologists. RESULTS: In the internal validation, the proposed model had an AUC, sensitivity, and specificity of 0.881, 85.4%, and 73.3% and 0.968, 91.1, and 95.0 for supine and erect positions, respectively. In the external validation at the two institutions, the AUCs were 0.835 and 0.852 for the supine position and 0.909 and 0.944 for the erect position. In the reader study, the readers' performances improved with the assistance of the proposed model. CONCLUSION: The proposed model trained with the DISTL method can accurately detect pneumoperitoneum on abdominal radiography in both the supine and erect positions.
Assuntos
Aprendizado Profundo , Humanos , Estudos Retrospectivos , Radiografia Abdominal , Radiografia , Aprendizado de Máquina Supervisionado , Radiografia Torácica/métodosRESUMO
Background: The incidence and clinical importance of nonalcoholic fatty liver disease (NAFLD) has emerged. However, effective therapeutic strategies for NAFLD have yet to be found. Panax ginseng (P. ginseng) is a traditional herb in Eastern Asia with therapeutic effects in many chronic disorders. However, the precise effects of ginseng extract on NAFLD are currently unknown. In present study, the therapeutic effects of Rg3-enriched red ginseng extract (Rg3-RGE) on the progression of NAFLD were explored. Methods: Twelve-week-old C57BL/6 male mice were fed a chow or western diet supplemented with high sugar water solution with or without Rg3-RGE. Histopathology, immunohistochemistry, immunofluorescence, serum biochemistry, western blot analysis, and quantitative RT-PCR were used for in vivo experiment. Conditionally immortalized human glomerular endothelial cell (CiGEnC) and primary liver sinusoidal endothelial cells (LSECs) were used for in vitro experiments. Results: Eight weeks of Rg3-RGE treatment significantly attenuated the inflammatory lesions of NAFLD. Furthermore, Rg3-RGE inhibited the inflammatory infiltrate in liver parenchyma and the expression of adhesive molecules to LSECs. Moreover, the Rg3-RGE exhibited similar patterns on the in vitro assays. Conclusion: The results demonstrate that Rg3-RGE treatment ameliorates NAFLD progression by inhibiting chemotaxis activities in LSECs.
RESUMO
Healed coronary plaques, morphologically characterized by a layered phenotype, are signs of previous plaque destabilization and healing. Recent optical coherence tomography (OCT) studies demonstrated that layered plaque is associated with higher levels of local and systemic inflammation and rapid plaque progression. However, the diagnosis of layered plaque needs expertise in OCT image analysis and is susceptible to inter-observer variability. We developed a deep learning (DL) model for an accurate diagnosis of layered plaque. A Visual Transformer (ViT)-based DL model that integrates information from adjacent frames emulating the cardiologists who review consecutive OCT frames to make a diagnosis was developed and compared with the standard convolutional neural network (CNN) model. A total of 237,021 cross-sectional OCT images from 581 patients collected from 8 sites were used for training and internal validation, and 65,394 images from 292 patients collected from another site were used for external validation. In the five-fold cross-validation, the ViT-based model provided better performance (area under the curve [AUC]: 0.860; 95% confidence interval [CI]: 0.855-0.866) than the standard CNN-based model (AUC: 0.799; 95% CI: 0.792-0.805). The ViT-based model (AUC: 0.845; 95% CI: 0.837-0.853) also surpassed the standard CNN-based model (AUC: 0.791; 95% CI: 0.782-0.800) in the external validation. The ViT-based DL model can accurately diagnose a layered plaque, which could help risk stratification for cardiac events.
Assuntos
Aprendizado Profundo , Infarto do Miocárdio , Placa Aterosclerótica , Humanos , Estudos Transversais , Placa Aterosclerótica/diagnóstico por imagem , Redes Neurais de Computação , Tomografia de Coerência Óptica/métodosRESUMO
BACKGROUND: Most extramedullary plasmacytomas (EMPs) aresolitary and located in the head and neck region. They may also occur in the visceral parts of the body. OBJECTIVES: Here, we report a case of oral EMP followed by neoplastic plasma cell metastasis to both kidneys in a neutered male Pomeranian. METHODS: Oral plasmacytoma recurred 11 months aftersurgical removal of an oral mass and partial maxillectomy was performed. Eighteen months after partial maxillectomy, neoplastic masses were detected in both kidneys on computed tomography. The dog died 12 months after detection of bilateral kidney neoplasms. The resected neoplastic masses were routinely processed for histopathological observation and immunohistochemistry against pan-cytokeratin, desmin, CD3, and MUM-1. RESULTS: The recurred mass mainly consisted of well-differentiated plasma cells and contained a small portion of aggressive cells with malignant features. Monoclonal gammopathy was not observed on serumelectrophoresis performed to exclude multiple myeloma. The mass was composed of plasma cells with high nuclear pleomorphism and abundant mitotic figures. The neoplasm stained positive for MUM-1 with a more aggressive morphology than in oral EMP. CONCLUSION: Based on serum biomarker and pathological observations, a diagnosis of recurrence and metastasis of oral-to-renal EMP was established. To the best of our knowledge, metastasis of oral EMP into the bilateral kidneys, as described in the current case, has not been previously reported in dogs.
Assuntos
Doenças do Cão , Plasmocitoma , Masculino , Cães , Animais , Plasmocitoma/diagnóstico , Plasmocitoma/cirurgia , Plasmocitoma/veterinária , Boca/patologia , Tomografia Computadorizada por Raios X , Rim , Doenças do Cão/diagnóstico por imagem , Doenças do Cão/cirurgiaRESUMO
Selective etching of silicon oxide (SiO2) against silicon (Si) using anhydrous hydrogen fluoride (HF) vapor has been used for semiconductor device fabrication. We studied the underlying mechanism of the selective etching by density functional theory (DFT) calculation. We constructed surface slab models of SiO2 or Si with different degrees of fluorination and simulated the four steps of fluorination. The calculations show relatively low activation energies of 0.72-0.79 eV for the four steps of fluorination of SiO2, which are close to â¼0.69 eV observed in the experiment. The four-membered ring structure of -Si-O-H-F- in all transition states stabilized the system, resulting in relatively low activation energies. Thus, continuous etching of SiO2 by HF is plausible at near-room temperature. In contrast, the fluorinations of Si showed relatively high activation energies ranging from 1.22 to 1.56 eV due to the less stable transition state geometries. Thus, negligible etching of silicon by HF is expected by the near-room temperature process. Our calculation results explain well the experimental observation of the selective etching of SiO2 against Si by HF vapor.
RESUMO
In this paper, we introduce a new blue-emitting material, CuCrS2/ZnS QDs (CCS QDs). To obtain bright and stable photoluminescent probes, we prepared a core/shell structure; the synthesis was conducted in a one-pot system, using 1-dodecanethiol as a sulfur source and co-ligand. The CCS QDs exhibited a semi-spherical colloidal nanocrystalline shape with an average diameter of 9.0 nm and ZnS shell thickness of 1.6 nm. A maximum photoluminescence emission peak (PL max) was observed at 465 nm with an excitation wavelength of 400 nm and PLQY was 5% at an initial [Cr3+]/[Cu+] molar ratio of one in the core synthesis. With an off-stoichiometric modification for band gap engineering, the CCS QDs exhibited slightly blue-shifted PL emission spectra and PLQY was 10% with an increase in initial molar ratio of 2.0 (462 nm PL max). However, when the initial molar ratio exceeded two, the CCS QDs exhibited a lower photoluminescence quantum yield of 4.5% with 461 nm of PL max at the initial molar ratio of four due to the formation of non-emissive Cr2S3 nanoflakes.
RESUMO
The widespread application of artificial intelligence in health research is currently hampered by limitations in data availability. Distributed learning methods such as federated learning (FL) and split learning (SL) are introduced to solve this problem as well as data management and ownership issues with their different strengths and weaknesses. The recent proposal of federated split task-agnostic (F eSTA) learning tries to reconcile the distinct merits of FL and SL by enabling the multi-task collaboration between participants through Vision Transformer (ViT) architecture, but they suffer from higher communication overhead. To address this, here we present a multi-task distributed learning using ViT with random patch permutation, dubbed p -F eSTA. Instead of using a CNN-based head as in F eSTA, p -F eSTA adopts a simple patch embedder with random permutation, improving the multi-task learning performance without sacrificing privacy. Experimental results confirm that the proposed method significantly enhances the benefit of multi-task collaboration, communication efficiency, and privacy preservation, shedding light on practical multi-task distributed learning in the field of medical imaging.