RFMiD: Retinal Image Analysis for multi-Disease Detection challenge.

In the last decades, many publicly available large fundus image datasets have been collected for diabetic retinopathy, glaucoma, and age-related macular degeneration, and a few other frequent pathologies. These publicly available datasets were used to develop a computer-aided disease diagnosis system by training deep learning models to detect these frequent pathologies. One challenge limiting the adoption of a such system by the ophthalmologist is, computer-aided disease diagnosis system ignores sight-threatening rare pathologies such as central retinal artery occlusion or anterior ischemic optic neuropathy and others that ophthalmologists currently detect. Aiming to advance the state-of-the-art in automatic ocular disease classification of frequent diseases along with the rare pathologies, a grand challenge on "Retinal Image Analysis for multi-Disease Detection" was organized in conjunction with the IEEE International Symposium on Biomedical Imaging (ISBI - 2021). This paper, reports the challenge organization, dataset, top-performing participants solutions, evaluation measures, and results based on a new "Retinal Fundus Multi-disease Image Dataset" (RFMiD). There were two principal sub-challenges: disease screening (i.e. presence versus absence of pathology - a binary classification problem) and disease/pathology classification (a 28-class multi-label classification problem). It received a positive response from the scientific community with 74 submissions by individuals/teams that effectively entered in this challenge. The top-performing methodologies utilized a blend of data-preprocessing, data augmentation, pre-trained model, and model ensembling. This multi-disease (frequent and rare pathologies) detection will enable the development of generalizable models for screening the retina, unlike the previous efforts that focused on the detection of specific diseases.

Broadening horizons: the role of ferroptosis in polycystic ovary syndrome.

Polycystic ovarian syndrome (PCOS) is a common heterogeneous reproductive endocrine metabolic disorder in women of reproductive age characterized by clinical and biochemical hyperandrogenemia, ovulation disorders, and polycystic ovarian morphology. Ferroptosis is a novel type of cell death driven by iron accumulation and lipid peroxidation. Ferroptosis plays a role in maintaining redox balance, iron metabolism, lipid metabolism, amino acid metabolism, mitochondrial activity, and many other signaling pathways linked to diseases. Iron overload is closely related to insulin resistance, decreased glucose tolerance, and the occurrence of diabetes mellitus. There is limited research on the role of ferroptosis in PCOS. Patients with PCOS have elevated levels of ferritin and increased reactive oxygen species in ovarian GCs. Studying ferroptosis in PCOS patients is highly important for achieving personalized treatment. This article reviews the progress of research on ferroptosis in PCOS, introduces the potential connections between iron metabolism abnormalities and oxidative stress-mediated PCOS, and provides a theoretical basis for diagnosing and treating PCOS.

Ocular image-based deep learning for predicting refractive error: A systematic review.

Background: Uncorrected refractive error is a major cause of vision impairment worldwide and its increasing prevalent necessitates effective screening and management strategies. Meanwhile, deep learning, a subset of Artificial Intelligence, has significantly advanced ophthalmological diagnostics by automating tasks that required extensive clinical expertise. Although recent studies have investigated the use of deep learning models for refractive power detection through various imaging techniques, a comprehensive systematic review on this topic is has yet be done. This review aims to summarise and evaluate the performance of ocular image-based deep learning models in predicting refractive errors. Main text: We search on three databases (PubMed, Scopus, Web of Science) up till June 2023, focusing on deep learning applications in detecting refractive error from ocular images. We included studies that had reported refractive error outcomes, regardless of publication years. We systematically extracted and evaluated the continuous outcomes (sphere, SE, cylinder) and categorical outcomes (myopia), ground truth measurements, ocular imaging modalities, deep learning models, and performance metrics, adhering to PRISMA guidelines. Nine studies were identified and categorised into three groups: retinal photo-based (n â€‹= â€‹5), OCT-based (n â€‹= â€‹1), and external ocular photo-based (n â€‹= â€‹3).For high myopia prediction, retinal photo-based models achieved AUC between 0.91 and 0.98, sensitivity levels between 85.10% and 97.80%, and specificity levels between 76.40% and 94.50%. For continuous prediction, retinal photo-based models reported MAE ranging from 0.31D to 2.19D, and R 2 between 0.05 and 0.96. The OCT-based model achieved an AUC of 0.79-0.81, sensitivity of 82.30% and 87.20% and specificity of 61.70%-68.90%. For external ocular photo-based models, the AUC ranged from 0.91 to 0.99, sensitivity of 81.13%-84.00% and specificity of 74.00%-86.42%, MAE ranges from 0.07D to 0.18D and accuracy ranges from 81.60% to 96.70%. The reported papers collectively showed promising performances, in particular the retinal photo-based and external eye photo -based DL models. Conclusions: The integration of deep learning model and ocular imaging for refractive error detection appear promising. However, their real-world clinical utility in current screening workflow have yet been evaluated and would require thoughtful consideration in design and implementation.

Lessons learned from follow up of spleen-preserving distal pancreatectomy with artery saving and vein sacrificing.

Background: The Warshaw method as a technique for spleen-preserving distal pancreatectomy (SPDP) carries the risk of splenic infarction following splenic artery ligation. This study introduces a modified Warshaw method, which preserves the splenic artery while sacrificing the splenic vein, and compares its outcomes with the traditional Warshaw method. Methods: According to the bleeding status during vessel dissection, either the Warshaw method (group W) or the modified Warshaw method (group MW) was used. Guided by preoperative imaging, we utilized the planned modified Warshaw method (group PMW) when the splenic vein was embedded in the pancreatic parenchyma. Results: Group MW demonstrated a lower incidence of splenic infarction and engorged gastric collaterals than group W (6.3% vs. 69.8%, P<0.001; 25.0% vs. 55.8%, P=0.003, respectively). There were no significant differences in perioperative changes of splenic volume between the two groups. Group PMW experienced less estimated blood loss than group W (71.9±59.13 vs. 357.9±447.72 cc, P=0.006). Conclusions: The planned modified Warshaw method is an efficient and safe technique, resulting in lower estimated blood loss and favorable outcomes concerning splenic infarction and gastric collaterals than the Warshaw method without inducing congestive splenomegaly.

Integrated Genome-Wide Analysis of DNA Methylation and Gene Expression in the Hippocampi of 5xFAD Alzheimer's Disease Mouse Model.

BACKGROUND: DNA methylation forms 5-methylcytosine and its regulation in the hippocampus is critical for learning and memory. Indeed, dysregulation of DNA methylation is associated with neurological diseases. Alzheimer's disease (AD) is the predominant of dementia and a neurodegenerative disorder. METHODS: We examined the learning and memory function in 3- and 9-month-old wild-type and 5xfamiliar Alzheimer's disease (5xFAD) transgenic mice by performing the object recognition memory and Y-maze tests, and identified the hippocampal amyloid beta burden. To investigate the epigenetically regulated genes involved in the development or neuropathology of AD, we performed genome-wide DNA methylation sequencing and RNA sequencing analyses in the hippocampus of 9-month-old wild-type and 5xFAD tg mice. To validate the genes inversely regulated by epigenetics, we confirmed their methylation status and mRNA levels. RESULTS: At 9 months of age, 5xFAD tg mice showed significant cognitive impairment and amyloid-beta plaques in the hippocampus. DNA methylation sequencing identified a total of 13,777 differentially methylated regions, including 4484 of hyper- and 9293 of hypomethylated regions, that are associated with several gene ontology (GO) terms including 'nervous system development' and 'axon guidance'. In RNA sequencing analysis, we confirmed a total of 101 differentially expressed genes, including 52 up- and 49 downregulated genes, associated with GO functions such as 'positive regulation of synaptic transmission, glutamatergic' and 'actin filament organization'. Through further integrated analysis of DNA methylation and RNA sequencing, three epigenetically regulated genes were selected: thymus cell antigen 1, theta (Thy1), myosin VI (Myo6), and filamin A-interacting protein 1-like (Filip1l). The methylation level of Thy1 decreased and its mRNA levels increased, whereas that of Myo6 and Filip1l increased and their mRNA levels decreased. The common functions of these three genes may be associated with the neural cytoskeleton and synaptic plasticity. CONCLUSIONS: We suggest that the candidate genes epigenetically play a role in AD-associated neuropathology (i.e., amyloid-beta plaques) and memory deficit by influencing neural structure and synaptic plasticity. Furthermore, counteracting dysregulated epigenetic changes may delay or ameliorate AD onset or symptoms.

miR-133a and miR-135a Regulate All-Trans Retinoic Acid-Mediated Differentiation in Pediatric Acute Myeloid Leukemia by Inhibiting CDX2 Translation and Serve as Prognostic Biomarkers.

Background: Acute myeloid leukemia (AML) is a type of blood cancer characterized by excessive growth of immature myeloid cells. Unfortunately, the prognosis of pediatric AML remains unfavorable. It is imperative to further our understanding of the mechanisms underlying leukemogenesis and explore innovative therapeutic approaches to enhance overall disease outcomes for patients with this condition. Methods: Quantitative reverse-transcription PCR was used to quantify the expression levels of microRNA (miR)-133a and miR-135a in 68 samples from 59 pediatric patients with AML. Dual-luciferase reporter transfection assay, Cell Counting Kit-8 assay, and western blot analysis were used to investigate the functions of miR-133a and miR-135a. Results: Our study found that all-trans-retinoic acid (ATRA) promoted the expression of miR-133a and miR-135a in AML cells, inhibited caudal type homeobox 2 (CDX2) expression, and subsequently inhibited the proliferation of AML cells. Additionally, miR-133a and miR-135a were highly expressed in patients with complete remission and those with better survival. Conclusions: miR-133a and miR-135a may play an antioncogenic role in pediatric AML through the ATRA-miRNA133a/135a-CDX2 pathway. They hold promise as potentially favorable prognostic indicators and novel therapeutic targets for pediatric AML.

Three-Year Outcomes of the Paul Glaucoma Implant for Treatment of Glaucoma.

PRCIS: In our case series, the 3-year failure for Paul Glaucoma Implant (PGI) implantation was 14.6%. At 3 years postoperatively, there was a significant reduction in mean intraocular pressure (IOP) and the number of glaucoma medications used. OBJECTIVE: To determine the 3-year efficacy and safety of the PGI, a novel glaucoma tube shunt in patients with glaucoma. METHODS: Retrospective review of all patients who had undergone PGI implantation in a single tertiary institution in Singapore between May 1, 2017 and January 1, 2022. Data were extracted from electronic health records (Computerized Patient Support System 2 and Epic). The primary outcome measure was failure, defined as IOP >18 mm Hg or <6 mm Hg on 2 consecutive visits after 3 months, reoperation for IOP-related indication, explantation of implant, or loss of light perception vision. Complete success was defined as the absence of failure without medications at 36 months, and qualified success similarly, but with medications. Postoperative mean IOP, mean number of IOP-lowering medications used, and visual acuity were also assessed. RESULTS: Forty-eight eyes in 48 patients were identified. Thirty-one patients (64.6%) had primary open angle and angle closure glaucoma, and 18 (37.5%) had previous existing tube implants or trabeculectomy. At 3 years postoperatively, 7 cases (14.6%) fulfilled the criteria for failure and 36 (75%) met the criteria for complete success. The mean IOP at 36 months was 14.9 ± 4.11 mm Hg, from the mean preoperative IOP of 20.6 ± 6.13 mm Hg ( P < 0.001). The mean number of IOP-lowering medications used was reduced from 3.13 ± 0.959 preoperatively to 0.167 ± 0.476 at 36 months ( P < 0.001). The most common postoperative complication was hypotony (n = 17, 35.4%), of which the majority were self-limiting, followed by hyphema (n = 5, 10.4%) and tube exposure (n = 4, 8.3%). CONCLUSION: The PGI demonstrated sustained IOP reduction and a reduction of medication burden at 3 years postoperatively.

Pyrolysis temperature and time of rice husk biochar potentially control ammonia emissions and Chinese cabbage yield from urea-fertilized soils.

Current agricultural practices are increasingly favoring the biochar application to sequester carbon, enhance crop growth, and mitigate various environmental pollutants resulting from nitrogen (N) loss. However, since biochar's characteristics can vary depending on pyrolysis conditions, it is essential to determine the optimal standard, as they can have different effects on soil health. In this study, we categorized rice husk biochars basis on their pH levels and investigated the role of each rice husk biochar in reducing ammonia (NH3) emissions and promoting the growth of Chinese cabbage in urea-fertilized fields. The findings of this study revealed that the variation in pyrolysis conditions of rice husk biochars and N rates affected both the NH3 emissions and crop growth. The neutral (pH 7.10) biochar exhibited effective NH3 volatilization reduction, attributed to its high surface area (6.49 m2 g-1), outperforming the acidic (pH 6.10) and basic (pH 11.01) biochars, particularly under high N rates (640 kg N ha-1). Chinese cabbage yield was highest, reaching 4.00 kg plant-1, with the basic biochar application with high N rates. Therefore, the neutral rice husk biochar effectively mitigate the NH3 emissions from urea-treated fields, while the agronomic performance of Chinese cabbage enhanced in all biochar amendments.

Effects of Livact Granule on Liver Function Recovery After Donor Right Hemi-Hepatectomy.

BACKGROUND: Living donor liver transplantation (LDLT) is currently widespread due to organ shortage. Because LDLT is a high-risk surgery for the donor, donor safety becomes an important issue. In adult LDLT, right lobe grafts are usually used, posing a greater risk to the donor than a left lobe. Reports have demonstrated that branched-chain amino acids help patients recover after hepatectomy. This study aimed to evaluate the effect of Livact granule on donor safety and recovery. METHODS: From January 2016 to December 2021, LDLT was performed on 258 patients at our center. Among them, 148 were in the non-Livact group, and 110 were in the Livact group. Six of 110 patients in the Livact group stopped taking the granules due to nausea and vomiting, leaving 104 patients in the Livact group to be analyzed. Various preoperative and postoperative factors were evaluated to assess donor safety and recovery. RESULTS: In the non-Livact group, the mean donor age was 35.8; in the Livact group, it was 40. There were no differences between the 2 groups in preoperative liver function tests and no difference in future liver remnant or steatosis. There was no difference in total bilirubin level between the 2 groups at 5 days postoperatively; however, in the Livact group, the prothrombin time international normalized ratio was lower, and albumin was higher. The days taken for total bilirubin to normalize were the same in both groups, but fewer days were needed for Livact to realize an international normalized ratio. More patients in the non-Livact group were discharged with the Jackson-Pratt drain because the drainage did not decrease. CONCLUSIONS: In donor right hepatectomy patients, taking Livact granules and branched-chain amino acids helps donor recovery. For donor safety, administration of Livact granules during the perioperative period should be considered.

The Usefulness of High-Magnification Loupe for Reducing Biliary Complications After Liver Transplantation.

BACKGROUND: Despite advances in surgical techniques, biliary complications are still considered to be a technical "Achilles' heel" of liver transplantation (LT). The purpose of this study was to evaluate the effect of loupe magnification in reducing biliary complications after LT. MATERIALS AND METHODS: From April 2017 to February 2022, LT was performed on 307 patients in our center. Among them, except for 3 patients who underwent hepaticojejunostomy, 304 adult patients with LT were enrolled. They were divided into 3 groups according to the loupe magnification: 2.5 times (×2.5 group, n = 105), 3.5 times (×3.5 group, n = 95), and 5.0 times (×5.0 group, n = 105). RESULTS: Biliary complications occurred in 63 (20.7%) patients. Anastomosis site leakage occurred in 37 patients (12.2%), and stricture occurred in 52 patients (17.1%). Anastomosis site leakage occurred in 15 patients (14.3%) in the ×2.5 group, 15 patients (16.0%) in the ×3.5 group, and 7 patients (6.7%) in the ×5.0 group (P = .097). Biliary stricture occurred in 26 patients (24.8%) in the ×2.5 group, 15 patients (16.0%) in the ×3.5 group, and 11 patients (10.5%) in the ×5.0 group (P = .021). Total biliary complications occurred in 31 patients (29.5%) in the ×2.5 group, 19 patients in the ×3.5 group (20.2%), and 13 patients in the ×5.0 group (12.4%) (P = .009). CONCLUSION: The use of a high magnification loupe can reduce biliary complications in liver transplantation. Further large-scale analyses of clinical data or randomized controlled trials are required to support this study.

The Potential Transcriptomic and Metabolomic Mechanisms of ATO and ATRA in Treatment of FLT3-ITD Acute Myeloid Leukemia.

BACKGROUND: Acute myeloid leukemia (AML) with Fms-like tyrosine kinase 3 gene internal tandem duplication (FLT3-ITD) mutations has a poor prognosis. The combination of arsenic trioxide (ATO) and all-trans retinoic acid (ATRA) has a synergistic killing effect on leukemia cells with FLT3-ITD mutation. However, the mechanism, especially the changes of gene expression and metabolic activity remain unclear. Here we explore the transcriptome and metabolomics changes of FLT3-ITD AML cells treated with ATO/ATRA. METHODS: RNA-seq was used to identify differential expressed genes (DEGs), and ultra-high performance liquid chromatography-quadrupole electrostatic field orbital trap mass spectrometry (UHPLC-QE-MS) nontargeted metabolomics method was used to screen out the differential metabolites in FLT3-ITD mutant cell lines treated with ATRA and ATO. KEGG pathway database was utilized for pathway exploration and Seahorse XF24 was used to detect extracellular acidification rate (ECAR). Metabolic polymerase chain reaction (PCR) array and real-time quantitative PCR (RT-qPCR) were used to detect mRNA levels of key metabolic genes of glycolysis and fatty acid after drug treatment. RESULTS: A total of 3873 DEGs were identified and enriched in 281 Gene Ontology (GO) terms, among which 210 were related to biological processes, 43 were related to cellular components, and 28 were related to molecular functions. Besides, 1794 and 927 differential metabolites were screened in positive and negative ion mode separately, and 59 different metabolic pathways were involved, including alanine-aspartate-glutamate metabolic pathway, arginine, and proline metabolic pathway, glycerophospholipid metabolic pathways, etc. According to KEGG Pathway analysis of transcriptome combined with metabolome, glycolysis/gluconeogenesis pathway and fatty acid metabolism pathway were significantly founded enriched. ATRA + ATO may inhibit the glycolysis of FLT3-ITD AML cells by inhibiting FLT3 and its downstream AKT/HK2-VDAC1 signaling pathway. CONCLUSIONS: The gene transcription profile and metabolites of FLT3-ITD mutant cells changes significantly after treatment, which might be related to the anti-FLT3-ITD AML effect. The screened DEGs, differential metabolites pathway are helpful in studying the mechanism of anti-leukemia effects and drug targets.

Integrated Analysis of DNA Methylation and Gene Expression Profiles in a Rat Model of Osteoarthritis.

Osteoarthritis (OA) is common and affected by several factors, such as age, weight, sex, and genetics. The pathogenesis of OA remains unclear. Therefore, using a rat model of monosodium iodoacetate (MIA)-induced OA, we examined genomic-wide DNA methylation using methyl-seq and characterized the transcriptome using RNA-seq in the articular cartilage tissue from a negative control (NC) and MIA-induced rats. We identified 170 genes (100 hypomethylated and upregulated genes and 70 hypermethylated and downregulated genes) regulated by DNA methylation in OA. DNA methylation-regulated genes were enriched in functions related to focal adhesion, extracellular matrix (ECM)-receptor interaction and the PI3K-Akt and Hippo signaling pathways. Functions related to extracellular matrix organization, extracellular matrix proteoglycans, and collagen formation were involved in OA. A molecular and protein-protein network was constructed using methylated expression-correlated genes. Erk1/2 was a downstream target of OA-induced changes in DNA methylation and RNA expression. We found that the integrin subunit alpha 2 (ITGA2) gene is important in focal adhesion, alpha6-beta4 integrin signaling, and the inflammatory response pathway in OA. Overall, gene expression changes because DNA methylation influences OA pathogenesis. ITGA2, whose gene expression changes are regulated by DNA methylation during OA onset, is a candidate gene. Our findings provide insights into the epigenetic targets of OA processes in rats.

Comparative Morphology of Island and Inland Agastache rugosa and Their Gastroprotective Effects in EtOH/HCl-Induced Gastric Mucosal Gastritis.

Agastache rugosa Kuntze (Lamiaceae; Labiatae), a medicinal and functional herb used to treat gastrointestinal diseases, grows well both on islands and inland areas in South Korea. Thus, we aimed to reveal the morphological and micromorphological differences between A. rugosa grown on island and inland areas and their pharmacological effects on gastritis in an animal model by combining morphological and mass spectrophotometric analyses. Morphological analysis showed that island A. rugosa had slightly smaller plants and leaves than inland plants; however, the density of all types of trichomes on the leaves, petioles, and stems of island A. rugosa was significantly higher than that of inland plants. The essential oil component analysis revealed that pulegone levels were substantially higher in island A. rugosa than in inland A. rugosa. Despite the differences between island and inland A. rugosa, treatment with both island and inland A. rugosa reduced gastric damages by more than 40% compared to the gastritis induction group. In addition, expression of inflammatory protein was reduced by about 30% by treatment of island and inland A. rugosa. The present study demonstrates quantitative differences in morphology and volatile components between island and inland plants; significant differences were not observed between the gastritis-inhibitory effects of island and inland A. rugosa, and the efficacy of island A. rugosa was found to be similar to that of A. rugosa grown in inland areas.

Chondroprotective effects of Protaetia brevitarsis seulensis larvae as an edible insect on osteoarthritis in mice.

Osteoarthritis (OA) is a common chronic joint inflammatory disease characterized by progressive destruction of the articular cartilage, bone remodeling, and excessive chronic pain. Most therapeutic approaches do not rescue the progression of OA effectively or provide relief of symptoms. Protaetia brevitarsis seulensis larva (PBSL), which is attracting attention, is an edible insect with very high nutritional value and herbal medicine for the treatment of blood stasis, hepatic disease, and various inflammatory diseases. However, the effect of PBSL on OA has not yet been investigated. This study aimed to demonstrate the effects of PBSL water extract on the progression of OA using monosodium iodoacetate (MIA)-induced mice and SW1353 chondrocytes or murine macrophages. We injected MIA into the intraarticular area of mice following pretreatment with either saline or PBSL (200 mg/kg) for 2 weeks, and then locomotor activity, microcomputed tomography and histopathological analysis, quantitative reverse transcriptase-polymerase chain reaction analysis, and western blot analysis were performed. To determine the molecular effects of PBSL, we used interleukin-1ß (IL-1ß)-induced SW1353 chondrosarcoma or lipopolysaccharide (LPS)-stimulated macrophages. Pretreatment with PBSL diminished the symptoms of OA. Physical activity, articular cartilage damage, and the generation of microfractures were rescued by pretreatment with PBSL in the mouse model. Pretreatment with PBSL suppressed the progress of OA through the regulation of articular cartilage degradation genes and inflammation in both in vivo and in vitro models. Our results demonstrated that PBSL has value as edible insect that can be used in the development of functional foods for OA.

Self-supervised enhanced thyroid nodule detection in ultrasound examination video sequences with multi-perspective evaluation.

Objective.Ultrasound is the most commonly used examination for the detection and identification of thyroid nodules. Since manual detection is time-consuming and subjective, attempts to introduce machine learning into this process are ongoing. However, the performance of these methods is limited by the low signal-to-noise ratio and tissue contrast of ultrasound images. To address these challenges, we extend thyroid nodule detection from image-based to video-based using the temporal context information in ultrasound videos.Approach.We propose a video-based deep learning model with adjacent frame perception (AFP) for accurate and real-time thyroid nodule detection. Compared to image-based methods, AFP can aggregate semantically similar contextual features in the video. Furthermore, considering the cost of medical image annotation for video-based models, a patch scale self-supervised model (PASS) is proposed. PASS is trained on unlabeled datasets to improve the performance of the AFP model without additional labelling costs.Main results.The PASS model is trained by 92 videos containing 23 773 frames, of which 60 annotated videos containing 16 694 frames were used to train and evaluate the AFP model. The evaluation is performed from the video, frame, nodule, and localization perspectives. In the evaluation of the localization perspective, we used the average precision metric with the intersection-over-union threshold set to 50% (AP@50), which is the area under the smoothed Precision-Recall curve. Our proposed AFP improved AP@50 from 0.256 to 0.390, while the PASS-enhanced AFP further improved the AP@50 to 0.425. AFP and PASS also improve the performance in the valuations of other perspectives based on the localization results.Significance.Our video-based model can mitigate the effects of low signal-to-noise ratio and tissue contrast in ultrasound images and enable the accurate detection of thyroid nodules in real-time. The evaluation from multiple perspectives of the ablation experiments demonstrates the effectiveness of our proposed AFP and PASS models.

Highly selective etching of SiNxover SiO2using ClF3/Cl2remote plasma.

Highly selective etching of silicon nitride over silicon oxide is one of the most important processes especially for the fabrication of vertical semiconductor devices including 3D NAND (Not And) devices. In this study, isotropic dry etching characteristics of SiNxand SiO2using ClF3/Cl2remote plasmas have been investigated. The increase of Cl2percent in ClF3/Cl2gas mixture increased etch selectivity of SiNxover SiO2while decreasing SiNxetch rate. By addition of 15% Cl to ClF3/Cl2, the etch selectivity higher than 500 could be obtained with the SiNxetch rate of ∼8 nm min-1, and the increase of Cl percent to 20% further increased the etch selectivity to higher than 1000. It was found that SiNxcan be etched through the reaction from Si-N to Si-F and Si-Cl (also from Si-Cl to Si-F) while SiO2can be etched only through the reaction from Si-O to Si-F, and which is also in extremely low reaction at room temperature. When SiNx/SiO2layer stack was etched using ClF3/Cl2(15%), extremely selective removal of SiNxlayer in the SiNx/SiO2layer stack could be obtained without noticeable etching of SiO2layer in the stack and without etch loading effect.

Benchmarking large language models' performances for myopia care: a comparative analysis of ChatGPT-3.5, ChatGPT-4.0, and Google Bard.

BACKGROUND: Large language models (LLMs) are garnering wide interest due to their human-like and contextually relevant responses. However, LLMs' accuracy across specific medical domains has yet been thoroughly evaluated. Myopia is a frequent topic which patients and parents commonly seek information online. Our study evaluated the performance of three LLMs namely ChatGPT-3.5, ChatGPT-4.0, and Google Bard, in delivering accurate responses to common myopia-related queries. METHODS: We curated thirty-one commonly asked myopia care-related questions, which were categorised into six domains-pathogenesis, risk factors, clinical presentation, diagnosis, treatment and prevention, and prognosis. Each question was posed to the LLMs, and their responses were independently graded by three consultant-level paediatric ophthalmologists on a three-point accuracy scale (poor, borderline, good). A majority consensus approach was used to determine the final rating for each response. 'Good' rated responses were further evaluated for comprehensiveness on a five-point scale. Conversely, 'poor' rated responses were further prompted for self-correction and then re-evaluated for accuracy. FINDINGS: ChatGPT-4.0 demonstrated superior accuracy, with 80.6% of responses rated as 'good', compared to 61.3% in ChatGPT-3.5 and 54.8% in Google Bard (Pearson's chi-squared test, all p ≤ 0.009). All three LLM-Chatbots showed high mean comprehensiveness scores (Google Bard: 4.35; ChatGPT-4.0: 4.23; ChatGPT-3.5: 4.11, out of a maximum score of 5). All LLM-Chatbots also demonstrated substantial self-correction capabilities: 66.7% (2 in 3) of ChatGPT-4.0's, 40% (2 in 5) of ChatGPT-3.5's, and 60% (3 in 5) of Google Bard's responses improved after self-correction. The LLM-Chatbots performed consistently across domains, except for 'treatment and prevention'. However, ChatGPT-4.0 still performed superiorly in this domain, receiving 70% 'good' ratings, compared to 40% in ChatGPT-3.5 and 45% in Google Bard (Pearson's chi-squared test, all p ≤ 0.001). INTERPRETATION: Our findings underscore the potential of LLMs, particularly ChatGPT-4.0, for delivering accurate and comprehensive responses to myopia-related queries. Continuous strategies and evaluations to improve LLMs' accuracy remain crucial. FUNDING: Dr Yih-Chung Tham was supported by the National Medical Research Council of Singapore (NMRC/MOH/HCSAINV21nov-0001).

Classification of Type 2 Diabetes Incidence Risk and the Health Behavior of the 30-50-Year-Old Korean Adults: Latent Class Analysis.

This study aimed to categorize the risk of type 2 diabetes mellitus development (T2DD) in the 30-50-year-old (3050) Korean adults and establish a baseline framework of customized management to prevent the progression to diabetes. A total of 9515 participants were enrolled in the Korea National Health and Nutrition Examination Survey (KNHANES) 2016-2019. Latent class analysis (LCA) was performed based on the health behaviors that were obtained from the secondary data source and were considered to affect T2DD. The major results were compared by latent class, multinomial regression analysis was performed, and the predicted risk of T2DD was evaluated using a self-assessment tool for Korean adults. Data analysis was performed using SPSS (ver. 25.0) and Mplus (ver. 8.6). The latent classes were divided into four categories: negative abdominal obesity and high-risk health behavior (Class A) (28.2%), negative abdominal obesity and low-risk health behavior (Class B) (37.1%), positive abdominal obesity and high-risk health behavior (Class C) (10.7%), and positive abdominal obesity and low-risk health behavior (Class D) (23.9%). The predicted risk scores for T2DD were 6.27 (Class C), 4.50 (Class D), 3.58 (Class A), and 2.16 (Class B), with a higher score indicating a worse state. Significant differences were observed in the predicted risk of T2DD between the latent classes, and abdominal obesity increased the risk. When managing the 30s-50s Korean generation physical activity and abdominal obesity control are strongly recommended.

Deep Ensemble Learning for Retinal Image Classification.

Purpose: Vision impairment affects 2.2 billion people worldwide, half of which is preventable with early detection and treatment. Currently, automatic screening of ocular pathologies using convolutional neural networks (CNNs) on retinal fundus photographs is limited to a few pathologies. Simultaneous detection of multiple ophthalmic pathologies would increase clinical usability and uptake. Methods: Two thousand five hundred sixty images were used from the Retinal Fundus Multi-Disease Image Dataset (RFMiD). Models were trained (n = 1920) and validated (n = 640). Five selected CNN architectures were trained to predict the presence of any pathology and categorize the 28 pathologies. All models were trained to minimize asymmetric loss, a modified form of binary cross-entropy. Individual model predictions were averaged to obtain a final ensembled model and assessed for mean area under the receiver-operator characteristic curve (AUROC) for disease screening (healthy versus pathologic image) and classification (AUROC for each class). Results: The ensemble network achieved a disease screening (healthy versus pathologic) AUROC score of 0.9613. The highest single network score was 0.9586 using the SE-ResNeXt architecture. For individual disease classification, the average AUROC score for each class was 0.9295. Conclusions: Retinal fundus images analyzed by an ensemble of CNNs trained to minimize asymmetric loss were effective in detection and classification of ocular pathologies than individual models. External validation is needed to translate machine learning models to diverse clinical contexts. Translational Relevance: This study demonstrates the potential benefit of ensemble-based deep learning methods on improving automatic screening and diagnosis of multiple ocular pathologies from fundoscopy imaging.

Development of Models to Estimate Total Soil Carbon across Different Croplands at a Regional Scale Using RGB Photography.

A quick, accurate and cost-effective method for estimating total soil carbon is necessary for monitoring its levels due to its environmentally and agronomically irreplaceable importance. There are several impediments to both laboratory analysis and spectroscopic sensor technology because the former is both expensive and time-consuming whereas the initial cost of the latter is too high for farmers to afford. RGB photography obtained from digital cameras could be used to quickly and cheaply estimate the total carbon (TC) content of the soil. In this study, we developed models to predict soil TC contents across different cropland types including paddy, upland and orchard fields as well as the TC content of the soil combined from all the aforementioned cropland types on a regional scale. Soil colour measurements were made on samples from the Chungcheongnam-do province of South Korea. The soil TC content ranged from 0.045% to 6.297%. Modelling was performed using multiple linear regression considering the soil moisture levels and illuminance. The best soil TC prediction model came from the upland soil and gave training and validation r2 values of 0.536 and 0.591 with RMSE values of 0.712% and 0.441%, respectively. However, the most accurate equation is the one that produces the lowest RMSE value. Hence, although the model for the upland soil was the most stable of all, the paddy soil model which gave training and validation r2 values of 0.531 and 0.554 with RMSE values of 0.240% and 0.199%, respectively, was selected as the best soil TC prediction equation of all due to its comparatively high r2 value and the lowest RMSE of all equations.