OBJECTIVES: To assess the ability, as well as factors affecting the ability, of ultrasound examiners with different levels of ultrasound experience to detect correctly infiltration of ovarian cancer in predefined anatomical locations, and to evaluate the inter-rater agreement regarding the presence or absence of cancer infiltration, using preacquired ultrasound videoclips obtained in a selected patient sample with a high prevalence of cancer spread. METHODS: This study forms part of the Imaging Study in Advanced ovArian Cancer multicenter observational study (NCT03808792). Ultrasound videoclips showing assessment of infiltration of ovarian cancer were obtained by the principal investigator (an ultrasound expert, who did not participate in rating) at 19 predefined anatomical sites in the abdomen and pelvis, including five sites that, if infiltrated, would indicate tumor non-resectability. For each site, there were 10 videoclips showing cancer infiltration and 10 showing no cancer infiltration. The reference standard was either findings at surgery with histological confirmation or response to chemotherapy. For statistical analysis, the 19 sites were grouped into four anatomical regions: pelvis, middle abdomen, upper abdomen and lymph nodes. The videoclips were assessed by raters comprising both senior gynecologists (mainly self-trained expert ultrasound examiners who perform preoperative ultrasound assessment of ovarian cancer spread almost daily) and gynecologists who had undergone a minimum of 6 months' supervised training in the preoperative ultrasound assessment of ovarian cancer spread in a gynecological oncology center. The raters were classified as highly experienced or less experienced based on annual individual caseload and the number of years that they had been performing ultrasound evaluation of ovarian cancer spread. Raters were aware that for each site there would be 10 videoclips with and 10 without cancer infiltration. Each rater independently classified every videoclip as showing or not showing cancer infiltration and rated the image quality (on a scale from 0 to 10) and their diagnostic confidence (on a scale from 0 to 10). A generalized linear mixed model with random effects was used to estimate which factors (including level of experience, image quality, diagnostic confidence and anatomical region) affected the likelihood of a correct classification of cancer infiltration. We assessed the observed percentage of videoclips classified correctly, the expected percentage of videoclips classified correctly based on the generalized linear mixed model and inter-rater agreement (reliability) in classifying anatomical sites as being infiltrated by cancer. RESULTS: Twenty-five raters participated in the study, of whom 13 were highly experienced and 12 were less experienced. The observed percentage of correct classification of cancer infiltration ranged from 70% to 100% depending on rater and anatomical site, and the median percentage of correct classification for the 25 raters ranged from 90% to 100%. The probability of correct classification of all 380 videoclips ranged from 0.956 to 0.975 and was not affected by the rater's level of ultrasound experience. The likelihood of correct classification increased with increased image quality and diagnostic confidence and was affected by anatomical region. It was highest for sites in the pelvis, second highest for those in the middle abdomen, third highest for lymph nodes and lowest for sites in the upper abdomen. The inter-rater agreement of all 25 raters regarding the presence of cancer infiltration ranged from substantial (Fleiss kappa, 0.68 (95% CI, 0.66-0.71)) to very good (Fleiss kappa, 0.99 (95% CI, 0.97-1.00)) depending on the anatomical site. It was lowest for sites in the upper abdomen (Fleiss kappa, 0.68 (95% CI, 0.66-0.71) to 0.97 (95% CI, 0.94-0.99)) and highest for sites in the pelvis (Fleiss kappa, 0.94 (95% CI, 0.92-0.97) to 0.99 (95% CI, 0.97-1.00)). CONCLUSIONS: Ultrasound examiners with different levels of ultrasound experience can classify correctly predefined anatomical sites as being infiltrated or not infiltrated by ovarian cancer based on video recordings obtained by an experienced ultrasound examiner, and the inter-rater agreement is substantial. The likelihood of correct classification as well as the inter-rater agreement is highest for sites in the pelvis and lowest for sites in the upper abdomen. However, owing to the study design, our results regarding diagnostic accuracy and inter-rater agreement are likely to be overoptimistic. © 2025 The Author(s). Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.
- MeSH
- Video Recording MeSH
- Abdomen diagnostic imaging pathology MeSH
- Neoplasm Invasiveness diagnostic imaging MeSH
- Clinical Competence * MeSH
- Middle Aged MeSH
- Humans MeSH
- Ovarian Neoplasms * diagnostic imaging pathology MeSH
- Observer Variation MeSH
- Pelvis diagnostic imaging pathology MeSH
- Prevalence MeSH
- Aged MeSH
- Ultrasonography methods MeSH
- Check Tag
- Middle Aged MeSH
- Humans MeSH
- Aged MeSH
- Female MeSH
- Publication type
- Journal Article MeSH
- Multicenter Study MeSH
- Observational Study MeSH
- Research Support, Non-U.S. Gov't MeSH
OBJECTIVES: Artificial Intelligence (AI), particularly deep learning, has significantly impacted healthcare, including dentistry, by improving diagnostics, treatment planning, and prognosis prediction. This systematic mapping review explores the current applications of deep learning in dentistry, offering a comprehensive overview of trends, models, and their clinical significance. MATERIALS AND METHODS: Following a structured methodology, relevant studies published from January 2012 to September 2023 were identified through database searches in PubMed, Scopus, and Embase. Key data, including clinical purpose, deep learning tasks, model architectures, and data modalities, were extracted for qualitative synthesis. RESULTS: From 21,242 screened studies, 1,007 were included. Of these, 63.5% targeted diagnostic tasks, primarily with convolutional neural networks (CNNs). Classification (43.7%) and segmentation (22.9%) were the main methods, and imaging data-such as cone-beam computed tomography and orthopantomograms-were used in 84.4% of cases. Most studies (95.2%) applied fully supervised learning, emphasizing the need for annotated data. Pathology (21.5%), radiology (17.5%), and orthodontics (10.2%) were prominent fields, with 24.9% of studies relating to more than one specialty. CONCLUSION: This review explores the advancements in deep learning in dentistry, particulary for diagnostics, and identifies areas for further improvement. While CNNs have been used successfully, it is essential to explore emerging model architectures, learning approaches, and ways to obtain diverse and reliable data. Furthermore, fostering trust among all stakeholders by advancing explainable AI and addressing ethical considerations is crucial for transitioning AI from research to clinical practice. CLINICAL RELEVANCE: This review offers a comprehensive overview of a decade of deep learning in dentistry, showcasing its significant growth in recent years. By mapping its key applications and identifying research trends, it provides a valuable guide for future studies and highlights emerging opportunities for advancing AI-driven dental care.
- MeSH
- Deep Learning * MeSH
- Humans MeSH
- Dentistry * MeSH
- Check Tag
- Humans MeSH
- Publication type
- Journal Article MeSH
- Review MeSH
- Systematic Review MeSH
Článek se zaměřuje na zmapování současného postavení studijního koordinátora ve zdravotnických zařízeních v České republice. Hlavním cílem je zvýšit povědomí o činnostech, které studijní koordinátoři vykonávají, a identifikovat potřeby jejich vzdělávání. Studie rovněž zkoumá poptávku po formálním vzdělávacím programu, který by podpořil profesní rozvoj této klíčové role v klinickém výzkumu. Výsledky dvou dotazníkových průzkumů, kterých se aktivně zúčastnilo 42 zdravotnických zařízení a 120 studijních koordinátorů, ukazují, že pouze 45 % zařízení má formalizované oddělení klinických studií. Přibližně 40 % studijních koordinátorů je zařazeno do nezdravotnických kategorií, což komplikuje jejich odpovídající pracovní zařazení. Studijním koordinátorům chybí ucelený systém pregraduálního i postgraduálního vzdělávání. Závěry zdůrazňují potřebu formalizace této pracovní pozice včetně vytvoření profesní organizace, která by sjednocovala metodické pokyny a posílila institucionální podporu studijních koordinátorů.
The article focuses on mapping the current status of clinical research coordinators in healthcare institutions in the Czech Republic. The main objective is to raise awareness of the tasks performed by clinical research coordinators and to identify their educational needs. The study also examines the demand for a formal educational program that would support the professional development of this crucial role in clinical research. The results of two surveys, involving 42 healthcare institutions and 120 clinical research coordinators, show that only 45% of institutions have a formal Clinical Trials Unit. Approximately 40% of clinical research coordinators are classified in non-healthcare categories, complicating their appropriate job classification. Coordinators lack a comprehensive system for both undergraduate and postgraduate education. The conclusions emphasize the need for formalizing this position, including the establishment of a professional organisation to unify guidelines and strengthen institutional support for clinical research coordinators.
- Keywords
- studijní koordinátor,
- MeSH
- Clinical Studies as Topic * methods MeSH
- Humans MeSH
- Organization and Administration statistics & numerical data MeSH
- Surveys and Questionnaires MeSH
- Education, Professional classification methods MeSH
- Health Facilities * statistics & numerical data MeSH
- Check Tag
- Humans MeSH
- Publication type
- Research Support, Non-U.S. Gov't MeSH
- Review MeSH
BACKGROUND: Heart failure (HF) is becoming an increasingly prevalent issue, particularly among the elderly population. Lipids are closely associated with cardiovascular disease (CVD) pathology. Lipidomics as a comprehensive profiling tool is showing to be promising in the prediction of events and mortality due to CVD as well as identifying novel biomarkers. MATERIALS AND METHODS: In this study, eicosanoids and lipid profiles were measured in order to predict survival in patients with de novo or acute decompensated HF. Our study included 50 patients (16 females, mean age 73 years and 34 males, mean age 71 years) with de novo or acute decompensated chronic HF with a median follow-up of 7 months. Lipids were semiquantified using targeted lipidomic liquid chromatography-mass spectrometry (LC-MS/MS) analysis. Eicosanoid concentrations were determined using a commercially available sandwich ELISA assay. RESULTS: From 736 lipids and 3 eicosanoids, 39 significant lipids were selected (by using the Mann-Whitney U test after Benjamini-Hochberg correction) with the highest number of representatives belonging to the polyunsaturated (PUFA) phosphatidylcholines (PC). PC 42:10 (p = 1.44 × 10-4) was found to be the most statistically significantly elevated in the surviving group with receiver operating characteristics of AUC = 0.84 (p = 3.24 × 10-7). A multivariate supervised discriminant analysis based on the aforementioned lipid panel enabled the classification of the groups of surviving and non-surviving patients with 90 % accuracy. CONCLUSIONS: In the present study we describe a trend in PUFA esterified in PC that were systematically increased in surviving patients with HF. This trend in low-abundant and rarely identified PUFA PC (mainly very long chain PUFA containing PC such as PC 42:10 or PC 40:9 containing FA 22:6, FA 20:5 and FA 20:4) suggests candidate biomarkers.
- Publication type
- Journal Article MeSH
Manual visual review, annotation and categorization of electroencephalography (EEG) is a time-consuming task that is often associated with human bias and requires trained electrophysiology experts with specific domain knowledge. This challenge is now compounded by development of measurement technologies and devices allowing large-scale heterogeneous, multi-channel recordings spanning multiple brain regions over days, weeks. Currently, supervised deep-learning techniques were shown to be an effective tool for analyzing big data sets, including EEG. However, the most significant caveat in training the supervised deep-learning models in a clinical research setting is the lack of adequate gold-standard annotations created by electrophysiology experts. Here, we propose a semi-supervised machine learning technique that utilizes deep-learning methods with a minimal amount of gold-standard labels. The method utilizes a temporal autoencoder for dimensionality reduction and a small number of the expert-provided gold-standard labels used for kernel density estimating (KDE) maps. We used data from electrophysiological intracranial EEG (iEEG) recordings acquired in two hospitals with different recording systems across 39 patients to validate the method. The method achieved iEEG classification (Pathologic vs. Normal vs. Artifacts) results with an area under the receiver operating characteristic (AUROC) scores of 0.862 ± 0.037, 0.879 ± 0.042, and area under the precision-recall curve (AUPRC) scores of 0.740 ± 0.740, 0.714 ± 0.042. This demonstrates that semi-supervised methods can provide acceptable results while requiring only 100 gold-standard data samples in each classification category. Subsequently, we deployed the technique to 12 novel patients in a pseudo-prospective framework for detecting Interictal epileptiform discharges (IEDs). We show that the proposed temporal autoencoder was able to generalize to novel patients while achieving AUROC of 0.877 ± 0.067 and AUPRC of 0.705 ± 0.154.
The scarcity of high-quality annotations in many application scenarios has recently led to an increasing interest in devising learning techniques that combine unlabeled data with labeled data in a network. In this work, we focus on the label propagation problem in multilayer networks. Our approach is inspired by the heat diffusion model, which shows usefulness in machine learning problems such as classification and dimensionality reduction. We propose a novel boundary-based heat diffusion algorithm that guarantees a closed-form solution with an efficient implementation. We experimentally validated our method on synthetic networks and five real-world multilayer network datasets representing scientific coauthorship, spreading drug adoption among physicians, two bibliographic networks, and a movie network. The results demonstrate the benefits of the proposed algorithm, where our boundary-based heat diffusion dominates the performance of the state-of-the-art methods.
- MeSH
- Algorithms MeSH
- Supervised Machine Learning * MeSH
- Machine Learning MeSH
- Hot Temperature * MeSH
- Publication type
- Journal Article MeSH
BACKGROUND: Cell-free DNA (cfDNA) analysis holds great promise for non-invasive cancer screening, diagnosis, and monitoring. We hypothesized that mining the patterns of cfDNA shallow whole-genome sequencing datasets from patients with cancer could improve cancer detection. METHODS: By applying unsupervised clustering and supervised machine learning on large cfDNA shallow whole-genome sequencing datasets from healthy individuals (n = 367) and patients with different hematological (n = 238) and solid malignancies (n = 320), we identified cfDNA signatures that enabled cancer detection and typing. RESULTS: Unsupervised clustering revealed cancer type-specific sub-grouping. Classification using a supervised machine learning model yielded accuracies of 96% and 65% in discriminating hematological and solid malignancies from healthy controls, respectively. The accuracy of disease type prediction was 85% and 70% for the hematological and solid cancers, respectively. The potential utility of managing a specific cancer was demonstrated by classifying benign from invasive and borderline adnexal masses with an area under the curve of 0.87 and 0.74, respectively. CONCLUSIONS: This approach provides a generic analytical strategy for non-invasive pan-cancer detection and cancer type prediction.
- MeSH
- Humans MeSH
- Biomarkers, Tumor genetics MeSH
- Neoplasms * diagnosis genetics MeSH
- Whole Genome Sequencing MeSH
- Cell-Free Nucleic Acids * MeSH
- Check Tag
- Humans MeSH
- Publication type
- Journal Article MeSH
- Research Support, Non-U.S. Gov't MeSH
The current population worldwide extensively uses social media to share thoughts, societal issues, and personal concerns. Social media can be viewed as an intelligent platform that can be augmented with a capability to analyze and predict various issues such as business needs, environmental needs, election trends (polls), governmental needs, etc. This has motivated us to initiate a comprehensive search of the COVID-19 pandemic-related views and opinions amongst the population on Twitter. The basic training data have been collected from Twitter posts. On this basis, we have developed research involving ensemble deep learning techniques to reach a better prediction of the future evolutions of views in Twitter when compared to previous works that do the same. First, feature extraction is performed through an N-gram stacked autoencoder supervised learning algorithm. The extracted features are then involved in a classification and prediction involving an ensemble fusion scheme of selected machine learning techniques such as decision tree (DT), support vector machine (SVM), random forest (RF), and K-nearest neighbour (KNN). all individual results are combined/fused for a better prediction by using both mean and mode techniques. Our proposed scheme of an N-gram stacked encoder integrated in an ensemble machine learning scheme outperforms all the other existing competing techniques such unigram autoencoder, bigram autoencoder, etc. Our experimental results have been obtained from a comprehensive evaluation involving a dataset extracted from open-source data available from Twitter that were filtered by using the keywords "covid", "covid19", "coronavirus", "covid-19", "sarscov2", and "covid_19".
- MeSH
- COVID-19 * MeSH
- Humans MeSH
- Pandemics MeSH
- SARS-CoV-2 MeSH
- Social Media * MeSH
- Social Networking MeSH
- Machine Learning MeSH
- Check Tag
- Humans MeSH
- Publication type
- Journal Article MeSH
Precise classification of acute leukemia (AL) is crucial for adequate treatment. EuroFlow has previously designed an AL orientation tube (ALOT) to guide toward the relevant classification panel and final diagnosis. In this study, we designed and validated an algorithm for automated (database-supported) gating and identification (AGI tool) of cell subsets within samples stained with ALOT. A reference database of normal peripheral blood (PB, n = 41) and bone marrow (BM; n = 45) samples analyzed with the ALOT was constructed, and served as a reference for the AGI tool to automatically identify normal cells. Populations not unequivocally identified as normal cells were labeled as checks and were classified by an expert. Additional normal BM (n = 25) and PB (n = 43) and leukemic samples (n = 109), analyzed in parallel by experts and the AGI tool, were used to evaluate the AGI tool. Analysis of normal PB and BM samples showed low percentages of checks (<3% in PB, <10% in BM), with variations between different laboratories. Manual analysis and AGI analysis of normal and leukemic samples showed high levels of correlation between cell numbers (r2 > 0.95 for all cell types in PB and r2 > 0.75 in BM) and resulted in highly concordant classification of leukemic cells by our previously published automated database-guided expert-supervised orientation tool for immunophenotypic diagnosis and classification of acute leukemia (Compass tool). Similar data were obtained using alternative, commercially available tubes, confirming the robustness of the developed tools. The AGI tool represents an innovative step in minimizing human intervention and requirements in expertise, toward a "sample-in and result-out" approach which may result in more objective and reproducible data analysis and diagnostics. The AGI tool may improve quality of immunophenotyping in individual laboratories, since high percentages of checks in normal samples are an alert on the quality of the internal procedures.
- MeSH
- Leukemia, Myeloid, Acute diagnosis MeSH
- Algorithms * MeSH
- Immunophenotyping methods MeSH
- Leukocytes pathology MeSH
- Humans MeSH
- Flow Cytometry MeSH
- Check Tag
- Humans MeSH
- Publication type
- Journal Article MeSH
- Research Support, Non-U.S. Gov't MeSH
Breast meat from modern fast-growing big birds is affected with myopathies such as woody breast (WB), white striping, and spaghetti meat (SM). The detection and separation of the myopathy-affected meat can be carried out at processing plants using technologies such as bioelectrical impedance analysis (BIA). However, BIA raw data from myopathy-affected breast meat are extremely complicated, especially because of the overlap of these myopathies in individual breast fillets and the human error associated with the assignment of fillet categories. Previous research has shown that traditional statistical techniques such as ANOVA and regression, among others, are insufficient in categorising fillets affected with myopathies by BIA. Therefore, more complex data analysis tools can be used, such as support vector machines (SVMs) and backpropagation neural networks (BPNNs), to classify raw poultry breast myopathies using their BIA patterns, such that the technology can be beneficial for the poultry industry in detecting myopathies. Freshly deboned (3-3.5 h post slaughter) breast fillets (n = 100 × 3 flocks) were analysed by hand palpation for WB (0-normal; 1-mild; 2-moderate; 3-Severe) and SM (presence and absence) categorisation. BIA data (resistance and reactance) were collected on each breast fillet; the algorithm of the equipment calculated protein and fat index. The data were analysed by linear discriminant analysis (LDA), and with SVM and BPNN with 70::30: training::test data set. Compared with the LDA analysis, SVM separated WB with a higher accuracy of 71.04% for normal (data for normal and mild merged), 59.99% for moderate, and 81.48% for severe WB. Compared with SVM, the BPNN training model accurately (100%) separated normal WB fillets with and without SM, demonstrating the ability of BIA to detect SM. Supervised learning algorithms, such as SVM and BPNN, can be combined with BIA and successfully implemented in poultry processing to detect breast fillet myopathies.
- Publication type
- Journal Article MeSH