Hybrid Techniques for Diagnosis with WSIs for Early Detection of Cervical Cancer Based on Fusion Features

Applied Sciences, Journal Year: 2022, Volume and Issue: 12(17), P. 8836 - 8836

Published: Sept. 2, 2022

Cervical cancer is a global health problem that threatens the lives of women. Liquid-based cytology (LBC) one most used techniques for diagnosing cervical cancer; converting from vitreous slides to whole-slide images (WSIs) allows be evaluated by artificial intelligence techniques. Because lack cytologists and devices, it major promote automated systems receive diagnose huge amounts quickly accurately, which are useful in hospitals clinical laboratories. This study aims extract features hybrid method obtain representative achieve promising results. Three proposed approaches have been applied with different methods materials as follows: The first approach called VGG-16 SVM GoogLeNet SVM. second classify abnormal cell ANN classifier extracted GoogLeNet. A third cells an combine them hand-crafted features, using Fuzzy Color Histogram (FCH), Gray Level Co-occurrence Matrix (GLCM) Local Binary Pattern (LBP) algorithms. Based on mixed CNN FCH, GLCM, LBP (hand-crafted), reached best results cervix. network achieved accuracy 99.4%, specificity 100%, sensitivity 99.35%, AUC 99.89% precision 99.42%.

Language: Английский

---

Multi-Method Analysis of Histopathological Image for Early Diagnosis of Oral Squamous Cell Carcinoma Using Deep Learning and Hybrid Techniques

Cancers, Journal Year: 2023, Volume and Issue: 15(21), P. 5247 - 5247

Published: Oct. 31, 2023

Oral cancer is a fatal disease and ranks seventh among the most common cancers throughout whole globe. type of that usually affects head neck. The current gold standard for diagnosis histopathological investigation, however, conventional approach time-consuming requires professional interpretation. Therefore, early Squamous Cell Carcinoma (OSCC) crucial successful therapy, reducing risk mortality morbidity, while improving patient's chances survival. Thus, we employed several artificial intelligence techniques to aid clinicians or physicians, thereby significantly workload pathologists. This study aimed develop hybrid methodologies based on fused features generate better results OSCC. three different strategies, each using five distinct models. first strategy transfer learning Xception, Inceptionv3, InceptionResNetV2, NASNetLarge, DenseNet201 second involves pre-trained art CNN feature extraction coupled with Support Vector Machine (SVM) classification. In particular, were extracted various models, namely DenseNet201, subsequently applied SVM algorithm evaluate classification accuracy. final employs cutting-edge fusion technique, utilizing an art-of-CNN model extract deep aforementioned These underwent dimensionality reduction through principal component analysis (PCA). Subsequently, low-dimensionality are combined shape, color, texture gray-level co-occurrence matrix (GLCM), Histogram Oriented Gradient (HOG), Local Binary Pattern (LBP) methods. Hybrid was incorporated into enhance performance. proposed system achieved promising rapid OSCC histological images. accuracy, precision, sensitivity, specificity, F-1 score, area under curve (AUC) support vector machine GLCM, HOG, LBP 97.00%, 96.77%, 90.90%, 98.92%, 93.74%, 96.80%, respectively.

Language: Английский

---

Leveraging technology-driven strategies to untangle omics big data: circumventing roadblocks in clinical facets of oral cancer

Frontiers in Oncology, Journal Year: 2024, Volume and Issue: 13

Published: Jan. 3, 2024

Oral cancer is one of the 19most rapidly progressing cancers associated with significant mortality, owing to its extreme degree invasiveness and aggressive inclination. The early occurrences this can be clinically deceiving leading a poor overall survival rate. primary concerns from clinical perspective include delayed diagnosis, rapid disease progression, resistance various chemotherapeutic regimens, metastasis, which collectively pose substantial threat prognosis. Conventional practices observed since antiquity no longer offer best possible options circumvent these roadblocks. world current research has been revolutionized advent state-of-the-art technology-driven strategies that ray hope in confronting said challenges by highlighting crucial underlying molecular mechanisms drivers. In recent years, bioinformatics Machine Learning (ML) techniques have enhanced possibility detection, evaluation prognosis, individualization therapy. This review elaborates on application aforesaid unraveling potential hints omics big data address complexities existing facets oral cancer. first section demonstrates utilization ML disentangle impediments related diagnosis. includes technology-based optimize classification, staging via uncovering biomarkers signatures. Furthermore, breakthrough concepts such as salivaomics-driven non-invasive biomarker discovery omics-complemented surgical interventions are articulated detail. following part, identification novel disease-specific targets alongside therapeutic agents confront omics-based methodologies presented. Additionally, special emphasis placed drug resistance, precision medicine, repurposing. final section, we discuss approaches oriented toward unveiling prognostic constructing prediction models capture metastatic tumors. Overall, intend provide bird’s eye view omics, bioinformatics, currently being used through relevant case studies.

Language: Английский

---

Machine Learning-Based Genome-Wide Salivary DNA Methylation Analysis for Identification of Noninvasive Biomarkers in Oral Cancer Diagnosis

Cancers, Journal Year: 2022, Volume and Issue: 14(19), P. 4935 - 4935

Published: Oct. 8, 2022

This study aims to examine the feasibility of ML-assisted salivary-liquid-biopsy platforms using genome-wide methylation analysis at base-pair and regional resolution for delineating oral squamous cell carcinoma (OSCC) potentially malignant disorders (OPMDs). A nested cohort patients with OSCC OPMDs was randomly selected from among mucosal diseases. Saliva samples were collected, DNA extracted pellets processed reduced-representation bisulfite sequencing. Reads a minimum 10× coverage used identify differentially methylated CpG sites (DMCs) 100 bp regions (DMRs). The performance eight ML models three feature-selection methods (ANOVA, MRMR, LASSO) then compared determine optimal biomarker based on DMCs DMRs. total 1745 105 DMRs identified detecting OSCC. proportion hypomethylated hypermethylated similar (51% vs. 49%), while most (62.9%). Furthermore, more than annotated promoter (36% 16%) intergenic (50% 36%). Of all compared, linear SVM model 11 by LASSO had perfect AUC, recall, specificity, calibration (1.00) detection. Overall, techniques can be applied directly saliva discovery ML-based may useful in stratifying during disease screening monitoring.

Language: Английский

---

Insight into the Relationships Between Chemical, Protein and Functional Variables in the PBP/GOBP Family in Moths Based on Machine Learning

International Journal of Molecular Sciences, Journal Year: 2025, Volume and Issue: 26(5), P. 2302 - 2302

Published: March 5, 2025

During their lives, insects must cope with a plethora of chemicals, which few will have an impact at the behavioral level. To detect these use several protein families located in main olfactory organs, antennae. Inside antennae, odorant-binding proteins (OBPs), as most studied family, bind volatile chemicals to transport them. Pheromone-binding (PBPs) and general-odorant-binding (GOPBs) are two subclasses OBPs evolved moths putative role. Predictions for OBP-chemical interactions remained limited, functional data collected over years unused. In this study, chemical, were curated, related datasets created descriptors. Regression algorithms implemented performance evaluated. Our results indicate that XGBoostRegressor exhibits best (R2 0.76, RMSE 0.28 MAE 0.20), followed by GradientBoostingRegressor LightGBMRegressor. our knowledge, is first study showing correlation among data, particularly context PBP/GOBP family moths.

Language: Английский

---

Application of machine learning in dentistry: insights, prospects and challenges

Acta Odontologica Scandinavica, Journal Year: 2025, Volume and Issue: 84, P. 145 - 154

Published: March 27, 2025

Background: Machine learning (ML) is transforming dentistry by setting new standards for precision and efficiency in clinical practice, while driving improvements care delivery quality. Objectives: This review: (1) states the necessity to develop ML purpose of breaking limitations traditional dental technologies; (2) discusses principles ML-based models utilised practice care; (3) outlines application respects dentistry; (4) highlights prospects challenges be addressed. Data sources: In this narrative review, a comprehensive search was conducted PubMed/MEDLINE, Web Science, ScienceDirect, Institute Electrical Electronics Engineers (IEEE) Xplore databases. Conclusions: Learning has demonstrated significant potential with its intelligently assistive function, promoting diagnostic efficiency, personalised treatment plans related streamline workflows. However, data privacy, security, interpretability, ethical considerations were highly urgent addressed next objective creating backdrop future research rapidly expanding arena. Clinical significance: Development brought transformative impact fields dentistry, from diagnostic, plan Particularly, integrating tools will significantly enhance surgeries treatments.

Language: Английский

---

Validity and accuracy of artificial intelligence-based dietary intake assessment methods: a systematic review

British Journal Of Nutrition, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 13

Published: April 10, 2025

Abstract One of the most significant challenges in research related to nutritional epidemiology is achievement high accuracy and validity dietary data establish an adequate link between exposure health outcomes. Recently, emergence artificial intelligence (AI) various fields has filled this gap with advanced statistical models techniques for nutrient food analysis. We aimed systematically review available evidence regarding AI-based intake assessment methods (AI-DIA). In accordance PRISMA guidelines, exhaustive search EMBASE, PubMed, Scopus Web Science databases was conducted identify relevant publications from their inception 1 December 2024. Thirteen studies that met inclusion criteria were included Of identified, 61·5 % preclinical settings. Likewise, 46·2 used AI based on deep learning 15·3 machine learning. Correlation coefficients over 0·7 reported six articles concerning estimation calories traditional methods. Similarly, obtained a correlation above macronutrients. case micronutrients, four achieved mentioned above. A moderate risk bias observed ( n 8) analysed, confounding being frequently observed. AI-DIA are promising, reliable valid alternatives estimations. However, more comparing different populations needed, as well larger sample sizes, ensure experimental designs.

Language: Английский

---

Personalized Support for People with Disabilities Through Generative AI

Published: Jan. 1, 2025

Language: Английский

---

Few-shot learning based oral cancer diagnosis using a dual feature extractor prototypical network

Journal of Biomedical Informatics, Journal Year: 2024, Volume and Issue: 150, P. 104584 - 104584

Published: Jan. 8, 2024

Language: Английский

---

Enhancing cancer detection and prevention mechanisms using advanced machine learning approaches

Informatics in Medicine Unlocked, Journal Year: 2024, Volume and Issue: 50, P. 101579 - 101579

Published: Jan. 1, 2024

Cancer is a major global health challenge, emphasizing the critical need for early detection to enhance patient outcomes. This study thoroughly investigates applications of advanced machine learning methods cancer and prevention, aiming develop robust algorithms that can accurately identify cancerous cells assess severity based on key parameters. The authors synthesize insights from previous prevention research through an in-depth literature review. sets specific objectives, including creating evaluating innovative classifying cells. employed techniques analyze parameters, such as cell size, shape, nucleus characteristics, additional factors like texture, mitosis count, tumour progression, metastasis, gene expression patterns, biological markers. methodology distinguished by its effective use diverse data types automated feature extraction improve prediction accuracy. Advanced precision reliability current classification algorithms. underscores importance timely accurate detection, which enables intervention significantly improves survival rates. results discussion section meticulously analyzes findings, demonstrating approach's effectiveness in identifying assessing severity. valuable resource medical professionals, supporting early-stage at different stages. proposed show promising improving accuracy efficiency systems, paving way future advancements offering essential healthcare practitioners researchers.

Language: Английский

---