MammoViT: A Custom Vision Transformer Architecture for Accurate BIRADS Classification in Mammogram Analysis

Diagnostics, Год журнала: 2025, Номер 15(3), С. 285 - 285

Опубликована: Янв. 25, 2025

Background: Breast cancer screening through mammography interpretation is crucial for early detection and improved patient outcomes. However, the manual classification of mammograms using BIRADS (Breast Imaging-Reporting Data System) remains challenging due to subtle imaging features, inter-reader variability, increasing radiologist workload. Traditional computer-aided systems often struggle with complex feature extraction contextual understanding mammographic abnormalities. To address these limitations, this study proposes MammoViT, a novel hybrid deep learning framework that leverages both ResNet50’s hierarchical capabilities Vision Transformer’s ability capture long-range dependencies in images. Methods: We implemented multi-stage approach utilizing pre-trained ResNet50 model initial from mammogram significant class imbalance our four-class dataset, we applied SMOTE (Synthetic Minority Over-sampling Technique) generate synthetic samples minority classes. The extracted arrays were transformed into non-overlapping patches positional encodings Transformer processing. employs multi-head self-attention mechanisms local global relationships between image patches, each attention head different aspects spatial dependencies. was optimized Keras Tuner trained 5-fold cross-validation stopping prevent overfitting. Results: MammoViT achieved 97.4% accuracy classifying images across categories. model’s effectiveness validated comprehensive evaluation metrics, including report, confusion matrix, probability distribution, comparison existing studies. Conclusions: effectively combines architectures while addressing challenge imbalanced medical datasets. high robust performance demonstrate its potential as reliable tool supporting clinical decision-making breast screening.

Язык: Английский

---

Advancing Breast Cancer Subtype Prediction and Mutation Analysis: Integrating Deep Learning and Machine Learning Techniques in Genomic Research

Proceedings of international conference on intelligent systems and new applications., Год журнала: 2024, Номер unknown

Опубликована: Апрель 28, 2024

Breast cancer, a heterogeneous disease, can be classified into several subtypes, each associated with distinct genetic mutations and clinical outcomes. As per the research article by National Institutes of Health it was stated that 25% hereditary cases are due to mutation highly penetrant genes which leads 80% lifetime risk breast cancer [15]. This study aims apply advanced deep learning machine algorithms predict subtypes identify key contributing disease using comprehensive gene expression dataset. We analyzed dataset comprising 1904 samples, encompassing 331 175 mutations, sourced from public platform including PAM50 Claudin low categorizations. Due limited observations, SMOTE employed for data augmentation, Principal Component Analysis (PCA) used assess variance. Several models, Random Forest Classifier, Support Vector Machine, K-Nearest Neighbor, XGBoost, Stacked were applied alongside techniques like Convolutional Neural Network Multi-Layer Perceptron. The model demonstrated superior performance an accuracy 0.955, outperforming other models. models achieved accuracies 0.911 (CNN) 0.936 (MLP). KNN analysis revealed potential clusters based on data, silhouette metric identifying "siah1_mut," "nras_mut," "hras_mut" as significant mutations. optimal clustering score 0.997 two clusters. These may play pivotal roles in pathogenesis could serve targets therapeutic interventions. Our findings demonstrate effectiveness integrating stacked predicting subtypes. identification through provides valuable insights underpinnings guide future development targeted therapies. highlights computational approaches elucidating complex landscape genomics paves way personalized medicine oncology.

Язык: Английский

---

The Impact of Artificial Intelligence On the Quality of Healthcare Services in Saudi Arabia: A Systematic Review. (Preprint)

Опубликована: Ноя. 24, 2024

: Artificial intelligence is a rapidly evolving technology with the potential to revolutionize healthcare industry. In Saudi Arabia, sector has adopted AI technologies over past decade enhance service efficiency and quality, aligning Vision 2030's technological thrust. This review aims systematically examine intelligence's impact on quality in Arabia hospitals A meticulous comprehensive systematic literature was undertaken identify studies investigating AI's by collecting several articles from selected databases , PubMed, Google Scholar, Digital Library databases. The search terms used were "Artificial Intelligence", healthcare, health care Quality (AI) Arabia,(AI) providers. focused published last ten years, ensuring inclusion of most recent relevant research effects organizations. included quantitative qualitative analyses, providing robust understanding topic 12 this review.The findings suggest that significantly improved diagnostic accuracy, patient management, operational within system. have been various areas, such as radiology, cardiology, pathology. However, also highlights challenges data privacy, algorithmic bias, need for regulatory frameworks. underscored importance ongoing monitoring rigorous training personnel applications. Intelligence transform improving outcomes streamlining operations. can aid predicting outcomes, personalizing medicine, enhancing administrative efficiency. making necessary infrastructure investments, addressing biases algorithms must be addressed. benefit through better disease outbreak predictions, advanced medical professionals, support education. rise telehealth digital strategies important role future region. To maximize benefits AI, it critical tackle associated challenges. With proper implementation training, lead more accurate cost-effective services. NO Trial

Язык: Английский

---

A Systematic Literature Review on Mammography: Deep Learning in Redefining Breast Cancer Diagnosis for the Asian Perspective

Research Square (Research Square), Год журнала: 2024, Номер unknown

Опубликована: Апрель 2, 2024

Breast cancer (BC) is a leading cause of mortality in women worldwide, and its incidence prognosis vary greatly, especially low-income areas such as Asia Africa, compared to America Europe. This review examines the critical role computer-aided diagnostic systems utilizing Deep Learning (DL) techniques improving precision BC detection, which can help researchers practitioners better understand obstacles emerging trends field. The comprehensive analysis synthesis published articles provided by Preferred Reporting Items for Systematic Reviews Meta- Analyses (PRISMA). Following several searches, 287 were found eligible assessment followed data extraction synthesis. work provides insight into pathology corresponding mammographic appearance, publication trends, important contributors, themes using bibliometric addition thorough research articles. Focus maps also created identify shed light on body knowledge. In contrast other research, this sheds subtle identification breast density, mass, calcification from mammography images emphasizing evaluation image pre- processing, augmentation, segmentation, classification used 2018 2023, further highlighting related Asian dataset. Furthermore, contributions made authors, countries, experimental datasets, providing insights various approaches. Prospero registration: CRD42022478896

Язык: Английский

---