A Novel Encoder Decoder Architecture with Vision Transformer for Medical Image Segmentation

Journal of Electronics Electromedical Engineering and Medical Informatics, Journal Year: 2025, Volume and Issue: 7(1), P. 176 - 186

Published: Jan. 8, 2025

Brain tumor image segmentation is one of the most critical tasks in medical imaging for diagnosis, treatment planning, and prognosis. Traditional methods brain are mostly based on Convolution Neural Network (CNN), which have been proved very powerful but still limitations to effectively capture long-range dependencies complex spatial hierarchies MRI images. Variability shape, size, location tumors may affect performance get stuck into suboptimal outcomes. In these regards, new encoder-decoder architecture with VisionTranscoder(ViT) proposed, enhance detection classification. The proposed VisionTranscoder exploits a transformer's ability modeling global context through self-attention mechanisms, providing more inclusive interpretation intricate patterns images classification by capturing both local features. maintains Vision Transformer its encoder processing as sequences patches often outside view traditional CNNs. Then map rebuilt at high level fidelity decoder upsampling skips connections maintain detailed information. risk overfitting hugely reduced design advanced regularization techniques extensive data augmentation. dataset contains 7,023 human images, all four different classes: glioma, meningioma, no tumor, pituitary. Images from 'no tumor' class, indicating an scan without any detectable were taken Br35H . results show efficiency over wide set scans, producing accuracy 98.5% loss 0.05. This underlines it accurately segment classify overfitting.

Language: Английский

---

Collaborative Healthcare Data Management Framework using Parallel Computing and the Internet of Things

Journal of Electronics Electromedical Engineering and Medical Informatics, Journal Year: 2025, Volume and Issue: 7(1), P. 187 - 196

Published: Jan. 9, 2025

Healthcare data management has become a critical research area, primarily driven by the widespread adoption of personal health monitoring systems and applications. These generate an immense volume data, necessitating efficient reliable solutions for lossless sharing. This article introduces Collaborative Data Management Framework (CDMF) that leverages combined strengths parallel computing federated learning. The proposed CDMF is designed to achieve two primary objectives: reducing computational complexity in handling ensuring high sharing accuracy, regardless generation rate. framework employs streamline scheduling processing acquired at various intervals. approach minimizes delays operating on less complex algorithm, making it suitable high-frequency generation. Federated learning, other hand, plays pivotal role verifying distribution maintaining accuracy. By enabling decentralized learning ensures remains local devices while only necessary model updates. enhances privacy security, consideration healthcare management. It are verified based appropriate requests avoiding latency issues. decentralizing process, as raw does not leave systems. cooperative interaction between operates cyclic manner, allowing adapt dynamically increasing intervals varying rates. performance validated through improvements key metrics. First, achieves 15.08% enhancement which vital integrity reliability during transfers. Second, reduces computation 9.48%, even when maximum results highlight framework’s potential revolutionize addressing dual challenges scalability

Language: Английский

---

Uncertainty-guided and cross-modality attention network for liver tumor segmentation and quantification via integrating dynamic MRI

Knowledge-Based Systems, Journal Year: 2025, Volume and Issue: unknown, P. 113021 - 113021

Published: Jan. 1, 2025

Language: Английский

---

Harnessing Artificial Intelligence in Early Detection and Diagnosis of Alzheimer's Disease: Current and Future Applications

Indus journal of bioscience research., Journal Year: 2025, Volume and Issue: 3(2), P. 199 - 212

Published: Feb. 25, 2025

Alzheimer's Disease (AD) is a neurodegenerative disorder requiring early detection. This study compares AI models—Convolutional Neural Networks (CNN), Support Vector Machines (SVM), and Random Forest (RF)—in analyzing neuroimaging data (MRI, PET) to enhance AD prediction improve diagnosis using machine learning techniques. Through the application of multi-modal in form genetic, clinical, data, also investigates effectiveness combining different types predictability models for diagnosis. Feature importance analysis was performed methods like SHAP (SHAP (Shapley Additive Explanations) values determine most important variables model predictions, e.g., certain brain regions or genetic components. The generalizability real-world applicability by training on an independent dataset representing diverse clinical settings. performance each assessed variety statistical measures accuracy, precision, recall, F1-score, Area Under Curve (AUC). findings showed that CNN better compared SVM RF all metrics with highest accuracy (92%), precision (93%), recall (91%), AUC (0.95). suggest effectively detects subtle patterns, making it strong tool While well, superior accuracy. Cross-validation confirmed its generalizability, crucial use. Implementing models, especially CNN, may enable earlier detection, timely interventions, improved patient outcomes Alzheimer’s care. References

Language: Английский

---

Recent Advancements in Neuroimaging‐Based Alzheimer's Disease Prediction Using Deep Learning Approaches in e‐Health: A Systematic Review

Health Science Reports, Journal Year: 2025, Volume and Issue: 8(5)

Published: May 1, 2025

ABSTRACT Purpose Alzheimer's disease (AD) is a severe neurological that significantly impairs brain function. Timely identification of AD essential for appropriate treatment and care. This comprehensive review intends to examine current developments in deep learning (DL) approaches with neuroimaging diagnosis, where popular imaging types, reviews well‐known online accessible data sets, describes different algorithms used DL the correct initial evaluation are presented. Significance Conventional diagnostic techniques, including medical evaluations cognitive assessments, usually not identify stages Alzheimer's. Neuroimaging methods, when integrated have demonstrated considerable potential enhancing diagnosis categorization AD. models received significant interest due their capability its early phases automatically, which reduces mortality rate cost Method An extensive literature search was performed leading scientific databases, concentrating on papers published from 2021 2025. Research leveraging techniques such as magnetic resonance (MRI), positron emission tomography, functional (fMRI), so forth. The complies Preferred Reporting Items Systematic Reviews Meta‐Analyses (PRISMA) guidelines. Results Current show CNN‐based especially those utilizing hybrid transfer frameworks, outperform conventional methods. employing combination multimodal has enhanced precision. Still, challenges method interpretability, heterogeneity, limited exist issues. Conclusion considerably improved accuracy reliability neuroimaging. Regardless issues accessibility adaptability, studies into interpretability fusion provide clinical application. Further research should concentrate standardized rigorous validation architectures, understandable AI methodologies enhance effectiveness methods prediction.

Language: Английский

---

Enhancing Aspect-Based Sentiment Analysis Through Multi-Granularity Information Sharing

Published: July 26, 2024

Language: Английский

---

Boundary Feature-Based Leaf Disease Detection Using Differential Network

Published: July 26, 2024

Language: Английский

---

Novel Automatic Classification of Human Adult Lung Alveolar Type II Cells Infected with SARS-CoV-2 through the Deep Transfer Learning Approach

Mathematics, Journal Year: 2024, Volume and Issue: 12(10), P. 1573 - 1573

Published: May 17, 2024

Transmission electron microscopy imaging provides a unique opportunity to inspect the detailed structure of infected lung cells with SARS-CoV-2. Unlike previous studies, this novel study aims investigate COVID-19 classification at cellular level in response Particularly, differentiating between healthy and human alveolar type II (hAT2) Hence, we explore feasibility deep transfer learning (DTL) introduce highly accurate approach that works as follows: First, downloaded processed 286 images pertaining hAT2 obtained from public image archive. Second, provided two DTL computations induce ten models. The first computation employs five pre-trained models (including DenseNet201 ResNet152V2) trained on more than one million ImageNet database extract features images. Then, it flattens output feature vectors trained, densely connected classifier Adam optimizer. second similar manner, minor difference freeze layers for extraction while unfreezing jointly training next layers. results using five-fold cross-validation demonstrated TFeDenseNet201 is 12.37× faster superior, yielding highest average ACC 0.993 (F1 0.992 MCC 0.986) statistical significance (P<2.2×10−16 t-test) compared an 0.937 0.938 0.877) counterpart (TFtDenseNet201), showing no (P=0.093 t-test).

Language: Английский

---

WIMOAD: Weighted Integration of Multi-omics data for Alzheimer's Disease (AD) Diagnosis

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 27, 2024

As the most common subtype of dementia, Alzheimer's disease (AD) is characterized by a progressive decline in cognitive functions, especially memory, thinking, and reasoning ability. Early diagnosis interventions enable implementation measures to reduce or slow further regression disease, preventing individuals from severe brain function decline. The current framework AD depends on A/T/(N) biomarkers detection cerebrospinal fluid imaging data, which invasive expensive during data acquisition process. Moreover, pathophysiological changes accumulate amino acids, metabolism, neuroinflammation, etc., resulting heterogeneity newly registered patients. Recently, next generation sequencing (NGS) technologies have found be non-invasive, efficient less-costly alternative screening. However, existing studies rely single omics only. To address these concerns, we introduce WIMOAD, weighted integration multi-omics for diagnosis. WIMOAD synergistically leverages specialized classifiers patients' paired gene expression methylation multi-stage classification. scores were then stacked with MLP-based meta-models performance improvement. prediction results two distinct integrated optimized weights final decision-making model, providing higher than using Remarkably, achieves significantly alone classification tasks. model's overall also outperformed approaches, highlighting its ability effectively discern intricate patterns their correlations clinical results. In addition, stands out as biologically interpretable model leveraging SHapley Additive exPlanations (SHAP) elucidate contributions each output. We believe very promising tool accurate effective biomarker discovery across different progression stages, eventually will consequential impacts early treatment intervention personalized therapy design AD.

Language: Английский

---