Creating a deep learning model using a Swin Transformer and tree growth optimisation to classify brain tumour DOI Open Access
K Sankar,

V. Gokula Krishnan,

Saurav Kumar

et al.

Review of Computer Engineering Research, Journal Year: 2023, Volume and Issue: 10(3), P. 110 - 121

Published: Oct. 16, 2023

The brain, which has billions of cells, is the largest and most complex organ in human body. A brain tumor primary malignant intracranial central nervous system that develops frequently. They are frequently found too late for effective therapy. use minimally invasive procedures necessary to make a diagnosis monitor system's response There exist three distinct classifications tumors, namely benign, premalignant, malignant. This study concentrated on using deep learning identify tumors (BT) normal or abnormal pictures. Numerous methodologies have been employed augment quality images, encompassing image smoothing noise restoration procedures. present employs proposed Adaptive Weighted Frost filter as it identified optimal approach reduction BT photographs. Swin Transformer technology purpose classifying BT. efficiency Tree Growth Optimization (TGA) model transformer hyper parameter tweaking evaluated this work. Before our unique dataset extensive experimental comparisons, medical specialists carefully examined down pixel level. predicted achieved greatest F1 score 99.82% maximum accuracy, recall, 100%, respectively.

Language: Английский

A novel Swin transformer approach utilizing residual multi-layer perceptron for diagnosing brain tumors in MRI images DOI Creative Commons
İshak Paçal

International Journal of Machine Learning and Cybernetics, Journal Year: 2024, Volume and Issue: 15(9), P. 3579 - 3597

Published: March 5, 2024

Abstract Serious consequences due to brain tumors necessitate a timely and accurate diagnosis. However, obstacles such as suboptimal imaging quality, issues with data integrity, varying tumor types stages, potential errors in interpretation hinder the achievement of precise prompt diagnoses. The rapid identification plays pivotal role ensuring patient safety. Deep learning-based systems hold promise aiding radiologists make diagnoses swiftly accurately. In this study, we present an advanced deep learning approach based on Swin Transformer. proposed method introduces novel Hybrid Shifted Windows Multi-Head Self-Attention module (HSW-MSA) along rescaled model. This enhancement aims improve classification accuracy, reduce memory usage, simplify training complexity. Residual-based MLP (ResMLP) replaces traditional Transformer, thereby improving speed, parameter efficiency. We evaluate Proposed-Swin model publicly available MRI dataset four classes, using only test data. Model performance is enhanced through application transfer augmentation techniques for efficient robust training. achieves remarkable accuracy 99.92%, surpassing previous research models. underscores effectiveness Transformer HSW-MSA ResMLP improvements innovative diagnostic offering support diagnosis, ultimately outcomes reducing risks.

Language: Английский

Citations

47
Advancing multiple sclerosis diagnosis through an innovative hybrid AI framework incorporating Multi-view ResNet and quantum RIME-inspired metaheuristics DOI Creative Commons

Mohamed G. Khattap,

Mohammed Sallah, Abdelghani Dahou

et al.

Ain Shams Engineering Journal, Journal Year: 2025, Volume and Issue: 16(2), P. 103241 - 103241

Published: Jan. 13, 2025

Language: Английский

Citations

1
Emerging Trends in Fast MRI Using Deep-Learning Reconstruction on Undersampled k-Space Data: A Systematic Review DOI Creative Commons
Dilbag Singh, Anmol Monga, Hector Lise de Moura

et al.

Bioengineering, Journal Year: 2023, Volume and Issue: 10(9), P. 1012 - 1012

Published: Aug. 26, 2023

Magnetic Resonance Imaging (MRI) is an essential medical imaging modality that provides excellent soft-tissue contrast and high-resolution images of the human body, allowing us to understand detailed information on morphology, structural integrity, physiologic processes. However, MRI exams usually require lengthy acquisition times. Methods such as parallel Compressive Sensing (CS) have significantly reduced time by acquiring less data through undersampling k-space. The state-of-the-art fast has recently been redefined integrating Deep Learning (DL) models with these undersampled approaches. This Systematic Literature Review (SLR) comprehensively analyzes deep reconstruction models, emphasizing key elements proposed methods highlighting their strengths weaknesses. SLR involves searching selecting relevant studies from various databases, including Web Science Scopus, followed a rigorous screening extraction process using Preferred Reporting Items for Reviews Meta-Analyses (PRISMA) guidelines. It focuses techniques, residual learning, image representation encoders decoders, data-consistency layers, unrolled networks, learned activations, attention modules, plug-and-play priors, diffusion Bayesian methods. also discusses use loss functions training adversarial networks enhance Moreover, we explore applications, non-Cartesian reconstruction, super-resolution, dynamic MRI, joint learning coil sensitivity sampling, quantitative mapping, MR fingerprinting. paper addresses research questions, insights future directions, emphasizes robust generalization artifact handling. Therefore, this serves valuable resource advancing guiding development efforts better quality faster acquisition.

Language: Английский

Citations

22
Atom search optimization: a systematic review of current variants and applications DOI
Sylvère Mugemanyi, Zhaoyang Qu, François Xavier Rugema

et al.

Knowledge and Information Systems, Journal Year: 2025, Volume and Issue: unknown

Published: April 12, 2025

Language: Английский

Citations

0
Improving Brain Tumor Classification: An Approach Integrating Pre-Trained CNN Models and Machine Learning Algorithms DOI Creative Commons
Mohamed R. Shoaib, Jun Zhao, Heba M. Emara

et al.

Heliyon, Journal Year: 2024, Volume and Issue: unknown, P. e33471 - e33471

Published: June 1, 2024

Accurate detection of brain tumors is crucial for enhancing patient outcomes, yet the interpretation Magnetic Resonance Imaging (MRI) scans poses significant challenges. This study introduces a novel approach to tumor classification by exploring three pre-trained convolutional neural network (CNN) models: DenseNet201, EfficientNetB5, and InceptionResNetV2, combined with softmax activation feature extraction. These features are then subjected Principal Component Analysis (PCA) dimensionality reduction. Subsequently, machine learning models—Support Vector Machine (SVM), Multi-layer Perceptron (MLP), Gaussian Naive Bayes (GNB)—are employed classification. The results reveal that when SVM MLP, outperforms other models in terms accuracy, recall, precision. Specifically, DenseNet201 achieves 100% precision on Dataset-I 98% Dataset-II paired MLP. provides valuable insights into interplay between CNN models, extraction techniques, algorithms classification, highlighting efficacy

Language: Английский

Citations

3
Cross prior Bayesian attention with correlated inception and residual learning for brain tumor classification using MR images (CB-CIRL Net) DOI

B. Vijayalakshmi,

Sam Anand

Journal of Neuroscience Methods, Journal Year: 2025, Volume and Issue: unknown, P. 110392 - 110392

Published: Feb. 1, 2025

Language: Английский

Citations

0
OptEF-BCI: An Optimization-Based Hybrid EEG and fNIRS–Brain Computer Interface DOI Creative Commons
Muhammad Umair Ali,

Kwang Su Kim,

Karam Dad Kallu

et al.

Bioengineering, Journal Year: 2023, Volume and Issue: 10(5), P. 608 - 608

Published: May 18, 2023

Multimodal data fusion (electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS)) has been developed as an important neuroimaging research field in order to circumvent the inherent limitations of individual modalities by combining complementary information from other modalities. This study employed optimization-based feature selection algorithm systematically investigate nature multimodal fused features. After preprocessing acquired both (i.e., EEG fNIRS), temporal statistical features were computed separately with a 10 s interval for each modality. The create training vector. A wrapper-based binary enhanced whale optimization (E-WOA) was used select optimal/efficient subset using support-vector-machine-based cost function. An online dataset 29 healthy individuals evaluate performance proposed methodology. findings suggest that approach enhances classification evaluating degree complementarity between characteristics selecting most efficient subset. E-WOA showed high rate (94.22 ± 5.39%). exhibited 3.85% increase compared conventional algorithm. hybrid framework outperformed traditional (p < 0.01). These indicate potential efficacy several neuroclinical applications.

Language: Английский

Citations

7
Enhancing Skin Lesion Detection: A Multistage Multiclass Convolutional Neural Network-Based Framework DOI Creative Commons
Muhammad Umair Ali,

Majdi Khalid,

Hanan Alshanbari

et al.

Bioengineering, Journal Year: 2023, Volume and Issue: 10(12), P. 1430 - 1430

Published: Dec. 15, 2023

The early identification and treatment of various dermatological conditions depend on the detection skin lesions. Due to advancements in computer-aided diagnosis machine learning approaches, learning-based lesion analysis methods have attracted much interest recently. Employing concept transfer learning, this research proposes a deep convolutional neural network (CNN)-based multistage multiclass framework categorize seven types In first stage, CNN model was developed classify images into two classes, namely benign malignant. second then used with further lesions five subcategories (melanocytic nevus, actinic keratosis, dermatofibroma, vascular) malignant (melanoma basal cell carcinoma). frozen weights developed-trained correlated benefited using same type for subclassification classes. proposed technique improved classification accuracy online ISIC2018 dataset by up 93.4% class identification. Furthermore, high 96.2% achieved both Sensitivity, specificity, precision, F1-score metrics validated effectiveness framework. Compared existing models described literature, approach took less time train had higher rate.

Language: Английский

Citations

7
An Adaptation of Hybrid Binary Optimization Algorithms for Medical Image Feature Selection in Neural Network for Classification of Breast Cancer DOI
Olaide N. Oyelade, Enesi Femi Aminu, Hui Wang

et al.

Neurocomputing, Journal Year: 2024, Volume and Issue: unknown, P. 129018 - 129018

Published: Nov. 1, 2024

Language: Английский

Citations

2
Enhanced Magnetic Resonance Imaging-Based Brain Tumor Classification with a Hybrid Swin Transformer and ResNet50V2 Model DOI Creative Commons

Abeer Fayez Al Bataineh,

Khalid M.O. Nahar, Hayel Khafajeh

et al.

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(22), P. 10154 - 10154

Published: Nov. 6, 2024

Brain tumors can be serious; consequently, rapid and accurate detection is crucial. Nevertheless, a variety of obstacles, such as poor imaging resolution, doubts over the accuracy data, lack diverse tumor classes stages, possibility misunderstanding, present challenges to achieve an final diagnosis. Effective brain cancer crucial for patients’ safety health. Deep learning systems provide capability assist radiologists in quickly accurately detecting diagnoses. This study presents innovative deep approach that utilizes Swin Transformer. The suggested method entails integrating Transformer with pretrained model Resnet50V2, called (SwT+Resnet50V2). objective this modification decrease memory utilization, enhance classification accuracy, reduce training complexity. self-attention mechanism identifies distant relationships captures overall context. Resnet 50V2 improves both speed by extracting adaptive features from Transformer’s dependencies. We evaluate proposed framework using two publicly accessible magnetic resonance (MRI) datasets, each including four distinct classes, respectively. Employing data augmentation transfer techniques enhances performance, leading more dependable cost-effective training. achieves impressive 99.9% on binary-labeled dataset 96.8% four-labeled dataset, outperforming VGG16, MobileNetV2, EfficientNetV2B3, ConvNeXtTiny, convolutional neural network (CNN) algorithms used comparison. demonstrates transducer, when combined capable diagnosing tumors. leverages combination SwT+Resnet50V2 create diagnostic tool. Radiologists have potential accelerate improve tumors, improved patient outcomes reduced risks.

Language: Английский

Citations

1