PatchResNet: Multiple Patch Division–Based Deep Feature Fusion Framework for Brain Tumor Classification Using MRI Images

Journal of Digital Imaging, Год журнала: 2023, Номер 36(3), С. 973 - 987

Опубликована: Фев. 16, 2023

Язык: Английский

---

Enhancing EfficientNetv2 with global and efficient channel attention mechanisms for accurate MRI-Based brain tumor classification

Cluster Computing, Год журнала: 2024, Номер 27(8), С. 11187 - 11212

Опубликована: Май 20, 2024

Abstract The early and accurate diagnosis of brain tumors is critical for effective treatment planning, with Magnetic Resonance Imaging (MRI) serving as a key tool in the non-invasive examination such conditions. Despite advancements Computer-Aided Diagnosis (CADx) systems powered by deep learning, challenge accurately classifying from MRI scans persists due to high variability tumor appearances subtlety early-stage manifestations. This work introduces novel adaptation EfficientNetv2 architecture, enhanced Global Attention Mechanism (GAM) Efficient Channel (ECA), aimed at overcoming these hurdles. enhancement not only amplifies model’s ability focus on salient features within complex images but also significantly improves classification accuracy tumors. Our approach distinguishes itself meticulously integrating attention mechanisms that systematically enhance feature extraction, thereby achieving superior performance detecting broad spectrum Demonstrated through extensive experiments large public dataset, our model achieves an exceptional high-test 99.76%, setting new benchmark MRI-based classification. Moreover, incorporation Grad-CAM visualization techniques sheds light decision-making process, offering transparent interpretable insights are invaluable clinical assessment. By addressing limitations inherent previous models, this study advances field medical imaging analysis highlights pivotal role enhancing interpretability learning models diagnosis. research sets stage advanced CADx systems, patient care outcomes.

Язык: Английский

---

Brain Glial Cell Tumor Classification through Ensemble Deep Learning with APCGAN Augmentation

International Journal of Computational and Experimental Science and Engineering, Год журнала: 2025, Номер 11(1)

Опубликована: Янв. 5, 2025

Classification of brain tumor plays a vital role in medical imaging for accurate diagnosis, treatment, and monitoring. Deep learning approaches have gained significant traction this industry because their ability to extract relevant features from images. The research suggests employing an ensemble classifier with weighted voting mechanism categorize glial cell malignancies such as Astrocytoma, Glioblastoma multiforme, Oligodendroglioma, Ependymoma. proposed technique employs three main classifiers: Convolutional Neural Network (CNN), Long Short Term Memory (C-LSTM), + Conditional Random Fields (DCNN+CRF). algorithms require huge amount input data avoid overfitting. Adaptive Progressive Generative Adversarial Networks (APCGANs) are used produce realistic artificial images efficiently train the methodology. Overall, method strategy consistently outperforms other tested (CNN, C-LSTM, DCNN+CRF). Ensemble attained accuracy 99.4 %, recall - 99.1%, precision- 98.0%, F1-score 99.2%. demonstrates superior performance accurately classifying tumors, making it promising algorithm analysis tasks.

Язык: Английский

---

A Review of Recent Advances in Brain Tumor Diagnosis Based on AI-Based Classification

Diagnostics, Год журнала: 2023, Номер 13(18), С. 3007 - 3007

Опубликована: Сен. 20, 2023

Uncontrolled and fast cell proliferation is the cause of brain tumors. Early cancer detection vitally important to save many lives. Brain tumors can be divided into several categories depending on kind, place origin, pace development, stage progression; as a result, tumor classification crucial for targeted therapy. segmentation aims delineate accurately areas A specialist with thorough understanding illnesses needed manually identify proper type tumor. Additionally, processing images takes time tiresome. Therefore, automatic techniques are required speed up enhance diagnosis Tumors quickly safely detected by scans using imaging modalities, including computed tomography (CT), magnetic resonance (MRI), others. Machine learning (ML) artificial intelligence (AI) have shown promise in developing algorithms that aid utilizing various modalities. The right method must used precisely classify patients treatment. This review describes multiple types tumors, publicly accessible datasets, enhancement methods, segmentation, feature extraction, classification, machine techniques, deep learning, through transfer study In this study, we attempted synthesize modalities automatically computer-assisted methodologies characterization ML DL frameworks. Finding current problems engineering currently use predicting future paradigm other goals article.

Язык: Английский

---

ResUNet+: A New Convolutional and Attention Block-Based Approach for Brain Tumor Segmentation

IEEE Access, Год журнала: 2023, Номер 11, С. 69884 - 69902

Опубликована: Янв. 1, 2023

The number of brain tumor cases has increased in recent years. Therefore, accurate diagnosis and treatment tumors are extremely important. Accurate detection regions is difficult, even for experts, because images low-contrast, noisy contain normal tissue-like structures. this study, a new convolution-based hybrid model was proposed to perform segmentation with high accuracy. In the model, instead applying convolution whole image, applied ROI detected different modalities. With approach, it determined that processing cost reduced, performance increased. tested on BraTS 2020, 2019, 2018 datasets. method study also compared SOTA methods using same dataset. As result comparison, dice scores 92.80%, 93.10%, 91.90% were respectively obtained tumors, enhance nuclei 2020 these results, can compete many models literature be preferred applications due its success especially advantage pre-processing structure.

Язык: Английский

---

An Ensemble Model for the Diagnosis of Brain Tumors through MRIs

Diagnostics, Год журнала: 2023, Номер 13(3), С. 561 - 561

Опубликована: Фев. 3, 2023

Automatic brain tumor detection in MR Images is one of the basic applications machine vision medical image processing, which, despite much research, still needs further development. Using multiple learning techniques as an ensemble system solutions that can be effective achieving this goal. In paper, a novel method for diagnosing tumors by combining data mining and has been proposed. proposed method, each initially pre-processed to eliminate its background region identify tissue. The Social Spider Optimization (SSO) algorithm then utilized segment MRI Images. segmentation allows more precise identification image. next step, distinctive features are extracted using SVD technique. addition removing redundant information, strategy boosts speed processing at classification stage. Finally, combination algorithms Naïve Bayes, Support vector K-nearest neighbor used classify detect tumors. Each three performs feature individually, final output model created integrating independent outputs voting results. results indicate diagnose BRATS 2014 dataset with average accuracy 98.61%, sensitivity 95.79% specificity 99.71%. Additionally, could BTD20 database 99.13%, 99% 99.26%. These show significant improvement compared previous efforts. findings confirm technique, well learning, improving efficiency method.

Язык: Английский

---

Exploring the efficacy of multi-flavored feature extraction with radiomics and deep features for prostate cancer grading on mpMRI

BMC Medical Imaging, Год журнала: 2023, Номер 23(1)

Опубликована: Ноя. 22, 2023

Abstract Background The purpose of this study is to investigate the use radiomics and deep features obtained from multiparametric magnetic resonance imaging (mpMRI) for grading prostate cancer. We propose a novel approach called multi-flavored feature extraction or tensor, which combines four mpMRI images using eight different fusion techniques create 52 datasets each patient. evaluate effectiveness in cancer compare it traditional methods. Methods used PROSTATEx-2 dataset consisting 111 patients’ T2W-transverse, T2W-sagittal, DWI, ADC images. merge T2W, images, namely Laplacian Pyramid, Ratio low-pass pyramid, Discrete Wavelet Transform, Dual-Tree Complex Curvelet Fusion, Weighted Principal Component Analysis. Prostate were manually segmented, extracted Pyradiomics library Python. also an Autoencoder extraction. five sets train classifiers: all features, linked with PCA, combination features. processed data, including balancing, standardization, correlation, Least Absolute Shrinkage Selection Operator (LASSO) regression. Finally, we nine classifiers classify Gleason grades. Results Our results show that SVM classifier PCA achieved most promising results, AUC 0.94 balanced accuracy 0.79. Logistic regression performed best when only 0.93 0.76. Gaussian Naive Bayes had lower performance compared other classifiers, while KNN high PCA. Random Forest well achieving Voting showed higher 2 highest performance, 0.95 0.78. Conclusion concludes proposed tensor can be effective method findings suggest may more than alone accurately classifying

Язык: Английский

---

Grade Classification of Tumors from Brain Magnetic Resonance Images Using a Deep Learning Technique

Diagnostics, Год журнала: 2023, Номер 13(6), С. 1153 - 1153

Опубликована: Март 17, 2023

To improve the accuracy of tumor identification, it is necessary to develop a reliable automated diagnostic method. In order precisely categorize brain tumors, researchers developed variety segmentation algorithms. Segmentation images generally recognized as one most challenging tasks in medical image processing. this article, novel detection and classification method was proposed. The proposed approach consisted many phases, including pre-processing MRI images, segmenting extracting features, classifying images. During portion an scan, adaptive filter utilized eliminate background noise. For feature extraction, local-binary grey level co-occurrence matrix (LBGLCM) used, for segmentation, enhanced fuzzy c-means clustering (EFCMC) used. After scan we used deep learning model classify into two groups: glioma normal. classifications were created using convolutional recurrent neural network (CRNN). technique improved from defined input dataset. scans REMBRANDT dataset, which 620 testing 2480 training sets, research. data demonstrate that newly outperformed its predecessors. CRNN strategy compared against BP, U-Net, ResNet, are three prevalent approaches currently being classification, system outcomes 98.17% accuracy, 91.34% specificity, 98.79% sensitivity.

Язык: Английский

---

DEHA-Net: A Dual-Encoder-Based Hard Attention Network with an Adaptive ROI Mechanism for Lung Nodule Segmentation

Sensors, Год журнала: 2023, Номер 23(4), С. 1989 - 1989

Опубликована: Фев. 10, 2023

Measuring pulmonary nodules accurately can help the early diagnosis of lung cancer, which increase survival rate among patients. Numerous techniques for nodule segmentation have been developed; however, most them either rely on 3D volumetric region interest (VOI) input by radiologists or use 2D fixed (ROI) all slices computed tomography (CT) scan. These methods only consider presence within given VOI, limits networks’ ability to detect outside VOI and also encompass unnecessary structures in leading potentially inaccurate segmentation. In this work, we propose a novel approach that utilizes inputted from radiologist computer-aided detection (CADe) system. Concretely, developed two-stage technique. Firstly, designed dual-encoder-based hard attention network (DEHA-Net) full axial slice thoracic scan, along with an ROI mask, were considered as segment slice. The output DEHA-Net, mask nodule, was adaptive (A-ROI) algorithm automatically generate masks surrounding slices, eliminated need any further inputs radiologists. After extracting axis, at second stage, investigated sagittal coronal views employing DEHA-Net. All estimated into consensus module obtain final nodule. proposed scheme rigorously evaluated image database consortium resource initiative (LIDC/IDRI) dataset, extensive analysis results performed. quantitative showed method not improved existing state-of-the-art terms dice score but significant robustness against different types, shapes, dimensions nodules. framework achieved average score, sensitivity, positive predictive value 87.91%, 90.84%, 89.56%, respectively.

Язык: Английский

---

Developments in Brain Tumor Segmentation Using MRI: Deep Learning Insights and Future Perspectives

IEEE Access, Год журнала: 2024, Номер 12, С. 26875 - 26896

Опубликована: Янв. 1, 2024

The human brain is an incredible and wonderful organ that governs all body actions. Due to its great importance, any defect in the shape of regions should be reported quickly reduce death rate. abnormal region segmentation helps plan monitor treatment. most critical procedure isolating normal tissues from each other. So far, remarkable imaging modalities are being used diagnose abnormalities at their early stages, magnetic resonance (MRI) renowned noninvasive among those modalities. This paper investigates current landscape tumor (BTS) by exploring emerging deep learning (DL) methods for MRI analysis. findings offer a comprehensive comparison recent DL approaches, emphasizing effectiveness handling diverse types while addressing limitations associated with data scarcity robust validation. has shown vital improvement BTS, so our primary focus include significant models analyze MRI. However, outperforms traditional methods; still, there several limitations, especially related types, lack datasets, weak validations. future perspectives DL-based BTS present potential revolutionizing diagnosis treatment tumors.

Язык: Английский

---