Aug 2026· Al-Farooq Journal of Sciences· Vol 2, pp. 1452-1470· 0 citations· 25 references
TL;DR
This comprehensive review systematically examines the architecture, functionality, and clinical effectiveness of ANN-based models applied to a diverse spectrum of medical imaging modalities, encompassing mammography, magnetic resonance imaging (MRI), computed tomography (CT), fundus photography, and dermoscopy.
Abstract
The integration of artificial neural networks (ANNs), particularly convolutional neural networks (CNNs), into medical imaging has fundamentally transformed the paradigm of early disease diagnosis. This comprehensive review systematically examines the architecture, functionality, and clinical effectiveness of ANN-based models applied to a diverse spectrum of medical imaging modalities, encompassing mammography, magnetic resonance imaging (MRI), computed tomography (CT), fundus photography, and dermoscopy. Evidence synthesized from the current literature demonstrates that state-of-the-art deep learning architectures—including ResNet, VGG, DenseNet, InceptionV3, and U-Net—consistently achieve diagnostic accuracy exceeding 93% across oncological, neurological, ophthalmological, and pulmonary applications. Noteworthy findings include a 98.2% classification accuracy for brain tumor MRI analysis, 97.4% for breast cancer mammographic detection, and area under the receiver operating characteristic curve (AUC) values surpassing 0.97 in multi-class diagnostic tasks. The paper further critically evaluates the comparative performance of leading architectures through structured tables and graphical analyses, highlights the role of transfer learning in addressing data scarcity, and identifies persistent challenges including model interpretability, data heterogeneity, class imbalance, and regulatory compliance barriers. Future directions toward federated learning, explainable AI (XAI), and multimodal fusion frameworks are discussed as pathways to clinically deployable ANN-based diagnostic tools.
Brain tumors represent one of the most life-threatening diseases, and their early and accurate detection is critical for improving patient outcomes. Magnetic Resonance Imaging (MRI) is the most reliable imaging technique for identifying brain tumors, yet manual interpretation by radiologists is time-consuming and prone to errors. To address these challenges, this study investigates the application of deep learning architectures, including Convolutional Neural Network (CNN), VGG16, VGG19, ResNet50, and MobileNet, for brain tumor detection and classification. A publicly available MRI dataset consisting of glioma, meningioma, pituitary, and no-tumor cases was used. The models were trained and evaluated using accuracy, precision, recall, F1-score, ROC curves, and confusion matrices, while interpretability was assessed using Local Interpretable Model-Agnostic Explanations (LIME). Experimental results demonstrate that ResNet50 achieved the highest performance with 96.9% accuracy, followed closely by MobileNet at 96.6%, whereas CNN performed less effectively at 87.9%. The findings confirm that advanced deep learning architectures not only achieve high classification accuracy but also improve interpretability, thereby offering reliable and clinically applicable solutions for automated brain tumor diagnosis.
H. Uzel, Feyyaz Alpsalaz, Yıldırım Özüpak et al.· Computers and Electronics in...· 0 citations
Brain tumor (BT) classification plays a vital role in computer-aided diagnosis, as early detection directly influences treatment planning and patient survival. Advances in artificial intelligence, particularly in machine learning (ML) and deep learning (DL), have greatly improved automated tumor identification from magnetic resonance imaging (MRI), reducing reliance on manual assessment. This study proposes a hybrid framework that combines generative adversarial networks (GANs), convolutional neural networks (CNNs), and Vision Transformers (ViTs) for accurate tumor classification. GANs are used to generate realistic synthetic MRI images, addressing data scarcity and increasing dataset diversity. CNNs extract discriminative deep features, accelerating training while limiting overfitting. These features are then processed by a ViT model, capable of capturing complex spatial relationships within medical images. Experimental results show that the proposed ViT–CNN–GAN approach achieves 97.36% accuracy, outperforming conventional methods. Overall, this framework demonstrates the potential of advanced DL models to strengthen MRI-based tumor diagnosis and support more reliable clinical decision-making.
Iliass Zine-dine, J. Riffi, Khalid El Fazazy et al.· International Journal of Onl...· 0 citations
Abstract Introduction Pediatric cancer imposes a significant global burden, with ∼400 000 new cases annually. Magnetic resonance imaging (MRI) is a fundamental technique for timely diagnosis; however, interpretation can be subjective. In this context, convolutional neural networks (CNNs) have emerged as a promising tool for early and accurate detection of brain tumors in pediatric patients. Objective To develop and validate a CNN model for tumor detection tasks (tumor vs nontumor) of pediatric brain MRI images using T1-weighted imaging. Methodology T1-weighted brain MRI images from 285 pediatric patients were included (140 with confirmed tumors and 145 controls). A CNN architecture with convolutional layers, max-pooling, and dropout regularization was implemented. The model was trained and evaluated using cross-validation, employing the following metrics: accuracy, precision, recall, and F1-score. Model performance was compared across training configurations of 10-50 and 100 epochs to determine the optimal setup. Results The model trained for 40 epochs achieved the best overall performance, with a precision of 99%, a recall of 100%, and an F1-score of 99%. These metrics demonstrate excellent tumor detection while eliminating false negatives, a crucial aspect in pediatric oncology. Conclusion The CNN demonstrates strong potential as a diagnostic aid for detection brain tumors using pediatric MRI images. Its integration into clinical environments could contribute to more objective interpretation, optimize diagnostic workflows, and provide imaging follow-up before and after treatment.
Brain tumors are abnormal cell growths in the brain that require accurate and timely diagnosis, where Magnetic Resonance Imaging (MRI) plays a critical role in detecting and characterizing tumor structures. However, accurate interpretation of Magnetic Resonance Imaging (MRI) scans is challenging due to their complexity and the limited availability of expert radiologists. This challenge is further compounded by the lack of interpretability in many existing deep learning-based diagnostic systems. Therefore, the need for an automated and interpretable brain tumor classification system arises, which is the problem this study aims to address. In this research, a brain tumor classification system integrated with Explainable Artificial Intelligence (XAI) was developed using MRI images. The model was designed to classify brain tumors into glioma, meningioma, pituitary tumor, and no-tumor categories while providing visual explanations for its predictions using Grad-Class Activation Mapping (Grad-CAM). The performance of the system was evaluated for each tumor category using accuracy, precision, specificity, recall, F1-Score, false positive rate and also the Receiver Operating Characteristic and Area Under Curve (ROC-AUC). Experimental results show that the developed CNN model achieved an overall classification accuracy of 90.6% with an AUC-ROC value of 0.9892, indicating strong discriminative capability across tumor classes. The Grad-CAM visualizations consistently highlighted tumor-affected regions in the MRI images, confirming that the model based its predictions on clinically relevant anatomical structures. The developed model demonstrated effective classification performance and improved interpretability, making it suitable as a reliable decision-support tool for automated brain tumor diagnosis
T. H. Stephen, A. Oke, A. S. Falohun et al.· LAUTECH Journal of Engineeri...· 0 citations
Brain tumors represent an important clinical disease of the nervous system that requires early and accurate diagnosis for improved patient prognoses. Magnetic Resonance Imaging (MRI) is still the modality of choice for the evaluation of brain tumor due to its better soft tissue contrast and non-invasive nature. Nevertheless, manual interpretation of MRI examinations is difficult and subject to inter-observer variation. The present study aims at a thorough comparative evaluation of deep learning architectures for fully automated, multi-class brain tumor classification based on MRI data. A unified experimental framework is designed to empirically evaluate a custom convolutional neural network (CNNs) trained with the scratch method and four pre-trained transfer learning methods, including VGG16, ResNet50, DenseNet121, and MobileNetV3. The models are trained and validated using a large, consolidated data set of four classes: glioma, meningioma, pituitary tumour and healthy brain images. In order to make sure the comparability between models is true, same preprocessing, data augmentation and dataset partition strategy and the evaluation measure are followed by all the models. Performance is measured in terms of accuracy, precision, recall, F1 score, confusion matrices and roc-auc analysis. Results show that transfer learning models significantly exceed the custom CNN. Among the architectures evaluated, ResNet50 can achieve the best classification performance, test accuracy is 98.38% and macro-averaged F1-score is 0.9837, which can be regarded as good classification and generalisability. These results highlight the power of deep residual learning for solving some of the difficulties in classifying brain MRI, such as inter-class similarity and feature heterogeneity. The study adds a rigorous benchmarking mechanism and empirical evidence for adopting ResNet50 as a robust model for multi-class brain tumour diagnosis. Future research may extend this work by develop a web-based application that enables users to upload MRI images for automated brain tumor detection and classification.
Prabha Kumaresan, Xin Tian Lim· International Journal on Rob...· 0 citations
Magnetic resonance imaging (MRI) is a crucial component of the medical diagnostic and therapeutic approach for brain tumors, as early detection of anomalous tissue considerably enhances patient outcomes. MRI images may obscure significant tumor characteristics owing to noise, inadequate contrast, and erratic intensity distributions. This work examines the application of ResNet169, EfficientNetB0, and a hybrid fused model (ResNetEffi169B0) to address challenges in deep learning-based image improvement and categorization. The performance metrics included F1-score, Accuracy, Precision, Sensitivity, Entropy, SSIM, and MSE to test the models. The results show that EfficientNetB0 has the best PSNR (11.45), SSIM (0.308), and accuracy (0.601) when it comes to classification and improvement. The ResNet169 model's high sensitivity of 0.463 showed that it could dependably find tumor regions. The hybrid approach used the best features of both styles to create balanced results that made diagnoses more consistent and images clearer. The fusion method enhances the quality and structural data of magnetic resonance imaging (MRI) scan slices, leading to more precise classification and enhanced tumor visibility. This study examines the potential of hybrid deep learning models to enhance computer-aided diagnostic tools in medical imaging for improved brain tumor identification.
Balas, Subalatha.M· 2026 6th International Confe...· 0 citations