Enhancing Brain Tumor MRI Classification Using Dual-GAN Data Augmentation and Deep Ensemble Feature Extraction
International Journal of Medical Toxicology and Forensic Medicine,
Vol. 16 (2026),
1 January 2026
,
Page 1-9
https://doi.org/10.22037/ijmtfm.v16.53538
Abstract
Background: Accurate computer-assisted classification of brain tumors from magnetic resonance imaging (MRI) remains difficult because medical imaging datasets are often imbalanced. Deep learning algorithms tend to over-represent majority classes, with fewer samples available for under-represented tumor classes, making it less reliable to diagnose clinically important minority cases. To address class imbalance and enhance discriminative feature representation and model interpretability, the authors propose a DualGAN–DeepEFE framework.
Methods: The proposed framework combines the dual-stage generative augmentation with deep ensemble feature extraction. Two complementary generative adversarial networks (GANs) are used to synthesize realistic MRI samples for under-represented tumor classes without altering anatomical structure or fine-grained lesion characteristics. The enhanced, balanced dataset is then fed into the pre-trained ResNet50 and VGG16 networks. Both backbones are used for feature extraction, and their features are concatenated to provide a complete representation of the tumor, which is then passed to a fully connected classifier for four-class prediction. To improve visual interpretability, Grad-CAM is used to highlight image regions that influence each prediction.
Results: Using an open four-class brain MRI dataset, the model achieved a validation accuracy of 94.16%, a test accuracy of ~94%, and a macro F1-score of 94.5%. The class-wise ROC-AUC values were all above 0.96, indicating good discrimination across all tumor classes.
Conclusion: It can be inferred from the findings that the use of dual-stage synthetic augmentation with complementary deep feature fusion is effective in minimizing the effect of class imbalance and can maintain classification performance well. Furthermore, the Grad-CAM algorithm was used to localize regions associated with diagnostic information, which can be interpreted, thus aiding in the potential clinical use of the proposed DualGAN–DeepEFE framework.
- Brain tumor classification, generative adversarial networks, DualGAN-DeepEFE (Deep Ensemble Feature Extraction), data augmentation, deep feature ensemble, medical image analysis, class imbalance, explainable AI, Grad-CAM
How to Cite
References
[1] Roy P, Srijon FMS, Bhowmik P. An explainable ensemble approach for advanced brain tumor classification applying Dual-GAN mechanism and feature extraction techniques over highly imbalanced data. PLoS One. 2024;19(9):e0310748. [DOI: 10.1371/journal.pone.0310748]
[2] Ghassemi N, Shoeibi A, Rouhani M. Deep neural network with generative adversarial networks pre-training for brain tumor classification based on MR images. Biomed Signal Process Control. 2020;57:101678. [DOI: 10.1016/j.bspc.2019.101678]
[3] Ge C, Gu IYH, Jakola AS, Yang J. Enlarged training dataset by pairwise GANs for molecular-based brain tumor classification. IEEE Access. 2020;8:22560-70. [DOI: 10.1109/ACCESS.2020.2969805]
[4] Ge C, Gu IYH, Jakola AS, Yang J. Deep semi-supervised learning for brain tumor classification. BMC Med Imaging. 2020;20:87. [DOI: 10.1186/s12880-020-00485-0]
[5] Gab Allah AM, Sarhan AM, Elshennawy NM. Classification of brain MRI tumor images based on deep learning PGGAN augmentation. Diagnostics (Basel). 2021;11(12):2343. [DOI: 10.3390/diagnostics11122343]
[6] Ahmad B, Sun J, You Q, Palade V, Mao Z. Brain tumor classification using a combination of variational autoencoders and generative adversarial networks. Biomedicines. 2022;10(2):223. [DOI: 10.3390/biomedicines10020223]
[7] Mukherkjee D, Saha P, Kaplun D, Sinitca A, Sarkar R. Brain tumor image generation using an aggregation of GAN models with style transfer. Sci Rep. 2022;12:9141. [DOI: 10.1038/s41598-022-12646-y]
[8] Huang P, Li D, Jiao Z, Wei D, Cao B, Mo Z, et al. Common feature learning for brain tumor MRI synthesis by context-aware generative adversarial network. Med Image Anal. 2022;79:102472. [DOI: 10.1016/j.media.2022.102472]
[9] Yurtsever MME, Atay Y, Arslan B, Sagiroglu S. Development of brain tumor radiogenomic classification using GAN-based augmentation of MRI slices in the newly released gazi brains dataset. BMC Med Inform Decis Mak. 2024;24:285. [DOI: 10.1186/s12911-024-02699-6]
[10] Zandi M, Nasimzade S, Pourhossein B, Fazeli M. A Snapshot of Different Types of Under Research Vaccines A Gainst Covid 1 9: A Review. 2020. [Link]
[11] Kang J, Ullah Z, Gwak J. MRI-based brain tumor classification using ensemble of deep features and machine learning classifiers. Sensors (Basel). 2021;21(6):2222. [DOI: 10.3390/s21062222]
[12] Celik F, Celik K, Celik A. Enhancing brain tumor classification through ensemble attention mechanism. Sci Rep. 2024;14:22260. [DOI: 10.1038/s41598-024-73803-z]
[13] Ullah Z, Kim J. Hierarchical deep feature fusion and ensemble learning for enhanced brain tumor MRI classification. Mathematics. 2025;13(17):2787. [DOI: 10.3390/math13172787]
[14] Patro SGK, Govil N, Saxena S, Mishra BK, Zamani AT, Ben Miled A, et al. Brain tumor classification using an ensemble of deep learning techniques. IEEE Access. 2024;12:162094-106. [DOI: 10.1109/ACCESS.2024.3485895]
[15] Kaya Y, Akat E, Yıldırım S. Fusion-Brain-Net: a novel deep fusion model for brain tumor classification. Brain Behav. 2025;15(5):e70520. [DOI: 10.1002/brb3.70520]
[16] Badža MM, Barjaktarović MČ. Classification of brain tumors from MRI images using a convolutional neural network. Appl Sci. 2020;10(6):1999. [DOI: 10.3390/app10061999]
[17] Almalki YE, Ali MU, Kallu KD, Masud M, Zafar A, Alduraibi SK, et al. Isolated convolutional-neural-network-based deep-feature extraction for brain tumor classification using shallow classifier. Diagnostics (Basel). 2022;12(8):1793. [DOI: 10.3390/diagnostics12081793]
[18] Xu Z, Tang J, Qi C, Yao D, Liu C, Zhan Y, et al. Cross-domain attention-guided generative data augmentation for medical image analysis with limited data. Comput Biol Med. 2023;168:107744. [DOI: 10.1016/j.compbiomed.2023.107744]
[19] Chakrabarty N. Brain MRI images for brain tumor detection [dataset on the Internet]. Kaggle; 2019. [Link]
- Abstract Viewed: 75 times
- 53538 Downloaded: 23 times