Brain Tumor Classification through MRIs using Transfer Learning with VGG16 Model

Abstract

In the medical field, image segmentation is a crucial and difficult task. A useful technique for identifying aberrant brain tissue is magnetic resonance imaging (MRI) scans. For radiologists, correctly identifying and categorizing brain tumors from MRI scans is still a difficult and time-consuming task. This study offers a clever technique for accurately identifying brain tumors. The study investigates the use of Convolutional Neural Networks (CNNs) in conjunction with optimization techniques to classify different types of brain tumors from MRI data. In particular, tumor features are categorized and tumor kinds are identified using transfer learning on the VGG16 model. This method seeks to increase MRI scanning efficacy and improve identification precision. When evaluated using MRI scans from the Figshare, SARTAJ, and Br35H datasets [31], the proposed approach, which utilizes transfer learning, enhanced the performance of the original VGG16 model, allowing for more accurate and robust classification than its baseline capabilities, improving from 91.38% [1] to over 95%.

Introduction

Brain tumors are abnormal cell growths in the brain, categorized as benign or malignant. Due to the brain’s confined space within the skull, such growths can be dangerous. MRI scans are crucial for tumor detection, segmentation, and treatment planning. However, tumor variability in size, shape, and location complicates accurate diagnosis and segmentation.

Deep learning, particularly Convolutional Neural Networks (CNNs), has significantly improved brain tumor detection by automating feature extraction and classification. CNNs excel at identifying subtle patterns in MRI images, enabling earlier and more precise diagnoses while reducing human error and addressing doctor shortages. However, CNNs require large labeled datasets, and performance drops with limited data or low-contrast images.

To address this, transfer learning is employed. Pre-trained CNN models (like VGG16, DenseNet201, EfficientNetB5, InceptionResNetV2) are fine-tuned on smaller MRI datasets. Contrast enhancement techniques (using PIL) further improve image clarity. These methods form the basis of a robust Computer-Aided Diagnosis (CAD) framework that efficiently classifies brain tumors.

The study utilizes a publicly available dataset of 7,023 MRI images (categorized as glioma, meningioma, pituitary tumor, and no tumor) to train and test CNN models. Data augmentation, normalization, and hyperparameter tuning are used to enhance performance. Standard evaluation metrics (accuracy, precision, recall, F1 score) and visualization tools (confusion matrix, ROC curve) validate the approach.

Main Contributions:

• Application of deep transfer learning and contrast enhancement for tumor diagnosis.

• A comprehensive CAD framework for automated tumor identification and classification.

• Comparative analysis of multiple CNN models on the same dataset.

• Recommendations for continuous improvement through real-world deployment and feedback.

Literature Survey Insights:

• Historical methods evolved from traditional ML to deep learning.

• Studies show CNNs outperform conventional methods with accuracies up to 98%.

• Hybrid models, like DCNet and BrainMRNet, and 3D segmentation techniques (e.g., V-Net + 3DU-Net) have shown promising results.

• Key limitations of earlier methods include reliance on handcrafted features and limited generalization.

Technical Implementation:

• Training is done on high-performance systems using frameworks like TensorFlow and PyTorch.

• The model architecture uses CNNs with transfer learning, trained on augmented and normalized MRI data.

• The dataset used includes labeled images from Kaggle, SARTAJ, and Br35H sources

Conclusion

This study presents an automated, intelligent approach for detecting and categorizing brain cancers based on a custom-trained VGG16 model. The system uses transfer learning to improve the performance of the deep learning-based solution, even with a little dataset. In the first phase, preprocessing procedures such as scaling, normalization, and augmentation are used to prepare MRI images for analysis. In the second step, the custom-trained VGG16 model extracts high- and low-level features to classify brain tumors. This model achieved a 95% accuracy rate, with an average precision, recall, and F1-score of 96% across all categories. Several factors contribute to pre-trained models\' high performance, including InceptionResNetV2, EfficientNetB, and DenseNet201, which obtained 99% and 100% accuracy rates, respectively. These models include deeper topologies and better residual and attention methods, allowing them to extract more complex characteristics from MRI images. For example, DenseNet201\'s capacity to reuse features via dense connections adds to its superior precision and recall. These findings highlight the potential of deep learning approaches, particularly pre-trained models and transfer learning, in improving brain tumor detection and diagnostic accuracy. However, constraints remain, such as biases in training data and the difficulty of generalizing across various populations. Future studies should evaluate model performance on bigger, heterogeneous datasets to ensure consistency across clinical contexts. Exploring hybrid techniques that combine advanced architectures and optimization algorithms with novel data augmentation strategies has the potential to improve accuracy and robustness.

References

[1] M. Agarwal, G. Rani, A. Kumar, P. Kumar, R. Manikandan, and A. H. Gandomi, “Deep learning for enhanced brain Tumor Detection and classification,” Results in Engineering, vol. 22, p. 102117, 2024. [2] S. Hasan, M. Yousif, and T. M. J. Al-Talib, “Retracted: Brain tumor classification using Probabilistic Neural Network,” Journal of Fundamental and Applied Sciences, vol. 10, no. 4S, pp. 667–670, 2018. [3] B. Pushpa and F. Louies, “Detection and classification of brain tumor using machine learning approaches,” International Journal of Research in Pharmaceutical Sciences, vol. 10, no. 3, pp. 2153–2162, 2019. [4] N. Varuna Shree and T. N. R. Kumar, “Identification and classification of brain tumor MRI images with feature extraction using DWT and probabilistic neural network,” Brain Inform, vol. 5, no. 1, pp. 23–30, 2018. [5] A. Rehman, S. Naz, M. I. Razzak, F. Akram, and M. Imran, “A deep learning-based framework for automatic brain tumors classification using transfer learning,” Circuits Syst Signal Process, vol. 39, no. 2, pp. 757–775, 2020. [6] Z. N. K. Swati et al., “Brain tumor classification for MR images using transfer learning and fine-tuning,” Computerized Medical Imaging and Graphics, vol. 75, pp. 34–46, 2019. [7] S. Deepak and P. M. Ameer, “Brain tumor classification using deep CNN features via transfer learning,” ComputBiol Med, vol. 111, p. 103345, 2019. [8] A. Ar?, O. F. Alcin, and D. Hanbay, “Brain MR image classification based on deep features by using extreme learning machines,” Biomed J Sci Tech Res, vol. 25, no. 3, 2020. [9] C. Zhang, X. Shen, H. Cheng, and Q. Qian, “Brain tumor segmentation based on hybrid clustering and morphological operations,” Int J Biomed Imaging, vol. 2019, no. 1, p. 7305832, 2019. [10] T. Kalaiselvi, T. Padmapriya, P. Sriramakrishnan, and V. Priyadharshini, “Development of automatic glioma brain tumor detection system using deep convolutional neural networks,” Int J Imaging SystTechnol, vol. 30, no. 4, pp. 926–938, 2020. [11] R. Sekaran, A. K. Munnangi, M. Ramachandran, and A. H. Gandomi, “3D brain slice classification and feature extraction using Deformable Hierarchical Heuristic Model,” ComputBiol Med, vol. 149, p. 105990, 2022. [12] A. Gumaei, M. M. Hassan, M. R. Hassan, A. Alelaiwi, and G. Fortino, “A hybrid feature extraction method with regularized extreme learning machine for brain tumor classification,” IEEE Access, vol. 7, pp. 36266–36273, 2019. [13] M. A. Azamet al., “A review on multimodal medical image fusion: Compendious analysis of medical modalities, multimodal databases, fusion techniques and quality metrics,” ComputBiol Med, vol. 144, p. 105253, 2022. [14] I. Aboussaleh, J. Riffi, K. el Fazazy, A. M. Mahraz, and H. Tairi, “3DUV-NetR+: A 3D hybrid semantic architecture using transformers for brain tumor segmentation with MultiModal MR images,” Results in Engineering, vol. 21, p. 101892, 2024. [15] A. K. Sharma et al., “HOG transformation based feature extraction framework in modified Resnet50 model for brain tumor detection,” Biomed Signal Process Control, vol. 84, p. 104737, 2023. [16] C. Li, F. Zhang, Y. Du, and H. Li, “Classification of brain tumor types through MRIs using parallel CNNs and firefly optimization,” Sci Rep, vol. 14, no. 1, p. 15057, 2024. [17] S. S. R. Phaye, A. Sikka, A. Dhall, and D. Bathula, “Dense and diverse capsule networks: Making the capsules learn better,” arXiv preprint arXiv:1805.04001, 2018. [18] A. Pashaei, H. Sajedi, and N. Jazayeri, “Brain tumor classification via convolutional neural network and extreme learning machines,” in 2018 8th International conference on computer and knowledge engineering (ICCKE), IEEE, 2018, pp. 314–319. [19] R. Hashemzehi, S. J. S. Mahdavi, M. Kheirabadi, and S. R. Kamel, “Detection of brain tumors from MRI images base on deep learning using hybrid model CNN and NADE,” Biocybern Biomed Eng, vol. 40, no. 3, pp. 1225–1232, 2020. [20] F. Özyurt, E. Sert, and D. Avc?, “An expert system for brain tumor detection: Fuzzy C-means with super resolution and convolutional neural network with extreme learning machine,” Med Hypotheses, vol. 134, p. 109433, 2020. [21] M. To?açar, B. Ergen, and Z. Cömert, “BrainMRNet: Brain tumor detection using magnetic resonance images with a novel convolutional neural network model,” Med Hypotheses, vol. 134, p. 109531, 2020. [22] T. Rahman and M. S. Islam, “MRI brain tumor detection and classification using parallel deep convolutional neural networks,” Measurement: Sensors, vol. 26, p. 100694, 2023. [23] M. I. Mahmud, M. Mamun, and A. Abdelgawad, “A deep analysis of brain tumor detection from mr images using deep learning networks,” Algorithms, vol. 16, no. 4, p. 176, 2023. [24] M. U. Ali, S. J. Hussain, A. Zafar, M. R. Bhutta, and S. W. Lee, “WBM-DLNets: wrapper-based metaheuristic deep learning networks feature optimization for enhancing brain tumor detection,” Bioengineering, vol. 10, no. 4, p. 475, 2023. [25] M. Hammad, M. ElAffendi, A. A. Ateya, and A. A. Abd El-Latif, “Efficient brain tumor detection with lightweight end-to-end deep learning model,” Cancers (Basel), vol. 15, no. 10, p. 2837, 2023. [26] A. B. Abdusalomov, M. Mukhiddinov, and T. K. Whangbo, “Brain tumor detection based on deep learning approaches and magnetic resonance imaging,” Cancers (Basel), vol. 15, no. 16, p. 4172, 2023. [27] P. Neelima, P. Nikilish, and R. S. Shankar, “Fine-Tuning based Deep Transfer Learning System used to Identify the Stage of Brain Tumour from MR-Images,” in 2023 Second International Conference on Augmented Intelligence and Sustainable Systems (ICAISS), IEEE, 2023, pp. 1003–1011. [28] S. Sharmin, T. Ahammad, M. A. Talukder, and P. Ghose, “A hybrid dependable deep feature extraction and ensemble-based machine learning approach for breast cancer detection,” IEEE Access, 2023. [29] A. Raza, M. S. Alshehri, S. Almakdi, A. A. Siddique, M. Alsulami, and M. Alhaisoni, “Enhancing brain tumor classification with transfer learning: Leveraging DenseNet121 for accurate and efficient detection,” Int J Imaging SystTechnol, vol. 34, no. 1, p. e22957, 2024. [30] K. R. Pedada, B. Rao, K. K. Patro, J. P. Allam, M. M. Jamjoom, and N. A. Samee, “A novel approach for brain tumour detection using deep learning based technique,” Biomed Signal Process Control, vol. 82, p. 104549, 2023 [31] Nickparvar, M.: Brain Tumor MRI Dataset. Kaggle. https://doi.org/10.34740/KAGGLE/DSV/2645886 (2021)

Copyright

Copyright © 2025 Neha Pokhriyal, Aadhar Jain. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.