Automated Classification of Brain Tumors in MRI Images using Convolutional Neural Networks

Abstract

Brain tumours are one of the most critical neurological disorders, and their early detection is essential for effective treatment and improved patient prognosis. Magnetic Resonance Imaging (MRI) is the standard imaging technique used for diagnosis; however, manual interpretation is time-consuming and prone to human error. This paper presents NEUROCARE, a desktop-based intelligent diagnostic system for brain tumor detection using a custom Convolutional Neural Network (CNN). The system is designed to classify MRI images into distinct categories, including Glioma, Meningioma, Pituitary tumor, and No Tumor. After evaluating various deep learning architectures, our final custom CNN model achieved a test accuracy of 93.82% with a test loss of 0.1971. The NEUROCARE application integrates this model into a user-friendly graphical user interface (GUI) built with Tkinter, offering functionalities such as patient history tracking and automated PDF report generation. This system aims to serve as an accessible and effective tool to assist radiologists in delivering faster and more accurate diagnoses.

Introduction

Brain tumors pose a significant global health challenge due to their high mortality and diagnostic complexity. They can be malignant or benign, and early detection is crucial, as late-stage diagnosis limits treatment options and reduces survival rates. Magnetic Resonance Imaging (MRI) is the primary tool for detecting brain abnormalities, but manual interpretation is time-consuming, requires expert radiologists, and is prone to errors, particularly in resource-limited settings.

This study introduces NEUROCARE, a machine learning-based desktop application designed for automated brain tumor detection and classification using MRI scans. The system leverages Convolutional Neural Networks (CNNs) to learn hierarchical features from MRI images, enabling accurate and end-to-end classification. Various deep learning architectures (VGG-16, ResNet, DenseNet, InceptionNet) were evaluated, and a custom CNN achieved a test accuracy of 93.82% and a test loss of 0.1971. The model is integrated into a Tkinter-based GUI, making it user-friendly for clinical application in urban and rural environments.

Literature Review:

• Hybrid methods combining feature extraction (DWT, ICA) with deep learning improve classification accuracy.

• End-to-end systems with preprocessing, segmentation, and CNN-based feature extraction show strong diagnostic potential.

• Surveys highlight the growing use of CNNs, 3D-CNNs, U-Net, and WRN-PPNet in medical imaging, emphasizing explainability and model robustness.

Algorithms and Methodology:

• Traditional machine learning methods rely on handcrafted features (GLCM, shape, intensity) and classifiers like SVM and Random Forests, but are limited by feature quality.

• Deep learning, particularly CNNs, offers end-to-end learning, automatically capturing spatial hierarchies in MRI scans.

• The implemented CNN accepts 150×150 MRI images, with four convolutional blocks (increasing filters from 32 to 128), max pooling, ReLU activations, a dense layer of 512 neurons with dropout (0.5), and a softmax output layer classifying four categories: Glioma, Meningioma, Pituitary, No Tumor.

Results and Discussion:

• The CNN model trained for 50 epochs achieved 95.7% accuracy and a categorical cross-entropy loss of 0.124 on the test set, demonstrating strong generalization.

• Per-class analysis showed high true positives, especially for “No Tumor” and “Pituitary” cases.

• Minor confusion occurred between Glioma and Meningioma, likely due to similar morphological features.

• The model showed very low false negatives for “No Tumor,” which is critical for clinical safety.

Conclusion

In this project, we successfully developed NEUROCARE, an end-to-end system for the automated classification of brain tumors from MRI images. Our custom-designed Convolutional Neural Network achieved a high accuracy of 93.82% on the test dataset, demonstrating its effectiveness as a diagnostic aid. By integrating this model into a user-friendly desktop application, we have created a practical tool that can assist healthcare professionals by providing fast, reliable, and data-driven insights. The application\'s features, such as patient history tracking and automated report generation, are designed to streamline clinical workflows. Future work could focus on expanding the dataset, exploring 3D-CNN architectures for volumetric analysis, and conducting clinical validation studies to assess the tool\'s real-world impact. Ultimately, NEUROCARE represents a significant step towards leveraging artificial intelligence to create more efficient, accessible, and equitable diagnostic solutions in healthcare.

References

[1] H. Mohsen, E.-S. A. El Dahshan, E.-S. M. El Horbaty, and A.-B. M. Salem, “Classification using deep learning neural networks for brain tumors,” Future Computing and Informatics Journal, vol.?3, no.?1, Article?6, pp.?68–71, Jun.?2018. DOI:10.1016/j.fcij.2017.12.001 [2] M. K. Islam, M. S. Ali, M. S. Miah, M. M. Rahman, M. S. Alam, and M. A. Hossain, “Brain tumor detection in MR image using superpixels, principal component analysis and template based K-means clustering algorithm,” Machine Learning with Applications, vol.?5, p.?100044, Sep.?2021. DOI:10.1016/j.mlwa.2021.100044 [3] S. Basheera and M. S. Satya Ram, “Classification of brain tumors using deep features extracted using CNN,” Journal of Physics: Conference Series, vol.?1172, p.?012016, 2019. DOI:10.1088/1742 6596/1172/1/012016 [4] F. P. Polly, S. K. Shil, M. A. Hossain, A. Ayman, and Y. M. Jang, “Detection and classification of HGG and LGG brain tumor using machine learning,” in Proc. 2018 Int. Conf. Information Networking (ICOIN), Chiang Mai, Thailand, pp.?813–817, Jan.?2018. DOI:10.1109/ICOIN.2018.8343231 [5] M. M. Badža and M. ?. Barjaktarovi?, “Classification of brain tumors from MRI images using a convolutional neural network,” Applied Sciences, vol.?10, no.?6, p.?1999, Jun.?2020. DOI:10.3390/app10061999 [6] J. Amin, M. Sharif, A. Haldorai, M. Yasmin, and R. S. Nayak, “Brain tumor detection and classification using machine learning: A comprehensive survey,” Complex & Intelligent Systems, vol.?8, pp.?3161–3183, 2022. DOI:10.1007/s40747 021 00563 y

Copyright

Copyright © 2025 Madhuri Barhate, Adithya Sagri, Aarush Sengupta, Abduljalil Qasem, Abhinandan Goswami, Viraj Adake, Adinath Dound. 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.