Brain Tumor Classification from MRI Using Image Enhancement and Convolutional Neural Network Techniques.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Brain tumor detection from MRI is complex and prone to errors.
• Deep learning (DL) offers automated solutions for medical image analysis.
• Convolutional Neural Networks (CNNs) are effective for image classification tasks.

Purpose of the Study:

• To develop and validate a novel deep learning methodology for classifying brain tumors (glioma, meningioma, pituitary) and non-tumor cases from MRI.
• To integrate image enhancement techniques for improved classification performance.
• To compare the proposed model against established pre-trained models.

Main Methods:

• Implementation of a novel model combining Gaussian-blur sharpening and CLAHE adaptive histogram equalization for image enhancement.
• Classification of brain tumors using a deep learning approach.
• Validation using benchmark datasets and comparison with VGG16, ResNet50, VGG19, InceptionV3, and MobileNetV2.

Main Results:

• The proposed method achieved a classification accuracy of 97.84%.
• Precision, recall, and F1-score exceeded 97.85%, indicating high performance.
• The model demonstrated strong generalization capabilities across different tumor types.

Conclusions:

• The developed methodology accurately classifies common brain tumor types with high precision.
• The technique shows significant potential as a valuable tool for physicians in brain tumor diagnosis.
• The integration of image enhancement with DL improves diagnostic accuracy and efficiency.