An Efficient Methodology for Brain MRI Classification Based on DWT and Convolutional Neural Network

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Signal Processing

Background:

• Brain Magnetic Resonance (MR) image classification is crucial for diagnosing neurological conditions.
• Accurate and efficient classification models are needed to aid clinical decision-making.

Purpose of the Study:

• To propose a novel model for brain MR image classification using discrete wavelet transform and convolutional neural networks.
• To evaluate the performance of the proposed model against state-of-the-art algorithms.

Main Methods:

• Preprocessing using a median filter to remove noise.
• Feature extraction via 3-level Harr wavelet decomposition for detail reduction and size minimization.
• Classification using a convolutional neural network (CNN) to categorize images as normal or abnormal.

Main Results:

• The proposed model achieved a high accuracy of 99% on a standard dataset.
• Performance evaluation demonstrated superior results compared to existing algorithms.
• The methodology proved effective for practical applications in brain MR image analysis.

Conclusions:

• The integrated discrete wavelet transform and CNN model offers a highly accurate and efficient solution for brain MR image classification.
• The proposed approach shows significant potential for real-world clinical use.
• This method outperforms current state-of-the-art techniques in brain MR image analysis.