Wavelet-Enhanced hybrid CNN Models for Precise Brain Tumor Classification

The detection of brain tumors is a crucial and urgent procedure because the most aggressive types can significantly reduce life expectancy. Timely treatment is made possible by early, precise identification, which also greatly enhances patient outcomes. Radiologists have always relied on in-depth MRI image processing to detect malignancies, but this manual method is laborious and prone to mistakes. A variety of technology-supported diagnosis have been developed to categorize MRI images. However, due to the significant geographical and anatomical heterogeneity in the surrounding regions, automatically classifying brain tumors is still difficult. To overcome the drawbacks of strictly spatial processing, this study suggests a hybrid strategy that combines wavelet transforms with convolutional neural networks (CNNs) to preserve spectral and spatial information. The model utilizes Daubechies and Symlet wavelet families to enhance feature extraction, capturing edges, textures, and subtle patterns indicative of tumors. Evaluated on the Cheng dataset the proposed Wavelet CNN architecture, obtained 95.04% accuracy, outperforming conventional CNN-based methods. This method shows how wavelet transforms and deep learning can be combined to create a more robust and dependable system for classifying brain tumors in medical imaging.