Detection And Characterization Of Mr Brain Images Using Hybrid Wavelet Transform And Optimized Clustering With Shaft Algorithm

Abstract

Image enhancement techniques are crucial in image processing applications where image quality and contrast play a significant role in human interpretation. This study focuses on enhancing the resolution of magnetic resonance (MR) brain images using a hybrid wavelet transform technique with interpolation. The technique utilizes the discrete wavelet transform (DWT) and introduces a new approach called Hybrid Wavelet Transform Resolution Enhancement (HWT-RE). The DWT-RE method decomposes the low-resolution (LR) image into subbands using DWT and applies interpolation to enhance the high-frequency subbands. The HWT-RE method combines both decimated and undecimated wavelet transforms to improve the performance of resolution enhancement. The proposed methods are evaluated based on Mean Squared Error (MSE) and Peak Signal-to-Noise Ratio (PSNR) metrics. The results demonstrate that the HWT-RE method with interpolation provides improved image quality compared to the DWT-RE method. The study contributes to the field of image enhancement in MR brain images and provides insights into the application of hybrid wavelet transform techniques for resolution enhancement. The detection and characterization of tumors in MR-brain images play a crucial role in medical diagnosis and treatment. We propose a novel method for tumor detection and estimation of tumor area using the shaft algorithm and binarization techniques. Our approach demonstrates improved performance compared to conventional segmentation techniques such as fuzzy c-means (FCM) and K-means segmentation in terms of precision, time, and accuracy. The proposed method utilizes the shaft algorithm for tumor detection and also includes the calculation of tumor areas using binarization by counting the number of white pixels in a segmented MR image. Additionally, we introduce a Hybrid Wavelet Transform Resolution Enhancement (HWT-RE) technique with interpolation to enhance the quality of MR brain images. Experimental results show that our proposed algorithm outperforms existing m