Hybrid Discrete Wavelet Enhancement Model for Brain NCCT Images

Abstract

NCCT brain images are widely used to diagnose the brain abnormalities. The continued advancement and widespread use of computed tomography in medical science has increased the harmful effect of high dose radiation to patients. Moreover, low dose radiation may result in image deterioration, increase level of noise and artifacts which effects the radiologists' decisions. Different image denoising algorithms may be employed to reduce noise in NCCT images. In this research, a novel HDWN approach has been developed to enhance NCCT image as well as denoise. The proposed method takes into account the inherent properties of noise as well as complementary information of different wavelet coefficients to evaluate the noise in less computing time. Moreover, a directional regularizer has been incorporated to control the uneven pattern of noise and to differentiate image details from noise. Experiments have been performed on real NCCT brain images collected for diagnostic center. The performance metrics PSNR, SSIM, MSE have been used to measure the results. The proposed method outperforms many denoising and image enhancement state of art methods in both quantitative and qualitative measures.