An optimized morphology transform-based diagnostic computed tomography image enhancement using edge map

In computed tomography (CT) imaging, a low-quality CT scan, or more precisely, a noisy low contrast CT image, may provide insufficient information for visual interpretation of impacted regions. The purpose of this article is to propose a novel optimal morphological transform-based method for improving the noise and contrast of CT images in the wavelet domain. The low-quality CT image is first transformed using the optimized morphology transform method, which determines the optimal value of the parameter included in the fundamental morphology transformation equation. The parameter's optimum value is determined using the particle swarm optimization (PSO) method. The proposed approach then employs the discrete wavelet transform (DWT) to decompose the input and transformed CT image into higher and lower sub-bands. Following that, the lower sub-bands are modified by applying the correction factor via singular value decomposition (SVD). The input image's higher sub-band (HH) is denoised using an edge map (EM)-based method. This produces enhanced CT images with modified lower and higher sub-band. Finally, the modified sub-bands and remaining unprocessed higher frequency sub-bands are processed using the inverse discrete wavelet transform (IDWT), resulting in an enhanced CT image. Experiments and validations are conducted on a CT images obtained from a publicly available database to assess the proposed scheme's effectiveness qualitatively and quantitatively. Significant quantitative study shows that the proposed approach surpasses existing image enhancement approaches in terms of signal-to-noise ratio, discrete entropy, enhancement measurement, and contrast ratio. The proposed method generates higher-quality CT images, which is advantageous for disease inspection and diagnosis.

Automated summary

This article proposes a novel optimal morphological transform-based method for improving the noise and contrast of CT images in the wavelet domain. The proposed approach combines optimized morphology transform and discrete wavelet transform to enhance the quality of CT images. Experimental results demonstrate that the proposed method outperforms existing image enhancement approaches and produces higher-quality CT images, which is beneficial for disease inspection and diagnosis purposes.