An adaptive enhancement method based on stochastic parallel gradient descent of glioma image

• Published in: IET image processing Vol. 17; no. 14; pp. 3976 - 3985
• Main Authors: Wang, Hongfei, Peng, Xinhao, Ma, ShiQing, Wang, Shuai, Xu, Chuan, Yang, Ping
• Format: Journal Article
• Language: English
• Published: Wiley 01.12.2023
• Subjects: artefacts suppression; contrast improvement; histogram modification; image denoising; image enhancement; parallel algorithms; stochastic parallel gradient descent
• ISSN: 1751-9659; 1751-9667; 1751-9667
• DOI: 10.1049/ipr2.12911

Brain tumour diagnosis is significant for both physicians and patients, but the low contrast and the artefacts of MRI glioma images always affect the diagnostic accuracy. The existing mainstream image enhancement methods are insufficient in improving contrast and suppressing artefacts simultaneously. To enrich the field of glioma image enhancement, this research proposed a glioma image enhancement method based on histogram modification and total variational using stochastic parallel gradient descent (SPGD) algorithm. Firstly, this method modifies the cumulative distribution function on the image histogram and performs gamma correction on the image according to the modified histogram to obtain a contrast‐enhanced image. Then, the method suppresses the artefacts of glioma images by total variational and wavelet denoising algorithm. To get better enhancement images, the optimal parameters in the proposed method are searched by the SPGD algorithm. The statistical studies performed on 580 real glioma images demonstrate that the authors’ approach can outperform the existing mainstream image enhancement methods. The results show that the proposed method increases the discrete entropy of the image by 8.9% and the contrast by 2.8% compared to original images. The enhanced images are produced by the proposed method with a natural appearance, appealing contrast, less degradation, and reasonable detail preservation. To enrich the field of glioma image enhancement, this paper proposed a glioma image enhancement framework based on histogram modification and total variational using stochastic parallel gradient descent (SPGD) algorithm. The results show that the proposed framework has superior performance compared to several existing glioma image enhancement methods.