Contrast Enhancement of Multiple Tissues in MR Brain Images With Reversibility

Contrast enhancement (CE) of magnetic resonance (MR) brain images is an important technique to bring out the tissue details for clinical diagnosis. Recently, a new form of image enhancement has been proposed to complete the task without any information loss. Specifically, information required to restore the original image is reversibly hidden into the enhanced image. Moreover, several image segmentation based algorithms have been proposed so that the region of interest can be exclusively enhanced. However, with the reversible algorithms, it is hard to properly enhance the tissues in MR brain images when they are relatively small or connected with each other. To address this issue, a hierarchical CE scheme is proposed for MR brain images with reversibility in this letter. Firstly, a deep convolutional neural network is used to segment multiple tissue classes automatically. Then, the segmented tissues are individually utilized to guide the CE procedure so that individual-tissue-enhanced images are generated. Compared with using the background information to guide the CE procedure, better tissue enhancement effects and visual quality are both obtained by our proposed hierarchical scheme. The evaluation results obtained over MR brain test images demonstrate the reversibility and adaptability of the proposed scheme for the enhancement of interested tissues.

Automated summary

A new hierarchical contrast enhancement scheme for MR brain images with reversibility is proposed, utilizing deep learning for automatic tissue segmentation and guidance in the enhancement process, resulting in better enhancement effects and visual quality compared to conventional methods. Evaluation results demonstrate the effectiveness of the proposed scheme in enhancing interested tissues in MR brain images.