Deducing magnetic resonance neuroimages based on knowledge from samples. (December 2017)

• Record Type: Journal Article
• Authors: Jiang, Yuwei; Liu, Feng; Fan, Mingxia; Li, Xuzhou; Zhao, Zhiyong; Zeng, Zhaoling; Wang, Yi; Xu, Dongrong
• Is Part Of: Computerized medical imaging and graphics. Volume 62(2017)
• Page Start: 1
• Page End: 14
• Publication Date: 2017-12
• Subjects: Relaxation time -- Analogical reasoning -- Knowledge-based deducing -- Contrast improvement -- Personalized imaging; Diagnostic imaging -- Periodicals; Imaging systems in medicine -- Periodicals; Diagnosis, Radioscopic -- Data processing -- Periodicals; Diagnostic Imaging -- Periodicals; Imagerie pour le diagnostic -- Périodiques; Diagnostic imaging; Periodicals; Electronic journals; Electronic journals; 616.0754
• DOI: 10.1016/j.compmedimag.2017.07.005
• Languages: English
• ISSNs: 0895-6111

Abstract

Highlights: A knowledge-based method for repairing magnetic resonance images of poor contrast. The method deduces relaxation properties using an analogical reasoning strategy. The method improves tissue contrast in the deduced images using optimal parameters. The method performs reliably and effectively without the need of rescanning the individuals. Abstract: Purpose: Because individual variance always exists, using the same set of predetermined parameters for magnetic resonance imaging (MRI) may not be exactly suitable for each participant. We propose a knowledge-based method that can repair MRI data of undesired contrast as if a new scan were acquired using imaging parameters that had been individually optimized. Methods: The method employed a strategy called analogical reasoning to deduce voxel-wise relaxation properties using morphological and biological similarity. The proposed framework involves steps of intensity normalization, tissue segmentation, relaxation time deducing, and image deducing. Results: This approach has been preliminarily validated using conventional MRI data at 3 T from several examples, including 5 normal and 9 clinical datasets. It can effectively improve the contrast of real MRI data by deducing imaging data using optimized imaging parameters based on deduced relaxation properties. The statistics of deduced images shows a high correlation with real data that were actually collected using the same set of imaging parameters. Conclusion: The proposed method of deducing MRI data using knowledge of relaxation times alternatively provides a way of repairing MRI data of less optimal contrast. The method is also capable of optimizing an MRI protocol for individual participants, thereby realizing personalized MR imaging.