Super-resolution reconstruction of neonatal brain magnetic resonance images via residual structured sparse representation

• Published in: Medical image analysis Vol. 55; pp. 76 - 87
• Main Authors: Zhang, Yongqin, Yap, Pew-Thian, Chen, Geng, Lin, Weili, Wang, Li, Shen, Dinggang
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 01.07.2019; Elsevier BV
• Subjects: Algorithms; Brain; Convex optimization; Dictionary learning; Image filters; Image reconstruction; Image resolution; Magnetic resonance imaging; Neonates; Newborn babies; Preschool children; Regularization; Representations; Resonance; Scanners; Signal to noise ratio; Sparse representation; Spatial discrimination; Spatial resolution; Sparse representation; 68-10; 62-35; Magnetic resonance imaging; Convex optimization; Dictionary learning
• ISSN: 1361-8415; 1361-8423; 1361-8431; 1361-8423
• DOI: 10.1016/j.media.2019.04.010

•Borrowing high-frequency information from toddler images in the form of dictionaries.•Incorporating a parameterized nonconvex regularization into super-resolution modeling.•Introducing a two-layer representation strategy to solve the problem of the discrepancies between test and training images.•Exploiting image nonlocal self-similarity for enhancing structured sparse representation. [Display omitted] Magnetic resonance images of neonates, compared with toddlers, exhibit lower signal-to-noise ratio and spatial resolution. In this paper, we propose a novel method for super-resolution reconstruction of neonate images with the help of toddler images, using residual-structured sparse representation with convex regularization. Specifically, we introduce a two-layer image representation, consisting of a base layer and a detail layer, to cater to signal variation across scanners and sites. The base layer consists of the smoothed version of the image obtained via Gaussian filtering. The detail layer is the difference between the original image and the base layer. High-frequency details in the detail layer are borrowed across subjects for super-resolution reconstruction. Experimental results on T1 and T2 images demonstrate that the proposed algorithm can recover fine anatomical structures, and generally outperform the state-of-the-art methods both qualitatively and quantitatively.