Joint dictionary learning reconstruction of compressed multi-contrast magnetic resonance imaging

This study deals with reconstruction of compressed multicontrast magnetic resonance image (MRI) reconstruction using joint dictionary learning. Usually pre-determined dictionaries are used for compressed sensing reconstructions. Here, we propose an alternating-minimization based algorithm for recovering image and sparsifying transformation from only data itself. The proposed method can also be viewed as a joint multicontrast reconstruction extension of a previous blind compressive sensing algorithm [1]. For evaluation, the algorithm is compared in terms of convergence speed and image quality to both individual dictionary learning based method [1], and a joint reconstruction algorithm using pre-determined dictionaries for MRI [2].