Convolutional neural networks (CNNs) demonstrate excellent performance when employed\nto reconstruct the images obtained by compressed-sensing magnetic resonance imaging (CS-MRI).\nOur study aimed to enhance image quality by developing a novel iterative reconstruction approach\nthat utilizes image-based CNNs and k-space correction to preserve original k-space data. In the\nproposed method, CNNs represent a priori information concerning image spaces. First, the CNNs are\ntrained to map zero-filling images onto corresponding full-sampled images. Then, they recover the\nzero-filled part of the k-space data. Subsequently, k-space corrections, which involve the replacement\nof unfilled regions by original k-space data, are implemented to preserve the original k-space data.\nThe above-mentioned processes are used iteratively. The performance of the proposed method was\nvalidated using a T2-weighted brain-image dataset, and experiments were conducted with several\nsampling masks. Finally, the proposed method was compared with other noniterative approaches\nto demonstrate its effectiveness. The aliasing artifacts in the reconstructed images obtained using\nthe proposed approach were reduced compared to those using other state-of-the-art techniques.\nIn addition, the quantitative results obtained in the form of the peak signal-to-noise ratio and structural\nsimilarity index demonstrated the effectiveness of the proposed method. The proposed CS-MRI\nmethod enhanced MR image quality with high-throughput examinations.
Loading....