Blind compressive sensing dynamic MRI with sparse dictionaries

Each row shows few spatial frames and the image time profile. We observe a similar behavior as seen in fig.1. k-t FOCUSS showed some temporal blur (see yellow arrow in (b)). BCS had better temporal fidelity but suffered from noisy artifacts (see arrows in (c) due to learning noisy patterns. Sparse BCS resulted in reconstructions with reduced noise like artifacts without compromising on the spatiotemporal fidelity. Fig.1 Comparisons using free breathing data: The first, second and third rows respectively correspond to a spatial frame, error image, and the image time profile. The sampling mask is shown in i.b. The error images are scaled up by ~5 fold for better visualization. The kt FOCUSS method resulted in motion blurring and loss of temporal resolution due to the large motion content (see arrows in ii). The BCS scheme maintained the motion content, but suffered largely from noisy artifacts due to learning of noisy bases (see arrows in iii.). In contrast, the sparse BCS scheme produced image quality with better spatiotemporal fidelity and reduced noisy patterns. 6791 Blind compressive sensing dynamic MRI with sparse dictionaries Sajan Goud Lingala and Mathews Jacob The University of Iowa, Iowa city, IA, United States