Towards 3D Dynamic MRI of the Lung using Blind Compressed Sensing

3-D dynamic MRI of the lung is a promising tool to assess lung function and mechanics. Compared to multi-slice 2D-DMRI, 3-D acquisitions enables the accurate estimation of lung volumes and its variations. However, its full potential is not clinically realized due to restricted spatio-temporal resolutions and volume coverage. In this work, we propose to employ a blind compressed sensing (BCS) scheme to overcome existing trade-offs with 3D-DMRI. The BCS scheme exploits the sparsity of the dynamic dataset in a dictionary of temporal bases that are estimated from the measurements. Since the bases are learnt from the data at hand, they are more representative of the temporal variations within the data, and are expected to provide sparser representations than compressed sensing (CS) schemes that utilize predetermined bases. In addition, it does not require any assumptions on the breathing conditions. Additionally, we propose to combine BCS with parallel imaging and golden angle (GA) radial sampling; the combination offers superior incoherence properties. With the BCS scheme, we show feasibilities of imaging the lung during normal breathing at spatial resolutions and time resolutions of upto (2.37mm2, 0.72 sec) slice coverage (16 slices, 4 mm thickness)

This abstract and the presentation materials are available to members only; a login is required.

Join Here