Feng Huang1, Wei Lin1, Chiel den Harder2, Gabrielle Beck2, Clemens Bos3, George Randy Duensing1, Arne Reykowski1
1Invivo Corporation, Gainesville, FL, United States; 2Advanced Solutions, MRI, Philips Healthcare, Best, Netherlands; 3MR Clinical Science, Philips Healthcare, Best, Netherlands
Most existing motion compensation techniques for brain MRI assume that the motion is rigid. In fact, many kinds of non-rigid motion such as eye movement (eye ball rolling), skin movement (frowning), and jaw movement (swallowing, yawning), can also cause serious spatially local artifacts in brain imaging. These inevitable problems have not been carefully addressed. To remove these artifacts, data rejection and reconstruction with remaining unpolluted data can be used. However, the reconstruction with partial data could result in potential artifacts, such as reduced SNR and loss of contrast. In this work, it is proposed to use the image reconstructed with the full k-space, which is locally artifact corrupted but with high SNR, as a regularization in reconstruction to achieve an image with low artifact level and high SNR.