Sequential combination of k‐t principle component analysis (PCA) and partial parallel imaging: k‐t PCA GROWL

Purpose k‐t principle component analysis (k‐t PCA) is a distinguished method for high spatiotemporal resolution dynamic MRI. To further improve the accuracy of k‐t PCA, a combination with partial parallel imaging (PPI), k‐t PCA/SENSE, has been tested. However, k‐t PCA/SENSE suffers from long reconstruction time and limited improvement. This study aims to improve the combination of k‐t PCA and PPI on both reconstruction speed and accuracy. Methods A sequential combination scheme called k‐t PCA GROWL (GRAPPA operator for wider readout line) was proposed. The GRAPPA operator was performed before k‐t PCA to extend each readout line into a wider band, which improved the condition of the encoding matrix in the following k‐t PCA reconstruction. k‐t PCA GROWL was tested and compared with k‐t PCA and k‐t PCA/SENSE on cardiac imaging. Results k‐t PCA GROWL consistently resulted in better image quality compared with k‐t PCA/SENSE at high acceleration factors for both retrospectively and prospectively undersampled cardiac imaging, with a much lower computation cost. The improvement in image quality became greater with the increase of acceleration factor. Conclusion By sequentially combining the GRAPPA operator and k‐t PCA, the proposed k‐t PCA GROWL method outperformed k‐t PCA/SENSE in both reconstruction speed and accuracy, suggesting that k‐t PCA GROWL is a better combination scheme than k‐t PCA/SENSE. Magn Reson Med 77:1058–1067, 2017. © 2016 International Society for Magnetic Resonance in Medicine