Conjugate Gradient and DL Reconstructions for UTE-MRI: Reducing Scan Time without Image Quality and Nodule Detection Capability Degradations

Motivation: Conjugate gradient reconstruction (CG Recon) and DLR can reduce scan time without nodule detection performance degradation than grid reconstruction method (Grid Recon) on UTE-MRI. Goal(s): The goal was to determine utilities of CG Recon and DLR for scan time reduction without nodule detection performance degradation as compared with Grid Recon without DLR. Approach: This clinical study examined 35 patients by UTE-MRIs, and nodule detection performance was statistically compared. Results: Az of both UTEoriginals reconstructed with DLR and UTE1/2 reduced as 1/2 and reconstructed with CG Recon and DLR were significantly larger than that of UTEoriginal reconstructed with Grid Recon without DLR (p=0.03). Impact: When compared with conventional grid reconstruction, conjugate gradient reconstruction and deep learning reconstruction are useful for reducing acquisition time without degradations of image quality and nodule detection on pulmonary MRI with ultra-short TE.