Combined Data Acquisition of Multi-Contrast Images

We propose a technique to reduce clinical MRI scan time by sharing k-space data between images. To improve data utilization, acquisition of multiple images of different contrast was combined into one single scan with variable acquisition parameters. We term this approach “combo” acquisition. As a proof of concept, simulations of MRI experiments using spin echo and fast spin echo sequences and based on Bloch equations were performed. Scan time reductions of 20%-52% were achieved. Artifacts were minimized through systematic optimization of the phase encoding order and variation schemes for selected acquisition parameters. Introduction Scan time reduction in MRI remains an important issue, especially in a clinical setting. Previous approaches to reduce scan time focused either on increasing the speed of acquisition of k-space data [l], [Z], [3] or on completing reduced data sets through computations [4], [ 5 ] . Our work attempts to prove the concept of combining the acquisition of images of different contrast into one acquisition through sharing of k-space data in conventional Fourier encoding. Different contrast weightings for multiple images were obtained by continuously varying acquisition parameters during acquisition. Results Results were obtained for multi-contrast combo acquisitions and compared with images from standard acquisitions in terms of image contrast, quality, and resolution. Fig. 1 shows images from a SE combo acquisition together with corresponding variation schemes for TR and TE. As seen in Fig. 1, images from the combo acquisitions had sufficient contrast definition and did not exhibit any major artifacts. Scan time reductions of 30%-52% were obtained for SE combo acquisitions compared to separate standard SE acquisitions. In addition, the quantitative evaluation of the images from the SE combo acquisition showed an only slightly widened PSF, thus, minor ringing artifacts across tissue boundaries, but no contrast degradation. The SNR was comparable to the one from standard acquisitions. Fig. 2 shows images obtained with a FSE combo acquisition and the corresponding variation schemes for TR, echo spacing (ESP), and ETL. Artifacts were sufficiently suppressed, and scan time was reduced by 20% when compared to a combination of a SE sequence for the T,-weighted image and a dual-contrast FSE sequence for the other two images that only acquired the lowest phase encoding quadrant for both images separately. Compared to a combination with a dual-contrast FSE sequence with no data sharing, scan time was reduced bv 37%. Fig. 1: Simulated images from SE combo acquisition: TI(top left), PD(top right), and Tz-weighted (bottom left), and corresponding parameter variation schemes (bottom right) for TR (solid) and TE (dashed), where data sharing between images is illustrated. Actual ordering of phase encoding is not shown. Methods Simulations of MRI data acquisition were performed using Bloch equations and a realistic brain phantom model from the McConnell Brain Imaging Centre of the McGill University in Montreal [6]. Multicontrast combo acquisitions employing spin echo (SE) or fast spin echo (FSE) sequences yielded TI-, proton density (PO)-, and T2weighted images of the same slice. Acquisition parameters TR and TE were continuously vaned for SE sequences; TR, echo spacing (ESP), and echo train length (ETL) for the FSE sequence. Varying acquisition parameters causes non-uniform and tissue-dependent data weighting in k-space, which inevitably introduces artifacts in the image domain. Thus, parameter variation and PE schemes were systematically optimized for combo acquisitions to minimize visible artifacts. Acquiring the zero phase encoding view with the same settings as for a standard sequence preserved desired contrast in the resulting images. Finally, results from combo acquisitions using the SE sequence were further evaluated by quantitatively analyzing the point spread function (PSF), image profiles across tissue boundaries, preservation of contrast, and the signal-to-noise-ratio (SNR) for each image. Fig. 2: Simulated images from FSE combo acquisition: TI(top left), PD(top right), and Tz-weighted (bottom left), and corresponding parameter variation schemes (bottom right) for TR (solid) and echo spacing (ESP) (dashed), where data sharing between images is illustrated. Actual ordering of phase encoding is not shown. Discussion Simulations of MRI experiments were run to test the concept of kspace data sharing between images of multiple contrasts in combo acquisitions. Image contrast was well preserved, and artifacts were significantly suppressed through systematic optimization of the method. Signal oscillations due to non steady state effects were numerically small. In a clinical setting, such combined acquisition also simplifies the protocol setup and increases the scan throughput. For future work, we envision that the concept of a combined acquisition of multiple images with different contrast mechanisms will be extended to a more general computerized optimization of parameter variation schemes, k-space sampling and data sharing in MR imaging. References[ I ] P. Mansfield, et al.,J; Phys. E: Scient. Instrum., 9: 271, 1976[2] J. Hennig, et al., MRM, 3(6): 823-33, 1986.[3] P. van der Meulen, et al., M U , 3(3): 297-9, 1985.[4] D. K. Sodickson, et al., MRM, 38(4): 591-603, 1997.[ S ] K. P. Pruessmann, et al., MRM, 42(5): 952-62, 1999.[6]D. L. Collins, et al., IEEE TMI, 17(3): 463-8, 1998.