Diffusion-weighted imaging (DWI) is a useful and noninvasive technique in detecting lesion and characterizing tissue type for brain and body applications (1-3). Recently, DWI has been shown to be useful in characterizing renal carcinoma with quantitative measurement of apparent diffusion coefficient (ADC) (4). Moreover, emerging clinical applications of kidney DWI, such as providing information on normal renal development in fetus (5) and evaluating focal lesion (6), enhance the clinical value of body DWI. However, with echo-planar imaging (EPI) based DWI protocols, the application of body DWI remains limited due to suboptimal EPI image quality. For instance, single-shot EPI suffers from geometric distortion that substantially degrades the image quality. Although the parallel imaging technique, such as sensitivity encoding (SENSE) (7), can help reduce geometric distortion, the undesirable noise amplification associated with SENSE reconstruction reduces the accuracy of DWI data. To address the geometric distortion problem without undesirable noise amplification, multi-shot EPI with multiplexed sensitivity encoding (MUSE) has been developed and shown to be useful in achieving high-resolution and high-quality DWI and DTI of brains (8, 9). Furthermore, the multi-band (MB) excitation can be incorporated into multi-shot EPI to improve the scan throughput. In this study, we propose to use multi-band multi-shot EPI to acquire kidney DWI data with reduced geometric distortion and bilateral coverage in sagittal-plane, demonstrating the feasibility of MB multi-shot DWI of body applications. The multi-band MUSE (MB-MUSE) reconstruction is used to simultaneously eliminate the aliasing artifact due to shot-to-shot phase variation and separate collapsed slices.A 2-band 4-shot interleaved DW-EPI pulse sequence utilizing two consecutive excitation and two refocusing RF pulses was used to acquire the DWI data from bilateral sides of kidney in sagittal-plane. Figure 1 schematically shows the coverage of bilateral sides of kidney achieved by using 2-band excitation with a gap between two slice groups. Figure 2a shows the image reconstructed from direct Fourier transform of 2-band 4-shot interleaved DW-EPI data that suffers aliasing artifact due to shot-to-shot phase variation. To simultaneously address the shot-to-shot phase variation problem and separate two collapsed slices, the multi-band MUSE reconstruction with adaptive partial Fourier reconstruction was applied to Nyquist corrected EPI data (9), producing aliasing-free multi-band images (Figures 2b and 2c). The data were acquired on a 3T MRI scanner (GE MR750, Waukesha, WI) from a healthy volunteer using a 32-channel phase-array abdominal coil. The other image parameters included: matrix size = 192 x 192, partial-Fourier factor = 62.5%, TE/TR = 67.9ms/2000ms, FOV = 28 cm, thickness = 4 mm, number of slice = 16 (i.e., 32 slices after MB reconstruction), and b-value = 500s/mm2 with 6 non-collinear diffusion directions, and total scan time = 58 sec.Figure 3 shows 2-band 4-shot T2-weighted (upper panel) and DW (lower panel) interleaved EPI images acquired in sagittal-plane and reconstructed from Nyquist corrected data with direct Fourier transform. Figures 4 and 5 show T2-weighted and DW images, respectively, of bilateral sides of kidney reconstructed from the same MB multi-shot EPI data set using MB-MUSE algorithm.This study demonstrates the feasibility of achieving high-resolution high-quality kidney DWI image with reduced geometric distortion. The new scan and reconstruction methods have the following strengths. First, with multi-band excitation, the scan throughput of multi-shot interleaved DW-EPI can be improved. Second, aliasing artifacts due to shot-to-shot phase variation in multi-band multi-shot DWI can be removed in post-processing using a MB-MUSE algorithm, without relying on navigator acquisition. Third, the SNR penalty (g-factor) of MB-MUSE reconstruction is substantially less pronounced than conventional MB-SENSE reconstruction, thereby enabling high-resolution and high-quality DWI. The major limitation of MB-MUSE protocol is that the number of shots cannot exceed the number of coil elements along the phase encoding direction, and the maximal MB factor depends on geometry of coil elements along the slice direction. In addition to bilateral acquisition of kidney DWI as demonstrated in this study, another application of our method is sagittal-plane DWI of bilateral sides of breast (with half of scan time as compared with signal-band acquisition). Further studies need to be conducted to investigate the quantitative accuracy of ADC mapping with MB-MUSE technique. In conclusion, the MB-MUSE technique is a clinical imaging tool suitable for high-quality and high-resolution body DWI with reduced geometric distortion.