Restriction Spectrum Imaging (RSI) is an advanced diffusion imaging technique that has potential to correct for B0 distortions and non-invasively predict tumor grade. This abstract is an initial evaluation of RSI in breast imaging. RSI has already been shown to be useful in brain imaging and prostate imaging [1,2]. Breast diffusion imaging in particular would benefit from this technique due to relatively greater sensitivity to B0 inhomogeneity because of the off-center location of the breasts, adjacent lungs and ambient air. By increasing sensitivity to the intracellular restricted component of the diffusion signal, this technique also allows for increased specificity for tumor rather than post treatment edema/hemorrhage and other benign findings. RSI optimizes the signal from the spherically restricted diffusion pool and attenuates the signal from all other diffusion pools. This theoretically allows for isolation of signal from tightly packed tumor cells. The output from these techniques has been termed the tumor ‘cellularity map’ (CM). By isolating the spherical restricted compartment from the cylindrically restricted and hindered compartments, RSI-CM provides significantly greater conspicuity in distinguishing tumor from normal appearing tissue when compared with conventional imaging measures. Eleven patients with a new diagnosis of breast cancer (biopsy proven) underwent multi-parametric MRI to evaluate extent of disease. The protocol included Gadolinium contrast enhanced MRI, standard diffusion-weighted imaging (b=0,800 s/mm2) imaging, and diffusion tensor (0, 500, 1500, 4000 s/mm2 in 6, 6, and 15 directions) imaging for restriction spectrum imaging cellularity maps (CM). RSI-CM are distortion-corrected. To correct for distortion, forward and reverse trajectories were collected at b = 0 s/mm2. The distortion-correction algorithm [3] for the RSI-MRI sequence utilizes the symmetry of the distortion from B0 inhomogeneity. By collecting images at b = 0 s/mm2 in both the forward and reverse phase encode trajectories, a deformation field map was calculated and used to correct the entire diffusion data set. RSI Z-score maps were made by normalizing the data to the normal appearing breast tissue across all data sets. Dynamic contrast-enhanced MRI, standard diffusion MRI, and RSI-MRI were compared and evaluated with known biopsy results. An ROI was drawn on the post contrast image in the area of known cancer for each patient. A second ROI was drawn on the post contrast images in a region of likely cancer free tissue. These ROIs were then used to evaluate the Z-score on the RSI maps and the ADC maps. Both the RSI maps and the ADC maps were corrected for B0 distortions.RSI-MRI detected breast cancer in all 11 subjects evaluated. The average Z-score measured in cancer was 13.4, with the min and max at 2.3 and 25, respectively (Figure 1). With the Z-score statistic, 2 is usually considered significant. The normal tissue Z-score across patients measured 0.19, well below the significant threshold. For the diffusion maps, the mean ADC for cancer was 1.2, while the mean for the normal tissue was 1.12. This was unexpected as ADC is usually less than cancer. This variation is likely due to a couple of cases where fat suppression failed on the diffusion acquisition. In Figure 2, please note that the patient’s breast cancer was difficult to evaluate and visualize on post contrast imaging, and standard diffusion imaging. RSI however clearly identified the cancer in the medial aspect of the patient’s left breast (arrow). Figure 3 demonstrates 2 examples of Z-score maps.As demonstrated in this abstract, RSI dramatically increases conspicuity of cancer. On average, the Z-score of healthy tissue and normal tissue is separated by 12 points. On the other hand the ADCs of cancer and normal tissue were not significantly different in this patient population. Future research includes evaluating whether the Z-score can predict tumor nuclear grade.