J. S. Kim, C. C. Conlin, M. A. Anderson, T. Barrett, S. Do, L. Georgiou, M. G. Harisinghani, M. A. Liss, R. Manger, D. J. Margolis, R. M. Marks, P. M. Murphy, N. Nakrour, M. Ohliger, T. Pawlicki, R. Rakow-Penner, M. Rojo Domingo, J. Scholey, S. N. Seyedin, Y. Song, N. Wehrli, S. Woolen, C. Zamboglou, T. M. Seibert, A. M. Dale
Purpose: B0-inhomogeneity distortion limits the utility of diffusion weighted imaging (DWI) for prostate cancer (PCa) lesion targeting for focal radiotherapy. Distortion correction methods like the Reversed Polarity Gradient (RPG) method and FSL-topup struggle due to signal-intensity changes from voxel stretch/overlap. We propose a technique called multi-b RPG (mRPG) for correcting B0-inhomogeneity distortions that eliminates confounding signal-intensity changes and demonstrate that it outperforms alternative methods. Methods and Materials: mRPG employs DWI volumes of opposite phase-encoding polarity acquired at different b-values. Signal-intensity changes from voxel stretch/overlap are eliminated by normalizing the volumes by the sum across all b-values. Tissue deformation is then estimated by registering normalized volumes of opposite phase-encoding polarity. Patients with known/suspected PCa were included if scanned with a protocol including two multi-b-value DWI series with opposite phase-encoding polarity. Degree of anatomic distortion was quantified by the Pearson correlation between opposite-polarity volumes; less distortion yields higher correlation. Correlation coefficients were computed for DWI and ADC volumes of each patient, before and after correction by RPG, topup, and mRPG. Absolute tissue displacement was measured along the posterior prostate border and within PCa lesions. Results: 1225 patients were available for analysis. mRPG yielded significant improvements in median correlation (p<0.001) between opposite-polarity DWI volumes (0.90 vs 0.83 uncorrected) and ADC volumes (0.78 vs 0.71). The increase achieved with mRPG was significantly larger (p<0.001) than with either RPG or topup on DWI (0.90 vs 0.84 with RPG; 0.81 with topup) or ADC (0.78 vs 0.71 with RPG; 0.67 with topup). Tissue displacement during correction ranged from 0.5-13.8mm along the posterior border of the prostate and 0.5-7.5mm within PCa lesions. Displacement [≥]4mm was observed in 30% of patients. Conclusions: mRPG is a feasible method for correcting B0-inhomogeneity distortions in prostate DWI that avoids drawbacks of alternative methods to achieve better distortion correction performance.