Click here to


Are you sure ?

Yes, do it No, cancel

Dosimetric Impact of Residual Distortion in a 3T MR Simulator Used for Stereotactic Radiosurgery Treatment Planning

Y Sheng1*, Y Cui1 , S Elguindi2 , J Kirkpatrick1 , F Yin1 , E Subashi2 , (1) Duke University Medical Center, Durham, NC, (2) Memorial Sloan Kettering Cancer Center, New York, NY


(Tuesday, 7/16/2019) 4:30 PM - 5:30 PM

Room: Exhibit Hall | Forum 2

Purpose: To evaluate the dosimetric impact of residual distortion in a 3T MR simulator used for stereotactic radiosurgery treatment planning.

Methods: Gradient-nonlinearity (GNL) distortion was measured using an anthropomorphic head phantom containing a 3D grid of rods. Intersection points identified in a high-resolution CT (voxel size: 0.6x0.6x0.6mm³) were used as the ground truth for distortion estimation. The phantom was scanned using a T1-weighted 3D SPGR sequence with 1 mm³ isotropic voxels. GNL-distortion was corrected using the vendor’s approach based on 2D/3D spherical harmonics deconvolution. Distortion magnitude was calculated as the distance between coordinates of control points in MR and CT image. A model for the distortion vector-field was generated by considering the magnitude of measured distortion as a function of distance-to-isocenter in left-right, anterior-posterior, and superior-inferior direction. The dosimetric impact of residual distortion was studied in 63 patients treated with SRS. MR-isocenter was identified on each plan and the residual distortion for both planning target volume (PTV) and gross tumor volume (GTV) was calculated. The effects of residual GNL distortion were studied using D99, V100, minimum GTV dose, and Paddick conformity index (CI).

Results: Median GNL-distortion was significantly lower when distortion correction was applied and did not depend on correction method (Kruskal-Wallis test, α=0.05). Mean surface distortion was between 0.17 and 1.31mm and was primarily dictated by target-to-MR-isocenter distance(R2=0.59). In the PTV, the median (inter-quartile range, IQR) difference for D99 and V100 was -1.21%(3.67%) and -0.39%(2.03%), respectively. Median (IQR) difference of GTV minimum dose was -0.32%(2.19%), resulting in 8(12.7%) GTVs receiving dose less than 100% Rx. For these 8 cases, median (IQR) target-to-MR-isocenter distance was 8.4cm(3.1cm). Median (IQR) difference of CI for all plans was -0.04(0.03).

Conclusion: Residual distortion must be evaluated for a 3T MR simulator used for SRS treatment planning.




TH- External beam- photons: intracranial stereotactic/SBRT

Contact Email