Room: Exhibit Hall | Forum 2
Purpose: Using MR images as a primary dataset for radiotherapy planning requires corrections fo geometric distortion and non-uniform intensity. We seek to determine: 1) the effectiveness of correction for gradient-non-linearity and susceptibility effects on both QUASAR GRID3D and CIRS phantoms; 2) the magnitude and location of regions of residual distortion before and after correction
Methods: Phantom Study: MR, CT and CBCT images of QUASAR GRID3D and CIRS head phantoms were acquired. Axial T1 MPRAGE images were acquired using MAGNETOM Siemens Aera 1.5T RT edition, and both vendorsâ€™ commercially available software used to analyze images. Patient Study: Ten patients were MRI scanned for stereotactic radiosurgery treatment. Correction algorithm: The field map images were acquired with TE1=9.52ms/TE2=4.76ms, two magnitude and one phase difference image set. Then, the field map was created in FSL software. A custom MATLAB program was used to calculate geometric distortion in the frequency encoding direction, and 3D interpolation was applied to convert it to the size of MPRAGE images. The MPRAGE images were warped according to the interpolated field map in the frequency encoding direction. MIM software was used to fuse corrected and non-corrected MR images, deformable registered, and to generate the difference distortion map.
Results: Maximum deviation improvements: GRID3D, 0.27 mm in y-direction, 0.07 mm in z-direction and 0.23 mm in x-direction. CIRS, 0.34 mm, 0.1 mm and 0.09 mm at 20, 40- and 60-mm diameters from isocenter. Patient data show a correction range of 0.2â€“1.2 mm based on location. The most-distorted areas are around air-cavities, e.g. sinuses and eyes.
Conclusion: The phantom data show the validity of our fast distortion correction algorithm. Patient-specific data are acquired in <2 min and analyzed and available for planning in less than a minute. The data show non-uniformity of distortion magnitude at the similar locations between all patients.