Evaluating stereotactic accuracy with patient-specific MRI distortion corrections for frame-based radiosurgery

Abstract

Purpose: This study examines how MRI distortions affect frame-based SRS treatments and assesses the need for clinical distortion corrections.

Methods: The study included 18 patients with 80 total brain targets treated using frame-based radiosurgery. Distortion within patients' MRIs were corrected using Cranial Distortion Correction (CDC) software, which utilizes the patient's CT to alter planning MRIs to reduce inherent intra-cranial distortion. Distortion was evaluated by comparing the original planning target volumes (PTV_ORIG) to targets contoured on corrected MRIs (PTV_CORR). To provide an internal control, targets were also re-contoured on uncorrected (PTV_RECON) MRIs. Additional analysis was done to assess if 1 mm expansions to PTV_ORIG targets would compensate for patient-specific distortions. Changes in target volumes, DICE and JACCARD similarity coefficients, minimum PTV dose (D_min), dose to 95% of the PTV (D95%), and normal tissue receiving 12 Gy (V_12Gy), 10 Gy (V_10Gy), and 5 Gy (V_5Gy) were calculated and evaluated. Student's t-tests were used to determine if changes in PTV_CORR were significantly different than intra-contouring variability quantified by PTV_RECON.

Results: PTV_RECON and PTV_CORR relative changes in volume were 6.19% ± 10.95% and 1.48% ± 32.92%. PTV_RECON and PTV_CORR similarity coefficients were 0.90 ± 0.08 and 0.73 ± 0.16 for DICE and 0.82 ± 0.12 and 0.60 ± 0.18 for JACCARD. PTV_RECON and PTV_CORR changes in D_min were -0.88% ± 8.77% and -12.9 ± 17.3%. PTV_RECON and PTV_CORR changes in D95% were -0.34% ± 5.89 and -8.68% ± 13.21%. The 1 mm expanded PTV_ORIG targets did not entirely cover 14 of the 80 PTV_CORR targets. Normal tissue changes (V_12Gy, V_10Gy, V_5Gy) calculated with PTV_RECON were (-0.09% ± 7.39%, -0.38% ± 5.67%, -0.08% ± 2.04%) and PTV_CORR were (-2.14% ± 7.34%, -1.42% ± 5.45%, -0.61% ± 1.93%). Except for V_10Gy, all PTV_CORR changes were significantly different (p < 0.05) than PTV_RECON.

Conclusion: MRIs used for SRS target delineation exhibit notable geometric distortions that may compromise optimal dosimetric accuracy. A uniform 1 mm expansion may result in geometric misses; however, the CDC algorithm provides a feasible solution for rectifying distortions, thereby enhancing treatment precision.

Keywords: MRI; SRS; distortions.

FIGURE 1

Cranial distortion correction of an MRI using the CDC algorithm. (a) shows the CDC interface displaying the distortion grid. (b) shows the original (red) and corrected (green) contours overlaid on the uncorrected MRI. (c) displays the final corrected MRI.

FIGURE 2

Normalized change in PTV volumes resulting from re‐contouring (RECON) and distortion corrections (CORR), with respect to the original (ORIG).

FIGURE 3

Histograms of DICE similarity coefficients between re‐contoured PTVs (RECON) and distortion‐corrected PTVs (CORR) compared to the original (ORIG) PTVs.

FIGURE 4

Histograms of JACCARD similarity coefficients between re‐contoured PTVs (RECON) and distortion‐corrected PTVs (CORR) compared to original (ORIG) PTVs.

FIGURE 5

Box‐and‐whisker plots of normalized changes in minimum PTV dose (ΔD _min) and dose to 95% of the PTV (ΔD95%) resulting from re‐contouring (RECON) and after distortion corrections (CORR), with respect to original (ORIG) PTVs. The x‐axis labels show p‐values from the Student's t‐test comparing RECON and CORR data.

FIGURE 6

Box‐and‐whisker plots of normalized changes in the volume of normal tissue (brain—PTVs) receiving 12 Gy (ΔV12Gy), 10 Gy (ΔV10Gy), and 5 Gy (ΔV5Gy) resulting from re‐contouring (RECON) and after distortion corrections (CORR), with respect to the original (ORIG) PTVs. The x‐axis labels show p‐values from the Student's t‐test comparing RECON and CORR data.