Prospective Study Evaluating Metal Artifact Reduction in MRI-Based Cervical Cancer Intracavitary Brachytherapy

Abstract

The accuracy of MRI-based brachytherapy with titanium tandem and ovoid applicators for cervical cancer is affected by metal artifacts in imaging. We hypothesized that orthopedic metal artifact reduction (O-MAR) MRI may reduce artifact associated with the titanium tandem, and that variations in applicator reconstruction due to artifact can affect brachytherapy dosimetry. We performed a prospective clinical study to test these hypotheses. We enrolled 12 patients receiving brachytherapy for cervical cancer on a prospective institutional review board-approved imaging protocol. We intended for each patient to receive six MRI-based brachytherapy treatments interdigitated with intensity modulated radiation. After tandem and ovoid implantation, patients received MRI simulation on a 1.5 T scanner with T2-weighted (T2W), diffusion weighted imaging (DWI), proton density weighted imaging (PDW), and PDW with O-MAR. The O-MAR method incorporated slice-selective excitation pulses with large bandwidth, view angle tilting, and “weak” slice encoding for metal artifact correction with 7 z-phase encodes. The tumor was contoured on the DWI, bladder, and rectum on T2W, and the tandem was reconstructed based on the PDW. The primary endpoint was reduction in the dimensions of the “bloom” artifact associated with the tip of the titanium tandem as measured in the PDW and O-MAR images. We evaluated dosimetry effects of variations in reconstruction due to the artifact by creating mock brachytherapy plans in which the tandem and dwell positions were shifted distally by a distance equal to half the improvement in the artifact length in O-MAR compared to PDW. Values are reported as mean and range. Differences were evaluated using the paired t-test. We acquired 62 MRI simulation scans with O-MAR, out of 72 brachytherapy fractions. The tandem tip “bloom” artifact mean length was 6.7 mm (3.6-8.9) without O-MAR and 3.4 (1.4-6.7) with O-MAR (P < 0.01), and the artifact mean width was 7.5 mm (4.4-9.4) without O-MAR and 3.7 mm (1.8-5.9) with O-MAR (P < 0.01). The mock brachytherapy plan representing potential variation in tandem reconstruction was generated with a mean shift of 1.7 mm (0.6-3.0), and resulted in a mean decrease of 1.3% (-4.5 to 3.2) in point A dose (P < 0.01), 4.4% (-1.9 to 18) in mean tumor dose (P < 0.01), 1.5% (-8.9 to 7.4) in tumor D100% (P < 0.01), and 2.8% (-1.4 to 8.5) in tumor D90% (P < 0.01) compared to baseline plans. Additionally, the maximum bladder dose was not significantly changed (P = 0.08), but there was a mean decrease of 0.5% (-14 to 4.0) in mean bladder dose (P = 0.01), 1.2% (0.3 to 7.0) in maximum rectum dose (P < 0.01), and 1.1% (-2.9 to 2.9) in mean rectum dose (P < 0.01) compared to baseline plans. O-MAR decreases the artifact associated with the titanium tandem. Variability in reconstruction due to the artifact at the tip of a tandem may lead to a mean tumor dose of 4.4% less than expected, highlighting the importance of accurate applicator reconstruction in brachytherapy treatment planning.