Adaptive MRI Guided Brachytherapy of Cervix Cancer: Impact of Reference Image Selection on Dose Accumulation

Abstract

Online imaging and deformable image registration (DIR) are required for an accurate dose accumulation of fractionated treatments. Dosimetry may depend on the selection of reference image, to which fractional doses are accumulated. This study reviews the impact of reference image selection on the accumulated dosimetry for adaptive MRI guided brachytherapy of cervix cancer. Twelve cervical cancer patients underwent one planning and two daily pelvic MRI scans during 3 day-long PDR brachytherapy. A brachytherapy plan was optimized on planning MRI respecting GEC-ESTRO recommendations and the plan was repeatedly used on following daily MRIs for delivered dose computation, resulting in three delivered dose distributions on each patient. To evaluate plan quality, D2cc of rectum/bladder wall and D90 of HRCTV/IRCTV were estimated by (1) simply taking average of each D2cc and D90 from three delivered dose distributions and more accurately measured from (2) DIR based dose accumulation using one of the MRI images as a reference (totaling three accumulations doses). Standard deviations of D2cc and D90 from three accumulations were considered as a measure of accumulation dose uncertainty. The evaluation parameters were compared among planned, averaged, and accumulated dose distributions. Relative to the clinical plan, the accumulated D90s of HRCTV/IRCTV varied by, on average, 2.1% (range 0.2∼3.0%) and 3.2% (0.8∼9.3%) respectively, depending on the selection of the reference image. The differences from planned dose were, on average, 3.3% (-12.3∼20.4%) and 1.8% (-9.5∼18.8%), respectively. Averaged D90s of HRCTV/IRCTV were less than the accumulated dose by 7.6% (3.4∼17.0%) and 7.7% (2.7∼13.5%), demonstrating a good conservative estimate for target coverage. Averaged D2cc estimated the accumulated D2cc well and were found within 2 STDs in all except one patient. 6 patients out of 12 had larger accumulated dose variation than difference from planned dose in at least one evaluation parameter. Accumulated dose using DIR was significantly different from the planned dose (p < 0.05, paired t-test) in PDR brachytherapy plans, demonstrating the need for adaptive planning. However, care should be taken in the interpretation of the result due to the dosimetric uncertainty of accumulation process depending on the selection of reference image. Averaging showed conservative estimate for target coverage and good D2cc agreement relative to accumulation.Poster Viewing Abstract 3485; TableEvaluation parameters on 12 patients relative to the clinical plan (%)Average∗Averaged from three delivered dose distributionsAccum 1†Accum n is DIR based dose accumulation using n-th MRI as a reference imageAccum 2Accum 3Mean of accumSTD of accum‡Standard deviation of accumulated dose in each patient.HRCTV D9089.2 (73.5∼100.0)98.8 (82.4∼115.6)95.7 (78.3∼111.7)95.8 (78.3∼109.7)96.7(79.7∼112.3)2.1 (0.2∼3.2)IRCTV D9090.6 (74.7∼99.7)100.8 (84.6∼108.2)96.9 (78.3∼106.5)97.0 (80.8∼114.5)98.2 (81.2∼109.5)3.2 (0.8∼9.3)Bladder wall D2cc95.0 (71.9∼119.4)93.7 (74.1∼111.8)94.7(73.8∼135.5)92.8 (69.8∼115.9)93.7 (73.0∼115.6)6.7 (1.7∼18.4)Rectum wall D2cc109.9 (93.0∼143.3)109.2(93.5∼124.0)109.2 (87.1∼139.3)108.7 (91.2∼152.8)109.1 (90.6∼138.7)4.8 (1.7∼14.4)() Shows range of the data∗ Averaged from three delivered dose distributions† Accum n is DIR based dose accumulation using n-th MRI as a reference image‡ Standard deviation of accumulated dose in each patient. Open table in a new tab