Encrypted login | home

Program Information

Comparison of Two Deformable Image Registration Algorithms for MRI-Guided CT-Based Brachytherapy in Locally Advanced Cervical Cancer

Y Rong

Y Rong*, B Dyer , C Wright , S Benedict , J Mayadev , University of California Davis Comprehensive Cancer Center, Sacramento, CA


WE-RAM3-GePD-J(B)-5 (Wednesday, August 2, 2017) 10:30 AM - 11:00 AM Room: Joint Imaging-Therapy ePoster Lounge - B

Purpose: This study aims to compare the accuracy and usefulness of two commercial deformable image registration (DIR) algorithms for MR-guided CT-based cervical cancer brachytherapy.

Methods: Twenty patients previously treated with the HDR brachytherapy were identified for this study. Each patient received pre-MRI (acquired prior to the applicator implantation) and post-CT (acquired after the applicator implantation). Pre-MRI was registered and deformed to post-CT using both the image value-based and biomechanical model-based (MORFEUS) DIR algorithms. MORFEUS requires controlling structures, which are rectum, bladder, uterus, and cervix for this study. Each post-CT has a physician-drawn CTV (CTV-BT), based on cervix location and disease extension. Additional CTV and GTV were delineated on the pre-MRI by the attending, and projected onto the post-CT using reversed deformation vector fields (CTV' and GTV', respectively). Evaluation included dice indices, mesh point difference (MPD), and center of mass (COM) difference. Clinical usefulness was scored by the same attending based on their value in assisting optimal target delineation on post-CTs.

Results: Dice index and MPD results for four controlling structures demonstrated high accuracy in image deformation for most of patients using MORFEUS. The average dice indices (range) between the physician-drawn and the DIR-mapped structures on the post-CT image for cervix, uterus, rectum, and bladder are 0.83(0.50-0.97), 0.87(0.44-0.99), 0.76(0.39-1.0), and 0.90(0.66-0.99), respectively. Comparing the differences in COM (in cm) CTV' and GTV' to the CTV-BT, results showed an average (range) of 1.25(0.31-2.80) and 1.08(0.19-2.20) in cm, respectively. The export scoring of clinically useful showed 16 cases that would have resulted in either a confirmation or modification of the CTV-BT for brachytherapy plans.

Conclusion: This is the first clinical study of evaluating DIR algorithms for MR-CT image deformation for Brachytherapy. Results of MORFEUS algorithm showed high accuracy and clinical usage with the pre-MRI in target definition for value-based cervical cancer brachytherapy.

Contact Email: