(S052) Inverse Optimization for Correlating 4DCT Ventilation Imaging and Radiation Dose
A numerical method for computing a DIR transformation according to a target ventilation image was used to generate a ventilation image that correlates precisely with the dose distribution while maintaining high DIR spatial accuracy. Thus, by employing this approach, the focus of future CT ventilation studies that are designed to assess radiation dose response is reduced to assessing the physical feasibility of the DIR transformation that generates the ventilation image predicted by the dose-response model.
Edward Castillo, PhD, Richard Castillo, PhD, Thomas Guerrero, MD, PhD; Beaumont Health System; UT Medical Branch
PURPOSE: Four-dimensional computed tomography (4DCT) ventilation imaging is an emerging modality with utility in thoracic radiotherapy treatment planning. Though recent studies have demonstrated its potential for quantifying the functional response of lung tissue to irradiation, ventilation image analysis is challenging and difficult to reproduce because of issues such as spatial inaccuracies in the required deformable image registration (DIR), cine mode phase-binning artifacts, and variations in the patient’s breathing. In order to address these issues, we have developed a numerical method for computing 4DCT ventilation images that correlate perfectly with a given radiation dose distribution or dose-response model while maintaining high-spatial accuracy in the corresponding DIR solution.
METHODS: Ventilation images are defined by a DIR spatial transformation that maps the position of inhale lung voxels to their corresponding position at exhale (or vice versa). A voxel’s volume change under the transformation, described mathematically by the Jacobian matrix, is the intensity value of the ventilation image and quantifies the air exchanged. Given an initial DIR estimate and a target ventilation image, we define an optimization problem describing the spatial transformation closest to our initial estimate (according to the L2-norm), with Jacobian values equal to those in the target ventilation image.
RESULTS: The inhale/exhale phases of a treatment-planning 4DCT for a patient with non–small-cell lung cancer were registered using a previously reported DIR method. The spatial accuracy of the DIR solution, given as the average (standard deviation) millimeter error with respect to 417 expert-determined landmark points, was 1.03 (1.01) mm. The radiation dose distribution image was used to generate a target ventilation image using a linear dose-response model applied to the original ventilation image. Our numerical method was then applied to the initial DIR solution to produce a spatial transformation and corresponding ventilation image that matched the target ventilation image within 1e-2.The average millimeter error for the new transformation was 1.11 (1.10).
CONCLUSION: A numerical method for computing a DIR transformation according to a target ventilation image was used to generate a ventilation image that correlates precisely with the dose distribution while maintaining high DIR spatial accuracy. Thus, by employing this approach, the focus of future CT ventilation studies that are designed to assess radiation dose response is reduced to assessing the physical feasibility of the DIR transformation that generates the ventilation image predicted by the dose-response model.
Proceedings of the 97th Annual Meeting of the American Radium Society - americanradiumsociety.org
Frontline Chemo-Free Regimen Supported in HR+/HER2+ Breast Cancer Therapy
January 1st 2024Combining anastrozole with palbociclib, trastuzumab, and pertuzumab as a frontline therapy for hormone receptor–positive, HER2-positive breast cancer may avoid some of the toxicities associated with chemotherapy, says Amy Tiersten, MD.
Oncology On-The-Go Podcast: ASCO 2023 Recap
June 19th 2023Experts from University of California, Los Angeles Health and Mayo Clinic discuss key data presented at the 2023 American Society of Clinical Oncology (ASCO) Annual Meeting in the gynecologic and gastrointestinal cancer spaces and how they may impact patient care.
