Premium
TU‐AB‐202‐07: A Novel Method for Registration of Mid‐Treatment PET/CT Images Under Conditions of Tumor Regression for Patients with Locally Advanced Lung Cancers
Author(s) -
Sharifi Hoda,
Zhang Hong,
Jin JianYyue,
Kong FengMing,
Chetty Indrin J,
Zhong Hualiang
Publication year - 2016
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4957429
Subject(s) - voxel , image registration , nuclear medicine , radiation therapy , lung cancer , medicine , radiology , computer science , artificial intelligence , image (mathematics) , pathology
Purpose: In PET‐guided adaptive radiotherapy (RT), changes in the metabolic activity at individual voxels cannot be derived until the duringtreatment CT images are appropriately registered to pre‐treatment CT images. However, deformable image registration (DIR) usually does not preserve tumor volume. This may induce errors when comparing to the target. The aim of this study was to develop a DIR‐integrated mechanical modeling technique to track radiation‐induced metabolic changes on PET images. Methods: Three patients with non‐small cell lung cancer (NSCLC) were treated with adaptive radiotherapy under RTOG 1106. Two PET/CT image sets were acquired 2 weeks before RT and 18 fractions after the start of treatment. DIR was performed to register the during‐RT CT to the pre‐RT CT using a B‐spline algorithm and the resultant displacements in the region of tumor were remodeled using a hybrid finite element method (FEM). Gross tumor volume (GTV) was delineated on the during‐RT PET/CT image sets and deformed using the 3D deformation vector fields generated by the CT‐based registrations. Metabolic tumor volume (MTV) was calculated using the pre‐ and during–RT image set. The quality of the PET mapping was evaluated based on the constancy of the mapped MTV and landmark comparison. Results: The B‐spline‐based registrations changed MTVs by 7.3%, 4.6% and −5.9% for the 3 patients and the correspondent changes for the hybrid FEM method −2.9%, 1% and 6.3%, respectively. Landmark comparisons were used to evaluate the Rigid, B‐Spline, and hybrid FEM registrations with the mean errors of 10.1 ± 1.6 mm, 4.4 ± 0.4 mm, and 3.6 ± 0.4 mm for three patients. The hybrid FEM method outperforms the B‐Spline‐only registration for patients with tumor regression Conclusion: The hybrid FEM modeling technique improves the B‐Spline registrations in tumor regions. This technique may help compare metabolic activities between two PET/CT images with regressing tumors. The author gratefully acknowledges the financial support from the National Institutes of Health Grant