Premium
TU‐F‐17A‐09: Four‐Dimensional Cone Beam CT Ventilation Imaging Can Detect Interfraction Lung Function Variations for Locally Advanced Lung Cancer Patients
Author(s) -
Kipritidis J,
Hugo G,
Weiss E,
Williamson J,
Keall P
Publication year - 2014
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 - 10.1118/1.4889336
Subject(s) - medicine , nuclear medicine , cone beam computed tomography , percentile , lung cancer , cone beam ct , radiation therapy , image registration , voxel , radiology , computed tomography , mathematics , pathology , statistics , artificial intelligence , computer science , image (mathematics)
Purpose: Four‐dimensional cone beam CT ventilation imaging (4D‐CBCT VI) is a novel functional lung imaging modality requiring validation. We hypothesize that 4D‐CBCT VI satisfies a necessary condition for validity: that intrafraction variations (e.g. due to poor 4D‐CBCT image quality) are substantially different to interfraction variations (e.g. due to changes in underlying function). We perform the first comparison of intrafraction (pre/post fraction) and interfraction (week‐to‐week) 4D‐CBCT VIs for locally advanced non small cell lung cancer (LA NSCLC) patients undergoing radiation therapy. Methods: A total of 215 4D‐CBCT scans were acquired for 19 LA NSCLC patients over 4‐6 weeks of radiation therapy, including 75 pairs of pre‐/post‐fraction scans on the same day. 4D‐CBCT VIs were obtained by applying state‐of‐the‐art, B‐spline deformable image registration to obtain the Jacobian determinant of deformation between the end‐exhale and end‐inhale phases. All VIs were deformably registered to the corresponding first day scan, normalized between the 10th and 90th percentile values and cropped to the ipsilateral lung only. Intrafraction variations were assessed by computing the mean and standard deviation of voxel‐wise differences between all same‐day pairs of pre‐/post‐fraction VIs. Interfraction differences were computed between first‐day VIs and treatment weeks 2, 4 and 6 for all 19 patients. We tested the hypothesis by comparing cumulative distribution functions (CDFs) of intrafraction and interfraction ventilation differences using two‐sided Kolmogorov‐Smirnov goodness‐of‐fit tests. Results: The (mean ± std. dev.) of intrafraction differences was (−0.007 ± 0.079). Interfraction differences for weeks 2, 4 and 6 were (−0.035 ± 0.103), (−0.006 ± 0.094) and (−0.019 ± 0.127) respectively. For week 2, the changes in CDFs for intrafraction and interfraction differences approached statistical significance (p=0.099). Conclusion: We have shown that 4D‐CBCT VI can satisfy a necessary condition for validity; intrafraction variations do not limit the ability to measure interfraction variations in LA NSCLC patients. This work is supported by a Cancer Institute NSW Early Career Fellowship, an NHMRC Australia Fellowship and NIH P01CA116602.