Validation of an elastic registration technique to estimate anatomical lung modification in Non-Small-Cell Lung Cancer Tomotherapy

Radiation Oncology. 2011;6(1):31 DOI 10.1186/1748-717X-6-31

 

Journal Homepage

Journal Title: Radiation Oncology

ISSN: 1748-717X (Online)

Publisher: BMC

LCC Subject Category: Medicine: Medicine (General): Medical physics. Medical radiology. Nuclear medicine | Medicine: Internal medicine: Neoplasms. Tumors. Oncology. Including cancer and carcinogens

Country of publisher: United Kingdom

Language of fulltext: English

Full-text formats available: PDF, HTML

 

AUTHORS


Persano Diego

Dell'Oca Italo

Ciavarro Cristina

Cattaneo Giovanni M

Faggiano Elena

Calandrino Riccardo

Rizzo Giovanna

EDITORIAL INFORMATION

Blind peer review

Editorial Board

Instructions for authors

Time From Submission to Publication: 13 weeks

 

Abstract | Full Text

<p>Abstract</p> <p>Background</p> <p>The study of lung parenchyma anatomical modification is useful to estimate dose discrepancies during the radiation treatment of Non-Small-Cell Lung Cancer (NSCLC) patients. We propose and validate a method, based on free-form deformation and mutual information, to elastically register planning kVCT with daily MVCT images, to estimate lung parenchyma modification during Tomotherapy.</p> <p>Methods</p> <p>We analyzed 15 registrations between the planning kVCT and 3 MVCT images for each of the 5 NSCLC patients. Image registration accuracy was evaluated by visual inspection and, quantitatively, by Correlation Coefficients (CC) and Target Registration Errors (TRE). Finally, a lung volume correspondence analysis was performed to specifically evaluate registration accuracy in lungs.</p> <p>Results</p> <p>Results showed that elastic registration was always satisfactory, both qualitatively and quantitatively: TRE after elastic registration (average value of 3.6 mm) remained comparable and often smaller than voxel resolution. Lung volume variations were well estimated by elastic registration (average volume and centroid errors of 1.78% and 0.87 mm, respectively).</p> <p>Conclusions</p> <p>Our results demonstrate that this method is able to estimate lung deformations in thorax MVCT, with an accuracy within 3.6 mm comparable or smaller than the voxel dimension of the kVCT and MVCT images. It could be used to estimate lung parenchyma dose variations in thoracic Tomotherapy.</p>