skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Automatic Delineation of On-Line Head-And-Neck Computed Tomography Images: Toward On-Line Adaptive Radiotherapy

Abstract

Purpose: To develop and validate a fully automatic region-of-interest (ROI) delineation method for on-line adaptive radiotherapy. Methods and Materials: On-line adaptive radiotherapy requires a robust and automatic image segmentation method to delineate ROIs in on-line volumetric images. We have implemented an atlas-based image segmentation method to automatically delineate ROIs of head-and-neck helical computed tomography images. A total of 32 daily computed tomography images from 7 head-and-neck patients were delineated using this automatic image segmentation method. Manually drawn contours on the daily images were used as references in the evaluation of automatically delineated ROIs. Two methods were used in quantitative validation: (1) the dice similarity coefficient index, which indicates the overlapping ratio between the manually and automatically delineated ROIs; and (2) the distance transformation, which yields the distances between the manually and automatically delineated ROI surfaces. Results: Automatic segmentation showed agreement with manual contouring. For most ROIs, the dice similarity coefficient indexes were approximately 0.8. Similarly, the distance transformation evaluation results showed that the distances between the manually and automatically delineated ROI surfaces were mostly within 3 mm. The distances between two surfaces had a mean of 1 mm and standard deviation of <2 mm in most ROIs. Conclusion: With atlas-basedmore » image segmentation, it is feasible to automatically delineate ROIs on the head-and-neck helical computed tomography images in on-line adaptive treatments.« less

Authors:
 [1];  [2];  [2];  [2]
  1. Department of Radiation Oncology, William Beaumont Hospital, Royal Oak, MI (United States). E-mail: tiezhi.zhang@beaumont.edu
  2. Department of Radiation Oncology, William Beaumont Hospital, Royal Oak, MI (United States)
Publication Date:
OSTI Identifier:
20951674
Resource Type:
Journal Article
Resource Relation:
Journal Name: International Journal of Radiation Oncology, Biology and Physics; Journal Volume: 68; Journal Issue: 2; Other Information: DOI: 10.1016/j.ijrobp.2007.01.038; PII: S0360-3016(07)00199-X; Copyright (c) 2007 Elsevier Science B.V., Amsterdam, Netherlands, All rights reserved; Country of input: International Atomic Energy Agency (IAEA)
Country of Publication:
United States
Language:
English
Subject:
62 RADIOLOGY AND NUCLEAR MEDICINE; COMPUTERIZED TOMOGRAPHY; HEAD; IMAGES; MANUALS; NECK; PATIENTS; RADIOTHERAPY

Citation Formats

Zhang Tiezhi, Chi Yuwei, Meldolesi, Elisa, and Yan Di. Automatic Delineation of On-Line Head-And-Neck Computed Tomography Images: Toward On-Line Adaptive Radiotherapy. United States: N. p., 2007. Web. doi:10.1016/j.ijrobp.2007.01.038.
Zhang Tiezhi, Chi Yuwei, Meldolesi, Elisa, & Yan Di. Automatic Delineation of On-Line Head-And-Neck Computed Tomography Images: Toward On-Line Adaptive Radiotherapy. United States. doi:10.1016/j.ijrobp.2007.01.038.
Zhang Tiezhi, Chi Yuwei, Meldolesi, Elisa, and Yan Di. Fri . "Automatic Delineation of On-Line Head-And-Neck Computed Tomography Images: Toward On-Line Adaptive Radiotherapy". United States. doi:10.1016/j.ijrobp.2007.01.038.
@article{osti_20951674,
title = {Automatic Delineation of On-Line Head-And-Neck Computed Tomography Images: Toward On-Line Adaptive Radiotherapy},
author = {Zhang Tiezhi and Chi Yuwei and Meldolesi, Elisa and Yan Di},
abstractNote = {Purpose: To develop and validate a fully automatic region-of-interest (ROI) delineation method for on-line adaptive radiotherapy. Methods and Materials: On-line adaptive radiotherapy requires a robust and automatic image segmentation method to delineate ROIs in on-line volumetric images. We have implemented an atlas-based image segmentation method to automatically delineate ROIs of head-and-neck helical computed tomography images. A total of 32 daily computed tomography images from 7 head-and-neck patients were delineated using this automatic image segmentation method. Manually drawn contours on the daily images were used as references in the evaluation of automatically delineated ROIs. Two methods were used in quantitative validation: (1) the dice similarity coefficient index, which indicates the overlapping ratio between the manually and automatically delineated ROIs; and (2) the distance transformation, which yields the distances between the manually and automatically delineated ROI surfaces. Results: Automatic segmentation showed agreement with manual contouring. For most ROIs, the dice similarity coefficient indexes were approximately 0.8. Similarly, the distance transformation evaluation results showed that the distances between the manually and automatically delineated ROI surfaces were mostly within 3 mm. The distances between two surfaces had a mean of 1 mm and standard deviation of <2 mm in most ROIs. Conclusion: With atlas-based image segmentation, it is feasible to automatically delineate ROIs on the head-and-neck helical computed tomography images in on-line adaptive treatments.},
doi = {10.1016/j.ijrobp.2007.01.038},
journal = {International Journal of Radiation Oncology, Biology and Physics},
number = 2,
volume = 68,
place = {United States},
year = {Fri Jun 01 00:00:00 EDT 2007},
month = {Fri Jun 01 00:00:00 EDT 2007}
}
  • Purpose: To assess the positioning accuracy using two-dimensional kilovoltage (2DkV) imaging and three-dimensional cone beam CT (CBCT) in patients with head and neck (H and N) cancer receiving radiation therapy. To assess the benefit of patient-specific headrest. Materials and Methods: All 21 patients studied were immobilized using thermoplastic masks with either a patient-specific vacuum bag (11 of 21, IMA) or standard clear plastic (10 of 21, IMB) headrests. Each patient was imaged with a pair of orthogonal 2DkV images in treatment position using onboard imaging before the CBCT procedure. The 2DkV and CBCT images were acquired weekly during the samemore » session. The 2DkV images were reviewed by oncologists and also analyzed by a software tool based on mutual information (MI). Results: Ninety-eight pairs of assessable 2DkV-CBCT alignment sets were obtained. Systematic and random errors were <1.6 mm for both 2DkV and CBCT alignments. When we compared shifts determined by CBCT and 2DkV for the same patient setup, statistically significant correlations were observed in all three major directions. Among all CBCT couch shifts, 4.1% {>=} 0.5 cm and 18.7% {>=} 0.3 cm, whereas among all 2DkV (MI) shifts, 1.7% {>=} 0.5 cm and 11.2% {>=} 0.3 cm. Statistically significant difference was found on anteroposterior direction between IMA and IMB with the CBCT alignment only. Conclusions: The differences between 2D and 3D alignments were mainly caused by the relative flexibility of certain H and N structures and possibly by rotation. Better immobilization of the flexible neck is required to further reduce the setup errors for H and N patients receiving radiotherapy.« less
  • Purpose: To study anatomic biologic contouring (ABC), using a previously described distinct halo, to unify volume contouring methods in treatment planning for head and neck cancers. Methods and Materials: Twenty-five patients with head and neck cancer at various sites were planned for radiation therapy using positron emission tomography/computed tomography (PET/CT). The ABC halo was used in all PET/CT scans to contour the gross tumor volume (GTV) edge. The CT-based GTV (GTV-CT) and PET/CT-based GTV (GTV-ABC) were contoured by two independent radiation oncologists. Results: The ABC halo was observed in all patients studied. The halo had a standard unit value ofmore » 2.19 {+-} 0.28. The mean halo thickness was 2.02 {+-} 0.21 mm. Significant volume modification ({>=}25%) was seen in 17 of 25 patients (68%) after implementation of GTV-ABC. Concordance among observers was increased with the use of the halo as a guide for GTV determination: 6 patients (24%) had a {<=}10% volume discrepancy with CT alone, compared with 22 (88%) with PET/CT (p < 0.001). Interobserver variability decreased from a mean GTV difference of 20.3 cm{sup 3} in CT-based planning to 7.2 cm{sup 3} in PET/CT-based planning (p < 0.001). Conclusions: Using the 'anatomic biologic halo' to contour GTV in PET/CT improves consistency among observers. The distinctive appearance of the described halo and its presence in all of the studied tumors make it attractive for GTV contouring in head and neck tumors. Additional studies are needed to confirm the correlation of the halo with presence of malignant cells.« less
  • Purpose: To propose an automatic atlas-based segmentation framework of the dental structures, called Dentalmaps, and to assess its accuracy and relevance to guide dental care in the context of intensity-modulated radiotherapy. Methods and Materials: A multi-atlas-based segmentation, less sensitive to artifacts than previously published head-and-neck segmentation methods, was used. The manual segmentations of a 21-patient database were first deformed onto the query using nonlinear registrations with the training images and then fused to estimate the consensus segmentation of the query. Results: The framework was evaluated with a leave-one-out protocol. The maximum doses estimated using manual contours were considered as groundmore » truth and compared with the maximum doses estimated using automatic contours. The dose estimation error was within 2-Gy accuracy in 75% of cases (with a median of 0.9 Gy), whereas it was within 2-Gy accuracy in 30% of cases only with the visual estimation method without any contour, which is the routine practice procedure. Conclusions: Dose estimates using this framework were more accurate than visual estimates without dental contour. Dentalmaps represents a useful documentation and communication tool between radiation oncologists and dentists in routine practice. Prospective multicenter assessment is underway on patients extrinsic to the database.« less
  • Purpose: To develop an automatic segmentation algorithm integrating imaging information from computed tomography (CT), positron emission tomography (PET), and magnetic resonance imaging (MRI) to delineate target volume in head and neck cancer radiotherapy. Methods: Eleven patients with unresectable disease at the tonsil or base of tongue who underwent MRI, CT, and PET/CT within two months before the start of radiotherapy or chemoradiotherapy were recruited for the study. For each patient, PET/CT and T1-weighted contrast MRI scans were first registered to the planning CT using deformable and rigid registration, respectively, to resample the PET and magnetic resonance (MR) images to themore » planning CT space. A binary mask was manually defined to identify the tumor area. The resampled PET and MR images, the planning CT image, and the binary mask were fed into the automatic segmentation algorithm for target delineation. The algorithm was based on a multichannel Gaussian mixture model and solved using an expectation–maximization algorithm with Markov random fields. To evaluate the algorithm, we compared the multichannel autosegmentation with an autosegmentation method using only PET images. The physician-defined gross tumor volume (GTV) was used as the “ground truth” for quantitative evaluation. Results: The median multichannel segmented GTV of the primary tumor was 15.7 cm{sup 3} (range, 6.6–44.3 cm{sup 3}), while the PET segmented GTV was 10.2 cm{sup 3} (range, 2.8–45.1 cm{sup 3}). The median physician-defined GTV was 22.1 cm{sup 3} (range, 4.2–38.4 cm{sup 3}). The median difference between the multichannel segmented and physician-defined GTVs was −10.7%, not showing a statistically significant difference (p-value = 0.43). However, the median difference between the PET segmented and physician-defined GTVs was −19.2%, showing a statistically significant difference (p-value =0.0037). The median Dice similarity coefficient between the multichannel segmented and physician-defined GTVs was 0.75 (range, 0.55–0.84), and the median sensitivity and positive predictive value between them were 0.76 and 0.81, respectively. Conclusions: The authors developed an automated multimodality segmentation algorithm for tumor volume delineation and validated this algorithm for head and neck cancer radiotherapy. The multichannel segmented GTV agreed well with the physician-defined GTV. The authors expect that their algorithm will improve the accuracy and consistency in target definition for radiotherapy.« less
  • Purpose: To quantify the uncertainties associated with incorporating diagnostic positron emission tomography/CT (PET/CT) and PET into the radiotherapy treatment-planning process using different image registration tools, including automated and manual rigid body registration methods, as well as deformable image registration. Methods and Materials: The PET/CTs and treatment-planning CTs from 12 patients were used to evaluate image registration accuracy. The PET/CTs also were used without the contemporaneously acquired CTs to evaluate the registration accuracy of stand-alone PET. Registration accuracy for relevant normal structures was quantified using an overlap index and differences in the center of mass (COM) positions. For tumor volumes, themore » registration accuracy was measured using COM positions only. Results: Registration accuracy was better with PET/CT than with PET alone. The COM displacements ranged from 3.2 {+-} 0.6 mm (mean {+-} 95% confidence interval, for brain) to 8.4 {+-} 2.6 mm (spinal cord) for registration with PET/CT data, compared with 4.8 {+-} 1.7 mm (brain) and 9.9 {+-} 3.1 mm (spinal cord) with PET alone. Deformable registration improved accuracy, with minimum and maximum errors of 1.1 {+-} 0.8 mm (brain) and 5.4 {+-} 1.4 mm (mandible), respectively. Conclusions: It is possible to incorporate PET and/or PET/CT acquired in diagnostic positions into the treatment-planning process through the use of advanced image registration algorithms, but precautions must be taken, particularly when delineating tumor volumes in the neck. Acquisition of PET/CT in the treatment-planning position would be the ideal method to minimize registration errors.« less