Volume Overlap Index Research Articles

Influence of MRI-CT fusion image on gross tumor volume delineation and dose distribution for radiation of primary tumors of cervical vertebra

Objective To compare the differences of gross tumor volume (GTV) and dose distribution between MRI-CT fusion imaging and CT-based imaging and investigate the dose difference in the therapeutic regime. Methods Ten patients diagnosed with primary tumors of the cervical vertebra between 2013 and 2014 were enrolled. Prior to radiotherapy, the imaging data of MRI examination (GE Discovery MR 750 3.0T) were collected, transfered into the Eclipse system and subject to fusion with CT images. GTV delineation, organ at risk (OAR) delineation and dose distribution discrepancy induced by target volume differences were analyzed and statistically compared between the MRI-CT fusion and CT-based images. GTV delineation and dose parameters among different radiologists between two approches were analyzed by analysis of variance (ANOVA) and paired t-test. Non-normally distributed variables were analyzed by Wilcoxon rank-sum test. The reliability of intraclass correlation coefficient (ICC) was assessed. Results The GTVMRI-CT volume was larger compared with the GTVCT volume. The volume overlap index was 0.84±0.17.The cordMRI-CT volume was significantly less than the cordCT volume (P=0.001). For 5 radiologists, the ICCMRI-CT was larger than ICCCT. The DmaxMRI-CT of the spinal cord was (46.00±1.09) Gy, significantly less than (52.39±1.34) Gy for the DmaxCT(P=0.014). Conclusions It is unlikely to miss the target delineation on MRI-CT fusion imaging. MRI-CT fusion imaging can minimize the discrepancy of interobserver radiologists and cause dosimetric advantages. Key words: Tomotherapy, X-ray computed; Magnetic resonance imaging; Gross tumor volume; Cervical neoplasm/intensity-modulated radiotherapy

Comparison of Magnetic Resonance Imaging and Computed Tomography for Breast Target Volume Delineation in Prone and Supine Positions

Todetermine whether T2-weighted MRI improves seroma cavity (SC) and whole breast (WB) interobserver conformity for radiation therapy purposes, compared with the gold standard of CT, both in the prone and supine positions. Eleven observers (2 radiologists and 9 radiation oncologists) delineated SC and WB clinical target volumes (CTVs) on T2-weighted MRI and CT supine and prone scans (4 scans per patient) for 33 patient datasets. Individual observer's volumes were compared using the Dice similarity coefficient, volume overlap index, center of mass shift, and Hausdorff distances. An average cavity visualization score was also determined. Imaging modality did not affect interobserver variation for WB CTVs. Prone WB CTVs were larger in volume and more conformal than supine CTVs (on both MRI and CT). Seroma cavity volumes were larger on CT than on MRI. Seroma cavity volumes proved to be comparable in interobserver conformity in both modalities (volume overlap index of 0.57(95% Confidence Interval (CI) 0.54-0.60) for CT supine and 0.52(95% CI 0.48-0.56) for MRI supine, 0.56(95% CI 0.53-0.59) for CT prone and 0.55(95% CI 0.51-0.59) for MRI prone); however, after registering modalities together the intermodality variation (Dice similarity coefficient of 0.41(95% CI 0.36-0.46) for supine and 0.38(0.34-0.42) for prone) was larger than the interobserver variability for SC, despite the location typically remaining constant. Magnetic resonance imaging interobserver variation was comparable to CT for the WB CTV and SC delineation, in both prone and supine positions. Although the cavity visualization score and interobserver concordance was not significantly higher for MRI than for CT, the SCs were smaller on MRI, potentially owing to clearer SC definition, especially on T2-weighted MR images.

Automated planning of breast radiotherapy using cone beam CT imaging.

Develop and clinically validate a methodology for using cone beam computed tomography (CBCT) imaging in an automated treatment planning framework for breast IMRT. A technique for intensity correction of CBCT images was developed and evaluated. The technique is based on histogram matching of CBCT image sets, using information from "similar" planning CT image sets from a database of paired CBCT and CT image sets (n = 38). Automated treatment plans were generated for a testing subset (n = 15) on the planning CT and the corrected CBCT. The plans generated on the corrected CBCT were compared to the CT-based plans in terms of beam parameters, dosimetric indices, and dose distributions. The corrected CBCT images showed considerable similarity to their corresponding planning CTs (average mutual information 1.0±0.1, average sum of absolute differences 185 ± 38). The automated CBCT-based plans were clinically acceptable, as well as equivalent to the CT-based plans with average gantry angle difference of 0.99°±1.1°, target volume overlap index (Dice) of 0.89±0.04 although with slightly higher maximum target doses (4482±90 vs 4560±84, P < 0.05). Gamma index analysis (3%, 3 mm) showed that the CBCT-based plans had the same dose distribution as plans calculated with the same beams on the registered planning CTs (average gamma index 0.12±0.04, gamma <1 in 99.4%±0.3%). The proposed method demonstrates the potential for a clinically feasible and efficient online adaptive breast IMRT planning method based on CBCT imaging, integrating automation.

SU‐E‐J‐78: Internal Target Volume Delineation for Lung Tumors in Patients Treated with Robotic Radiosurgery

Purpose:To compare different approaches for Internal Target Volume (ITV) delineation for patients treated with fiducial‐free robotic radiosurgery for primary and metastatic lung tumors.Methods:Ten patients undergoing Lung‐Optimized Treatment (LOT) for robotic radiosurgery were imaged with inhale and exhale breath‐hold CT scans and 8‐phase 4DCT scan. We evaluated the differences in internal target volume (ITV) delineated using three approaches: 1) maximum intensity projection (MIP) images reconstructed from 4DCT scan (ITV‐MIP); 2) linear interpolation of Gross Tumor Volumes (GTV) segmented on inhale and exhale breath‐hold scans (ITV‐BH); 3) linear interpolation of GTV segmented on inhale and exhale phases of 4DCT scan (ITV‐2Phase). All contours were independently generated by the same radiation oncologist using lung window settings. Patients had ITV‐MIP volumes ranging from 1.5 to 146.9 cc (mean 36.8 cc) located in various parts of the lung. Volume overlap and matching index (MI) were calculated and compared. The MI between two volumes was defined as the ratio of their intersection to their union. MI of 1 indicates the volumes are identical; MI of 0 indicates that there is no overlap.Results:The three approaches generated very different results. The average (SD) MI for ITV‐MIP and ITV‐BH was 0.52 (0.24); for ITV‐MIP and ITV‐2Phase it was 0.69 (0.13); and for ITV‐BH and ITV‐2Phase was 0.57 (0.21), (ANOVA, p=0.16). Relative to the ITV‐MIP, the percentage of volume overlap was 72% (26%) and 90% (7%) for ITV‐BH and ITV‐2Phase, respectively (t‐test, p=0.05).Conclusion:Differences between ITV‐BH and ITV‐MIP are due to inconsistent lung filling at breath‐hold and nonlinear tumor motion. Therefore, methods to check breath‐hold scanning against regular patient breathing patterns should be developed. Whenever possible, ITV‐BH generated by the LOT workflow should be verified by 4DCT data.

Evaluation of two synchronized external surrogates for 4D CT sorting

The purpose of this study was to quantify the performance and agreement between two different external surrogate acquisition systems: Varian's Real‐Time Position Management (RPM) and Philips Medical Systems' pneumatic bellows, in the context of waveform and 4D CT image analysis. Eight patient displacement curves derived from RPM data were inputted into a motion platform with varying amplitudes (0.5 to 3 cm) and patterns. Simultaneous 4D CT acquisition, with synchronized X‐ray on detection, was performed with the bellows and RPM block placed on the platform. Bellows data were used for online retrospective phase‐based sorting, while RPM data were used for off‐line reconstruction of raw 4D CT data. RPM and bellows breathing curves were resampled, normalized, and analyzed to determine associations between different external surrogates, relative amplitude differences, and system latency. Maximum intensity projection (MIP) images were generated, phantom targets were delineated, and volume differences, overlap index, and Dice similarity coefficient differences were evaluated. A prospective patient study of ten patients was performed and waveforms were evaluated for latency (i.e., absolute time differences) and overall agreement. 4D CT sorting quality and subtraction images were assessed. Near perfect associations between the RPM and bellows‐acquired breathing traces were found (Pearson′sr=0.987−0.999). Target volumes were 200.4±12ccand199.8±12.6cc for RPM and bellows targets, respectively, which was not significantly different (U=33,p>0.05). Negligible centroid variations were observed between bellows and RPM‐contoured MIP targets (largest discrepancy=−0.24±0.31mm in superior‐inferior direction). The maximum volume difference was observed for an RPM target 2.5 cc (1%) less than bellows, yielding the largest difference in centroid displacement (0.9 mm). Strong correlations in bellows and RPM waveforms were observed for all patients (0.947±0.037). Latency between external surrogates was <100ms for phantom and patient data. Negligible differences were observed between MIP, end‐exhale, and end‐inhale phase images for all cases, with delineated RPM and bellows lung volumes demonstrating a mean difference of −0.3±0.51%. Dice similarity coefficients and overlap indices were near unity for phantom target volumes and patient lung volumes. Slight differences were observed in waveform and latency analysis between Philips bellows and Varian's RPM, although these did not translate to differences in image quality or impact delineations. Therefore, the two systems were found to be equivalent external surrogates in the context of 4D CT for treatment planning purposes.PACS numbers: 87.57.Q‐, 07.07.Df

Preliminary experience with a novel method of three-dimensional co-registration of prostate cancer digital histology and in vivo multiparametric MRI

To assess a novel method of three-dimensional (3D) co-registration of prostate cancer digital histology and in-vivo multiparametric magnetic resonance imaging (mpMRI) image sets for clinical usefulness. A software platform was developed to achieve 3D co-registration. This software was prospectively applied to three patients who underwent radical prostatectomy. Data comprised in-vivo mpMRI [T2-weighted, dynamic contrast-enhanced weighted images (DCE); apparent diffusion coefficient (ADC)], ex-vivo T2-weighted imaging, 3D-rebuilt pathological specimen, and digital histology. Internal landmarks from zonal anatomy served as reference points for assessing co-registration accuracy and precision. Applying a method of deformable transformation based on 22 internal landmarks, a 1.6 mm accuracy was reached to align T2-weighted images and the 3D-rebuilt pathological specimen, an improvement over rigid transformation of 32% (p = 0.003). The 22 zonal anatomy landmarks were more accurately mapped using deformable transformation than rigid transformation (p = 0.0008). An automatic method based on mutual information, enabled automation of the process and to include perfusion and diffusion MRI images. Evaluation of co-registration accuracy using the volume overlap index (Dice index) met clinically relevant requirements, ranging from 0.81-0.96 for sequences tested. Ex-vivo images of the specimen did not significantly improve co-registration accuracy. This preliminary analysis suggests that deformable transformation based on zonal anatomy landmarks is accurate in the co-registration of mpMRI and histology. Including diffusion and perfusion sequences in the same 3D space as histology is essential further clinical information. The ability to localize cancer in 3D space may improve targeting for image-guided biopsy, focal therapy, and disease quantification in surveillance protocols.

TU‐C‐141‐05: A Variable‐Kernel Smoothing Technique for Improved Convergence During CT‐Cone Beam CT Deformable Image Registration

Purpose: To develop a fast and accurate technique for deformable image registration (DIR) of cone beam CT (CBCT) and CT images for use in online adaptive radiotherapy (ART). Methods: A novel technique for smoothing a deformation field in which the size of the Gaussian kernel is adaptively varied based on the rate of convergence is proposed and implemented for DIR based on the demons algorithm. The planning CT and daily CBCT acquired for 6 prostate cancer patients were registered using the newly developed technique. Histogram matching was used to compensate for intensity differences between the two modalities. The Pearson correlation coefficient (PCC) and volume overlap index (VOI) were used to quantify registration accuracy. Results: The iterative decrease of the smoothing kernel size allowed the algorithm to converge to an increasingly accurate solution, beyond the asymptotic limit for a constant kernel size. The mean VOI was calculated for bladder, prostate, and rectum with values of 91.9%, 68.7%, and 78.2%, respectively. The correlation coefficient was calculated for every fraction in the overlapping CT and CBCT scan volumes in each patient data set. Average PCC values were 0.9987, 0.9985, 0.9982, 0.9980, 0.9985, and 0.9985 for the six patients. A typical run time for a 512x512x70 image volume was 4.6 minutes. Conclusion: The presently developed DIR technique allowed for registration accuracies comparable to intra‐modality methods and validates its use for deformable CT‐CBCT registration. The method eliminates the need for multi‐resolution processing and continual upsampling of the image which can be computationally intensive. The combined speed and accuracy of this technique make it a promising tool for online ART.

Auto-segmentation of low-risk clinical target volume for head and neck radiation therapy

PurposeTo investigate atlas-based auto-segmentation methods to improve the quality of the delineation of low-risk clinical target volumes (CTVs) of unilateral tonsil cancers. Method and MaterialsSixteen patients received intensity modulated radiation therapy for left tonsil tumors. These patients were treated by a total of 8 oncologists, who delineated all contours manually on the planning CT image. We chose 6 of the patients as atlas cases and used atlas-based auto-segmentation to map each the atlas CTV to the other 10 patients (test patients). For each test patient, the final contour was produced by combining the 6 individual segmentations from the atlases using the simultaneous truth and performance level estimation algorithm. In addition, for each test patient, we identified a single atlas that produced deformed contours best matching the physician's manual contours. The auto-segmented contours were compared with the physician's manual contours using the slice-wise Hausdorff distance (HD), the slice-wise Dice similarity coefficient (DSC), and a total volume overlap index. ResultsNo single atlas consistently produced good results for all 10 test cases. The multiatlas segmentation achieved a good agreement between auto-segmented contours and manual contours, with a median slice-wise HD of 7.4 ± 1.0 mm, median slice-wise DSC of 80.2% ± 5.9%, and total volume overlap of 77.8% ± 3.3% over the 10 test cases. For radiation oncologists who contoured both the test case and one of the atlas cases, the best atlas for a test case had almost always been contoured by the oncologist who had contoured that test case, indicating that individual physician's practice dominated in target delineation and was an important factor in optimal atlas selection. ConclusionsMultiatlas segmentation may improve the quality of CTV delineation in clinical practice for unilateral tonsil cancers. We also showed that individual physician's practice was an important factor in selecting the optimal atlas for atlas-based auto-segmentation.

Comparative methods for PET image segmentation in pharyngolaryngeal squamous cell carcinoma

PurposeSeveral methods have been proposed for the segmentation of 18F-FDG uptake in PET. In this study, we assessed the performance of four categories of 18F-FDG PET image segmentation techniques in pharyngolaryngeal squamous cell carcinoma using clinical studies where the surgical specimen served as the benchmark.MethodsNine PET image segmentation techniques were compared including: five thresholding methods; the level set technique (active contour); the stochastic expectation-maximization approach; fuzzy clustering-based segmentation (FCM); and a variant of FCM, the spatial wavelet-based algorithm (FCM-SW) which incorporates spatial information during the segmentation process, thus allowing the handling of uptake in heterogeneous lesions. These algorithms were evaluated using clinical studies in which the segmentation results were compared to the 3-D biological tumour volume (BTV) defined by histology in PET images of seven patients with T3–T4 laryngeal squamous cell carcinoma who underwent a total laryngectomy. The macroscopic tumour specimens were collected “en bloc”, frozen and cut into 1.7- to 2-mm thick slices, then digitized for use as reference.ResultsThe clinical results suggested that four of the thresholding methods and expectation-maximization overestimated the average tumour volume, while a contrast-oriented thresholding method, the level set technique and the FCM-SW algorithm underestimated it, with the FCM-SW algorithm providing relatively the highest accuracy in terms of volume determination (−5.9 ± 11.9%) and overlap index. The mean overlap index varied between 0.27 and 0.54 for the different image segmentation techniques. The FCM-SW segmentation technique showed the best compromise in terms of 3-D overlap index and statistical analysis results with values of 0.54 (0.26–0.72) for the overlap index.ConclusionThe BTVs delineated using the FCM-SW segmentation technique were seemingly the most accurate and approximated closely the 3-D BTVs defined using the surgical specimens. Adaptive thresholding techniques need to be calibrated for each PET scanner and acquisition/processing protocol, and should not be used without optimization.

SU‐E‐J‐99: Multi‐Atlas Based Auto‐Segmentation of Low‐Risk Clinical Target Volume (CTV) for Head‐and‐Neck Radiotherapy

Purpose: To develop an auto‐segmentation approach using multiple atlases to improve the quality of target delineation for low‐risk CTV of unilateral tonsil cancers. Methods: Sixteen unilateral tonsil cancer patients received IMRT treatment for their left tonsil tumors. These patients were treated by a total of 8 oncologists, who delineated all contours manually on the planning CT. We chose 6 patients as atlases, and used deformable image registration to map each atlas CTV to the rest of 10 test patients. In each test patient, the final contour was produced by fusing 6 individually deformed contours using the Simultaneous Truth and Performance Level Estimation (STAPLE) algorithm. In addition, we also identified the best single atlas which best matched physicianˈs contour on each test case. The auto‐segmented contours were evaluated by comparing against the physician delineated contours using slice‐wise Hausdorff distance (HD) and Dice coefficient (DSC), and a total volume overlap index. Results: Evaluation results showed that not a single atlas could produce reasonably good results in all 10 testing cases. The multi‐atlas method achieved much better agreement than the single best atlas method. The multi‐atlas method produced the median slice‐wise HD of 7.35mm, the median slice‐wise DSC of 80.2%, and the total volume overlap of 77.8%. For radiation oncologists who contoured both the test case and one of the atlas cases, the best atlas almost always came from the same radiation oncologist, indicating that inter‐observer variation is dominated in target delineation. Our multi‐atlas result confirms that the use of multi‐atlas and STAPLE algorithm automatically reduces the inter‐observer variability. Conclusions: We developed and validated a multi‐atlas based auto‐segmentation method to improve the quality and usability of low‐risk target volume delineation for unilateral tonsil cancer radiotherapy.

Deformable planning CT to cone‐beam CT image registration in head‐and‐neck cancer

The purpose of this work was to implement and validate a deformable CT to cone-beam computed tomography (CBCT) image registration method in head-and-neck cancer to eventually facilitate automatic target delineation on CBCT. Twelve head-and-neck cancer patients underwent a planning CT and weekly CBCT during the 5-7 week treatment period. The 12 planning CT images (moving images) of these patients were registered to their weekly CBCT images (fixed images) via the symmetric force Demons algorithm and using a multiresolution scheme. Histogram matching was used to compensate for the intensity difference between the two types of images. Using nine known anatomic points as registration targets, the accuracy of the registration was evaluated using the target registration error (TRE). In addition, region-of-interest (ROI) contours drawn on the planning CT were morphed to the CBCT images and the volume overlap index (VOI) between registered contours and manually delineated contours was evaluated. The mean TRE value of the nine target points was less than 3.0 mm, the slice thickness of the planning CT. Of the 369 target points evaluated for registration accuracy, the average TRE value was 2.6 +/- 0.6 mm. The mean TRE for bony tissue targets was 2.4 +/- 0.2 mm, while the mean TRE for soft tissue targets was 2.8 +/- 0.2 mm. The average VOI between the registered and manually delineated ROI contours was 76.2 +/- 4.6%, which is consistent with that reported in previous studies. The authors have implemented and validated a deformable image registration method to register planning CT images to weekly CBCT images in head-and-neck cancer cases. The accuracy of the TRE values suggests that they can be used as a promising tool for automatic target delineation on CBCT.

SU‐GG‐J‐47: Verification of a Deformable Image Registration Algorithm for Head and Neck Cancer Therapy

Introduction: H&amp;N cancer patients may require frequent CT imaging for re‐planning due to significant anatomical changes to improve dosimetric accuracy. A challenging task is manual delineation of target(s) and normal structures on each image sets acquired on different days. The aim of this study was to verify a Small Deformation Inverse Consistent Linear Elastic (SICLE) deformable image registration for its ability to deformably‐map contours on H&amp;N image sets by comparing them to physician‐drawn contours. Methods and Materials: Six patients with H&amp;N cancer who had frequent CT imaging were selected. GTVs, parotids, cord and brainstem were physican‐drawn on both source and target image sets. Physician‐drawn contours on source image set were deformed by the DVFs to produce deformably‐mapped contours and dose on target images. Differences between the physician‐drawn and deformably‐mapped contours were compared using: “Differences in volumes(Vdiff)”, “differences in center of mass(COMdiff)”, “volume overlap index(VOI)”, “3D‐perpendicular‐distance‐maps”. To evaluate the clinical usefulness, plans were generated based on the physician‐drawn and deformably‐mapped contours. Furthermore, the dose from source image was deformed to target images using the DVFs. Plans were compared for their coverage of physician‐drawn contours. Results: While GTV and NodalGTV VOIs ranged from 0.57 to 0.72 and 0.57 to 0.79 respectively, for parotids, brainstem and cord, it varied from 0.42 to 0.82. Up to 1.7cm and 2.9cm discrepancies were observed in esophagus and cord COM in z direction. Visual comparison of physician‐drawn and deformably‐mapped GTVcontours showed a reasonably well agreement. Dosimetric impact of contour differences was not very significant with &lt;5% differences in PTV D95. Conclusion: SICLE algorithm can produce high quality contours on subsequent H&amp;N image sets. However, manual modification of deformably‐mapped contours may be necessary. Deformably‐mapped contours can be used as a QA tool for inspecting manually‐drawn contours, in addition to reducing or even eliminating the manual contouring workload.

Inter-observer and intra-observer reliability for lung cancer target volume delineation in the 4D-CT era

Background and purpose To investigate inter-observer and intra-observer target volume delineation (TVD) error in 4D-CT imaging of thoracic tumours. Materials and methods Primary and nodal gross tumour volumes (GTV) of 10 lung tumours on the 10 respiratory phases of a 4D-CT scan were contoured by six radiation oncologist observers. Inter-observer and intra-observer variability were assessed by the coefficient of variation (COV) and the volume overlap index (VOI). ANOVA was performed to assess differences in inter-observer and intra-observer variability based on patient case difficulty, respiratory phase, physician seniority, and physician observer. Results VOI analysis determined that inter-observer was a more significant source of error than intra-observer variability. VOI improved with the use of 4D-CT as compared to conventional CT. ANOVA analysis for COVs found case difficulty (easy versus difficult) to be significant for inter-observer primary tumour and intra-observer nodal disease delineation. Physician seniority, respiratory phase, and individual physician were not found to be significant for TVD error. Conclusion Variability in TVD is a major source of error in 4D-CT treatment planning. Development of measures to reduce inter-observer and intra-observer TVD variability are necessary in order to deliver high quality radiotherapy.

The value of magnetic resonance imaging in target volume delineation of base of tongue tumours – A study using flexible surface coils

Introduction Magnetic resonance imaging (MRI) provides superior diagnostic accuracy over computed tomography (CT) in oropharyngeal tumours. Precise delineation of the gross tumour volume (GTV) is mandatory in radiotherapy planning when a GTV boost is required. CT volume definition in this regard is poor. We studied the feasibility of using flexible surface (flex-L) coils to obtain MR images for MR–CT fusion to assess the benefit of MRI over CT alone in planning base of tongue tumours. Methods Eight patients underwent CT and MRI radiotherapy planning scans with an immobilisation device. Distortion-corrected T1-weighted post-contrast MR scans were fused to contrast-enhanced planning CT scans. GTV, clinical target and planning target volumes (CTV, PTV) and organs at risk (OAR) were delineated on CT, then on MRI with blinding to the CT images. The volumetric and spatial differences between MRI and CT volumes for GTV, CTV, PTV and OAR were compared. MR image distortions due to field inhomogeneity and non-linear gradients were corrected and the need for such correction was evaluated. Results The mean primary GTV was larger on MRI (22.2 vs. 9.5 cm 3, p = 0.05) than CT. The mean primary and nodal GTV (i.e. BOT and macroscopic nodes) was significantly larger on MRI (27.2 vs. 14.4 cm 3, p = 0.05). The volume overlap index (VOI) between MRI and CT for the primary was 0.34 suggesting that MRI depicts parts of the primary tumour not detected by CT. There was no significant difference in volume delineation between MR and CT for CTV, PTV, nodal CTV and nodal PTV. MRI volumes for brainstem and spinal cord were significantly smaller due to improved organ definition ( p = 0.002). Susceptibility and gradient-related distortions were not found to be clinically significant. Conclusion MRI improves the definition of tongue base tumours and neurological structures. The use of MRI is recommended for GTV dose-escalation techniques to provide precise depiction of GTV and improved sparing of spinal cord and brainstem.

Inter-observer and Intra-observer Reliability for Lung Cancer Target Volume Delineation in the 4D-CT Era

Abstract Background and purpose To investigate inter-observer and intra-observer target volume delineation (TVD) error in 4D-CT imaging of thoracic tumours. Materials and methods Primary and nodal gross tumour volumes (GTV) of 10 lung tumours on the 10 respiratory phases of a 4D-CT scan were contoured by six radiation oncologist observers. Inter-observer and intra-observer variability were assessed by the coefficient of variation (COV) and the volume overlap index (VOI). ANOVA was performed to assess differences in inter-observer and intra-observer variability based on patient case difficulty, respiratory phase, physician seniority, and physician observer. Results VOI analysis determined that inter-observer was a more significant source of error than intra-observer variability. VOI improved with the use of 4D-CT as compared to conventional CT. ANOVA analysis for COVs found case difficulty (easy versus difficult) to be significant for inter-observer primary tumour and intra-observer nodal disease delineation. Physician seniority, respiratory phase, and individual physician were not found to be significant for TVD error. Conclusion Variability in TVD is a major source of error in 4D-CT treatment planning. Development of measures to reduce inter-observer and intra-observer TVD variability are necessary in order to deliver high quality radiotherapy.

MO‐FF‐A3‐03: Quantitative Assessment of Automatic Anatomy Segmentation of 4D CT Images Using a Deformable Image Registration Algorithm

Purpose: Quantify the accuracy of Small Deformation Inverse Consistent Linear Elastic (SICLE) deformable image registration to auto‐segment contours on 4D thoracic CT images. Method and Materials: A process of automatic contouring based on SICLE deformable image registration of respiratory CT phases from 4DCT scans was integrated into our image‐guided‐radiotherapy‐computing‐framework. The gross tumor volumes (GTVs), right and left lungs, cord and esophagus were manually‐drawn by the same physician for all ten respiratory phases. Registrations were performed using three intensity windows: 0–900 (lung only), 900–1300 (remaining tissue excluding lung) and 0–1300 (whole image). For each respiratory phase, the differences between the manually‐drawn and deformably‐mapped contours for each phase were quantified using fractional volume differences (FVDiff), differences in center of mass coordinates (COMdiff) and volume overlap index (VOI) and 3D maps of perpendicular distance between manually‐drawn and deformably‐mapped boundaries. The results were also compared against the previously‐implemented viscous fluid‐based registration. Results: The VOI for GTV and heart varied from 0.85 to 0.90, from 0.9 to 0.97, respectively. As expected, the smallest FVDiff were found be for the lungs (&lt;0.027). The SICLE (0–900) and viscous fluid methods yielded similar GTV COM in superior‐inferior direction. COMdiff were largest for the esophagus (∼0.36 cm) since the esophagus is difficult to visualize on CT, especially in the lower thorax. Overall, visual comparison of GTV manual and auto contours showed good agreement. However, large discrepancies (up to 2cm) were found at the most inferior portion of the lung. This disagreement was also very evident from 3D maps of perpendicular distances between manually‐drawn and deformably‐mapped contours. Conclusion: Automatically‐generated contours using SICLE algorithm can produce high quality contours on subsequent respiratory CT phases. Deformable image registration may reduce or even eliminate the manual contouring workload and can be used for manually‐drawn contour quality assurance.Supported by NIHgrant‐P01CA11602

MO‐D‐BRB‐08: A Hybrid Strategy of Offline Adaptive Planning and Online Image Guidance for Prostate Cancer Radiotherapy

Purpose: In online image‐guided radiotherapy of prostate cancer, the setup error and inter‐fractional motion is eliminated through pre‐treatment imaging and couch correction at each fraction. However, the rotation and deformation is not corrected and only accounted for in planning margin. In this study, we propose a hybrid of online and offline adaptive image‐guidance strategy for both low risk patients (LRP, CTV=prostate) and intermediate risk patients (IRP, CTV=prostate+seminal vesicles). The benefit of margin reduction is evaluated geometrically. Method and Materials: Planning and treatment helical CT images from 25 patients over 412 fractions were used. Online image‐guidance was simulated by matching the center of mass of CTV in treatment CT to planning CT. Offline replanning was performed by constructing the internal target volume (ITV) from the union of position‐corrected CTVs from the first 5 treatment CTs. The volume overlap index (OI) of ITV and CTVs of the remaining fractions were compared with the OIs between PTV from planning CT and the treatment CTVs. Margins from 0 to 10 mm were investigated. Results: The mean ITV volumes are 62.5 and 90.1 cm3 for LRP and IRP, respectively, equivalent to 0.8 and 1.0 mm uniform margin to CTV0 (mean volumes of 55.7 and 78.0 cm3). The margins needed for 99% OI for ITV (V99% = 67.9 and 105.5 cm3) are 1.7 mm and 2.6 mm less than those for the planning CTV (V99% = 84.5 and 144.7 cm3) for LRP and IRP, respectively. Conclusion: The hybrid of online and offline adaptive radiotherapy protocol can effectively account for the patient‐specific interfraction organ motion and setup errors for prostate cancer patients. The planning margin can be reduced further using the hybrid strategy compared with online image guidance alone. Reduced irradiated volume will also lead to decreased toxicity in critical organs.

The Effectiveness of Rectal Balloon and Orthogonal Imaging for Proton Therapy of Prostate Cancer

The Effectiveness of Rectal Balloon and Orthogonal Imaging for Proton Therapy of Prostate Cancer

Performance Evaluation of Automatic Anatomy Segmentation Algorithm on Repeat or Four-Dimensional Computed Tomography Images Using Deformable Image Registration Method

Auto-propagation of anatomic regions of interest from the planning computed tomography (CT) scan to the daily CT is an essential step in image-guided adaptive radiotherapy. The goal of this study was to quantitatively evaluate the performance of the algorithm in typical clinical applications. We had previously adopted an image intensity-based deformable registration algorithm to find the correspondence between two images. In the present study, the regions of interest delineated on the planning CT image were mapped onto daily CT or four-dimensional CT images using the same transformation. Postprocessing methods, such as boundary smoothing and modification, were used to enhance the robustness of the algorithm. Auto-propagated contours for 8 head-and-neck cancer patients with a total of 100 repeat CT scans, 1 prostate patient with 24 repeat CT scans, and 9 lung cancer patients with a total of 90 four-dimensional CT images were evaluated against physician-drawn contours and physician-modified deformed contours using the volume overlap index and mean absolute surface-to-surface distance. The deformed contours were reasonably well matched with the daily anatomy on the repeat CT images. The volume overlap index and mean absolute surface-to-surface distance was 83% and 1.3 mm, respectively, compared with the independently drawn contours. Better agreement (>97% and <0.4 mm) was achieved if the physician was only asked to correct the deformed contours. The algorithm was also robust in the presence of random noise in the image. The deformable algorithm might be an effective method to propagate the planning regions of interest to subsequent CT images of changed anatomy, although a final review by physicians is highly recommended.

Automatic segmentation of phase‐correlated CT scans through nonrigid image registration using geometrically regularized free‐form deformation

Conventional radiotherapy is planned using free-breathing computed tomography (CT), ignoring the motion and deformation of the anatomy from respiration. New breath-hold-synchronized, gated, and four-dimensional (4D) CT acquisition strategies are enabling radiotherapy planning utilizing a set of CT scans belonging to different phases of the breathing cycle. Such 4D treatment planning relies on the availability of tumor and organ contours in all phases. The current practice of manual segmentation is impractical for 4D CT, because it is time consuming and tedious. A viable solution is registration-based segmentation, through which contours provided by an expert for a particular phase are propagated to all other phases while accounting for phase-to-phase motion and anatomical deformation. Deformable image registration is central to this task, and a free-form deformation-based nonrigid image registration algorithm will be presented. Compared with the original algorithm, this version uses novel, computationally simpler geometric constraints to preserve the topology of the dense control-point grid used to represent free-form deformation and prevent tissue fold-over. Using mean squared difference as an image similarity criterion, the inhale phase is registered to the exhale phase of lung CT scans of five patients and of characteristically low-contrast abdominal CT scans of four patients. In addition, using expert contours for the inhale phase, the corresponding contours were automatically generated for the exhale phase. The accuracy of the segmentation (and hence deformable image registration) was judged by comparing automatically segmented contours with expert contours traced directly in the exhale phase scan using three metrics: volume overlap index, root mean square distance, and Hausdorff distance. The accuracy of the segmentation (in terms of radial distance mismatch) was approximately 2 mm in the thorax and 3 mm in the abdomen, which compares favorably to the accuracies reported elsewhere. Unlike most prior work, segmentation of the tumor is also presented. The clinical implementation of 4D treatment planning is critically dependent on automatic segmentation, for which is offered one of the most accurate algorithms yet presented.