Index of content:
Volume 33, Issue 6, June 2006
- Joint Imaging/Therapy Moderated Poster Session: Exhibit Hall F
- Moderated Poster ‐ Area 2 (Joint): Correction Strategies
33(2006); http://dx.doi.org/10.1118/1.2240178View Description Hide Description
Purpose:IMRT delivery could be affected by residual target motion, the intrafractional target uncertainties resulting from motion that follows image‐guided procedures such as dual x‐ray image acquisitions with patient positional corrections. This study investigates the 6D spatial characteristics of residual target motion and a method for minimizing potential IMRT delivery errors. Method and Materials: A recursive dose painting algorithm was developed to reorder MLC segments, so that most segments were delivered repetitively at a fraction of its prescribed MU. The rationale is that there could be residual target motion that would cause significant delivery errors, so to “paint” the target voxels recursively in small dose segments instead of painting the whole area once with a larger dose segment, could produce a more accurate delivery. The residual target motions used were determined from implanted spine fiducials detected using the dual x‐ray images taken every 10–20 seconds during treatment (Cyberknife, Accuray). Six‐degree of freedom spatial characteristics of the residual target uncertainties were extracted. These residual motions were then fed into the recursive dose painting method for dose evaluation. Results: The residual target motion study showed that although the average was less than 3 mm during the whole delivery, there existed large irregularities in distributions in both rotational and translational directions. For most cases, there were also systematic motion errors of 0.5–1 mm and large spurious motions, sometimes of 5 mm or more. When using the actual error‐time curves, the recursive delivery significantly reduced dose variations caused by such errors (by as much as 27%). For the majority of cases studied, a cycle of three to six repetitive dose paintings was found sufficient to achieve such improvements. Conclusion: Residual target motion is irregularly distributed as treatment time progresses. Recursive dose painting provides a solution to decrease the dose errors caused by such uncertainties.
SU‐EE‐A2‐02: Feasibility of Tracking Head Position Under An Obscuring Immobilization Mask Using a Bite Block and a 3‐D Surface Imaging System33(2006); http://dx.doi.org/10.1118/1.2240186View Description Hide Description
Purpose: To assess the accuracy and feasibility of measuring head motion under a thermoplastic immobilization mask when using a 3‐D surface imaging system. Materials and Methods: Small dome‐shaped objects were arranged asymmetrically on a styrofoam platform that was attached to a bite block system. The assembly was mounted on the couch of a linear accelerator to a micromanipulator with slow motion controls. This arrangement is operationally similar to a bite block affixed to the maxilla of a intracranial radiation therapy patient. The micromanipulator allowed for motions with six degrees of freedom. We were able to achieve sub‐millimeter translational adjustments of the bite block assembly and angulation. The platform and objects were “tracked” with the AlignRT® 3‐D surface imaging system (VisionRT, London, UK) in order to compare mechanical translational and rotational movements of known magnitude with the changes reported by the AlignRT system. While translational motions are only reported with millimeter resolution on the computer control screen, we obtained the system records sub‐millimeter from stored data file records. Results: Translational agreement between the micromanipulator and the AlignRT system was 0.1±0.1 mm in all three axes. Rotational agreement was within 0.5 degrees for pitch and roll. Agreement for yaw was not determined, however the display for couch rotation is 0.1 degree and has a stability of ±0.1 degrees. Conclusion: The AlignRT surface imaging system has superior accuracy that is sufficient for stereotactic radiosurgery guidance using a bite block as we have designed the experiment. Combining a bite block similar to what we designed with the AlignRT 3‐D system shows promise in monitoring head position under an occluding immobilization mask.
SU‐EE‐A2‐03: Evaluation of Auto‐Segmentation Tools for the Target Definition for the Treatment of Lung Cancer33(2006); http://dx.doi.org/10.1118/1.2240194View Description Hide Description
Purpose: With the advent of more sophisticated image devices in the treatment room, image guided radiotherapy(IGRT) and adaptive radiotherapy(ART) have become distinct possibilities. IGRT and ART techniques in their various stages have been implemented in clinics. One of the ART techniques using the daily acquired CTimages involves re‐planning due to the target shape variation during the treatment. Lungcancer volumes of some cases are observed to undergo significant changes where re‐planning is a necessity. To be able to define target efficiently can help the treatment flow significantly. This study evaluates various auto or semi‐auto contouring tools either commercially available or under development for their accuracy and ease of use. Method and Materials: Three methods are included in the study. Two are commercially available (Focal,CMS): auto threshold (of gray level); and auto Segmentation where gray level, the edges and prior shape information are used. The third method is the ITK‐SNAP program that uses a powerful level set(snake) segmentation algorithm to segment anatomical structures in three dimensions. Results: Ten image sets from helical and cone beam CTs are included in the study. The acceptable contours are defined as those with distance to agreement to those drawn by radiation oncologists less than 3 mm. For target volume surrounded by normal lung, the percentage slices of contours that do not need manual adjustment are 41–62%, 23–39%, 62–78% for threshold, auto‐segmentation and SNAP respectively. For cone beam CT, these numbers are approximately 10% lower. SNAP can also be used for target volume with no clear boundary, although the percentage success is much lower. Conclusion: More sophisticated auto‐segmentation tools need to be available routinely with more flexibility for users to adjust algorithm parameters in order for them to be useful for routine clinical ART purposes.
SU‐EE‐A2‐04: A Method of Online MLC Aperture Adjustment for Treatment of Patients with Set Up Variations33(2006); http://dx.doi.org/10.1118/1.2240215View Description Hide Description
Purpose: To investigate MLC aperture adjustments to compensate patient setup variations, replacing couch shift methods for precision treatment delivery. Method and Materials: Patient setup variations can be described by 3‐D translational shifts. A scheme of adjusting MLC apertures to compensate for translational displacements of the patient has been developed. Patient shift information, such as provided by commercial image matching software, establishes the aperture shift vector. The projection of this vector into and orthogonal to the BEV plane was used to determine the displacement vector and divergence for the aperture. Modified beam apertures were generated and MLC leaf positions were determined through a polynomialinterpolation.
Dosimetric plan comparisons were made within Pinnacle 7.6c. Static field and segmented IMRT patient plans were investigated for pelvic as well as head and neck sites. Arbitrary shift vectors ranging from 3 mm up to 30 mm have been investigated. Non‐integral MLC leaf width shifts parallel to the leaf bank face to characterize leaf width effects. Results: Conformal plans show dose variations from the original plan of up to 3% in the pelvis for shifts of up to 30 mm. Dose coverage of the PTV was maintained except in the superior‐inferior direction, where coverage in the last 3 mm of target fell as low as 92% of the prescribed dose. Results for prostate and oropharynx IMRT plans showed little increase in maximum critical structure doses, and small increases in mean doses. DVH's for IMRT plans confirmed minimal impact on critical structure doses. Conclusion: An alternative method of on‐line adaptive treatment delivery has been explored which eliminates the need to adjust the patient position. This can potentially increase treatment accuracy and efficiency through minimizing patient disturbances and reducing the time between imaging and treatment.
33(2006); http://dx.doi.org/10.1118/1.2240220View Description Hide Description
Purpose: To investigate a novel technique for on‐line adaptive radiation therapy(ART) that uses direct aperture optimization (DAO) Methods and Materials: A model simulating the geometry of a prostate case was created. The prostate, rectum and bladder are represented by an ellipsoid, cylinder and sphere, respectively (the dimensions and positions of these structures are based on patient image data). This configuration represented the “original geometry” and was used to create the original IMRT treatment plan. The plan was created using an in‐house DAO system with seven beams (40, 80, 110, 250, 280, 310, 355 gantry angles) and six apertures per beam. Four different “deformed geometries” were created by systematically deforming the original geometry to various degrees (0.25, 0.50, 0.75 and 1.00 cm maximum deformations of rectum and prostate). For each deformed geometry, a new treatment plan was created by modifying (adapting) the original treatment plan using DAO. The quality of the resulting plans, together with the optimization time efficiency of the plan adaptation, was used to assess the suitability of DAO for on‐line ART. The effects of altering different DAO parameters were investigated by varying the maximum leaf step size, maximum aperture weight change and optimization cooling schedule. Results: The plans created by adapting the original treatment plan met the imposed dose constraints for all four deformed geometries. Adapting the original treatment plan was much faster than performing a completely new re‐optimization. Furthermore, by appropriately limiting selective DAO parameters the convergence to an acceptable plan was significantly accelerated. The optimal choice of DAO parameter limits was correlated to the degree of geometry deformation. Conclusion: This study demonstrated that DAO is highly suitable for online ART. The treatment plan adaptation was efficient and the resulting plans met the imposed dose constraints.
SU‐EE‐A2‐06: Benefit of 3D Image‐Guided Stereotactic Localization in the Hypofractionated Treatment of Lung Cancer33(2006); http://dx.doi.org/10.1118/1.2240225View Description Hide Description
Purpose: To investigated the benefit of image‐guided stereotactic localization for lungcancer patients treated with hypofractionated radiotherapy.Materials & Methods: A stereotactic body localizer (SBL) was used for patient immobilization, image registration among multi‐phase CT simulation, and image‐guided stereotactic localization. The simulation scans consist of 3 sets of CTimages taken during free breathing and 2 breath‐holding scans (at maximum inhalation and exhalation). Target delineation was performed on all 3 sets of images and the combination of the targets forms a composite gross‐target volume (GTV). Treatment planning was performed on the planning‐target volume (PTV) using 3 mm margin to account for the presumed reliability of the CT localization. Prior to each treatment, a localization CT scan using a CT‐on‐rails was obtained. Couch shifts were made based on the changes of the stereotactic coordinates of three pre‐selected bony landmarks. In this retrospective treatmentdose verification, we performed image fusion between the simulation CT scan and each pre‐treatment CT scan to obtain the same target and critical structure information. The same treatment plans were re‐loaded onto each pre‐treatment CT scan with their respective stereotactic coordinate system. The changes in dose distributions were assessed by dose‐volume histograms of the PTV and the critical structures before and after isocenter corrections which were prompted by image guided stereotactic localization. We compared D95, D99, and V95 for the PTV and GTV, and V20and V30 for the ipsilateral lung.Results: Our retrospective study for 10 patients with 40 dose reconstructions showed that the average D95, D99, and V95 of the PTVs are 92.1%, 88.1%, and 95.8% of the planned values before isocenter corrections. With the corrections, these values are all improved to 100% of the planned values. Conclusions: 3D image guidance is crucial for stereotactic radiotherapy of lungtumors when small margins are used.
- Moderated Poster ‐ Area 2 (Joint): Modeling of Intra‐Fraction Organ Motion
SU‐DD‐A2‐01: On the Accuracy of a Moving Average Algorithm for Tracking Respiratory Motion During Radiation Therapy Treatment Delivery33(2006); http://dx.doi.org/10.1118/1.2240135View Description Hide Description
Introduction:Real‐time motion tracking (RTT) treatmentdelivery has several advantages toward the improvement of accuracy for radiotherapy. However, currently there are certain limitations to this technique. The purpose of this study was to investigate an alternative treatment scenario using a moving average algorithm (MA) for treatment which could potentially be approaching the accuracy of RTT. Method: A comparison was performed between three different treatment scenarios.
(3) Static beam delivery(SB) .
Where Xest (t) and Xact (t) are the estimated and actual position at time t, n in seconds is the averaging period (5–25 seconds range). The data used for this analysis was 331 respiration‐motion traces from 24 lung‐cancer patients acquired using three different breathing types (free breathing(FB), audio coaching(A) and audio‐visual biofeedback(AV)). The metrics used for comparison were the group systematic error(M), the standard deviation(SD) of the systematic error(Σ), and the root mean square of the random error(σ). The averaging period was varied to study the effect on the various metrics. Margins were calculated using the formula by Stroom et al.(IJROBP 1999;43(4)) Results: M and Σ are negligible for both MA[M ∈ (−0.01,0), Σ ∈ (0,0.01)] and RTT[ M ∈ (0), Σ ∈ (0)] compared to SB[ M ∈ (−0.15,−0.02), Σ ∈ (0.05–0.20)]. MA(0.48–0.54) has a slightly reduced σ than SB(0.53–0.57). Negligible improvements were found by varying the average periods for M and Σ . σ was found to be insensitive to the different averaging periods(0.53–0.56 for A). From the margin calculations FB is most affected by the different treatment scenarios. (All values in cm). Conclusions: MA has accuracy advantages over SB and practical advantages over RTT. MA significantly reduces M and Σ compared with SB. MA and SB require less margins for AV than that for FB and A. The margins required for RTT are independent of breathing training type. There is a group systematic error caused by intrafraction motion during FB.
SU‐DD‐A2‐02: Variability of Waveforms and Probability Distributions in External Respiratory‐Surrogate Marker Data33(2006); http://dx.doi.org/10.1118/1.2240136View Description Hide Description
Purpose: To investigate intra‐ and inter‐fractional, inter‐subject variability in the motion patterns of external respiratory‐surrogate markers. A strong correlation between the motion of external markers and internal targets has been previously reported. Method and Materials: Varian real‐time position management (RPM) system is used clinically to monitor external marker motion. We analyzed over 450 RPM datasets (traces) from 186 4D‐CT, and 6 gated radiotherapy subjects (mean length: 235 seconds). Aperiodic (long‐term) motion components were subtracted by applying high‐pass filtering to Fourier transform of the data. Probability distribution functions (PDF) of the marker position were constructed, and variability bounds were calculated for the realized distributions. Trace‐average waveforms (TAW) were constructed from cumulative PDF, calculated separately for leading and trailing edges of motion cycles within the trace. Results: Inter‐ and intra‐fractional variability of PDF were reduced where the aperiodic motion components were subtracted from the data. The distribution of aperiodic shifts was approximately Gaussian over multiple fractions. Comparison between the data from various subjects showed that the PDF (when normalized to the mean amplitude of individual traces) was remarkably stable, indicating rather limited inter‐fractional and inter‐subject variability. While intra‐fractional variability of PDF appeared to be typically larger than either inter‐fraction or inter‐subject, as a wide variety of waveforms were realized within each trace. Conclusion: The marker position PDF and its variability bounds, constructed based on a single trace (e.g., pre‐treatment 4D‐CT), may serve as a conservative estimate of the expected variability in the PDF realized during a fractionated treatment. This information can be used in robust optimization of treatment planning for moving targets. The TAW may potentially be useful in subject classification by “respiratory personality”, and prediction of the realized PDF for a given expected uncertainty in the trace extrema positions (full exhale and inhale). Supported in part by the NCI grant 5P01‐CA21239‐25.
33(2006); http://dx.doi.org/10.1118/1.2240137View Description Hide Description
The purpose of the present work is to develop and validate a series of tests to assess the quality of four‐dimensional (4D) computed tomography(CT)imaging as applied to radiation treatment planning. Using a commercial motion phantom and two programmable moving platforms with a CT phantom, we acquired 4D CT datasets on two commercial multislice helical CT scanners using different approaches to 4D CTimage reconstruction. 4D CTimagedata sets were obtained as the platform moves in different patterns designed to simulate various breathing patterns. Known inserts were contoured and statistics were generated to evaluate properties important to radiation therapy, namely phase‐binning accuracy, geometric accuracy, volume accuracy, and CT number accuracy. Phase‐binning accuracy varied by as much as 5% for a 4D procedure in which images were reconstructed, then binned, but exhibited no variation for a 4D procedure in which projections were binned prior to reconstruction. Geometric distortion was found to be small as was volume error. Partial volume effects in the direction perpendicular to the axial planes of reconstruction affected volume accuracy, however. CT numbers were reproduced accurately, but 4D images exhibited significantly more noise than static CTimages. Characterization of such properties can be used to better understand and optimize the various parameters that affect 4D CTimage acquisition.
SU‐DD‐A2‐04: A Simple Method to Reconstruct a Representative Mid‐Ventilation CT Scan From 4D Respiration Correlated CT Scans for Radiotherapy Treatment Planning of Lung Cancer Patients33(2006); http://dx.doi.org/10.1118/1.2240138View Description Hide Description
Purpose: Four‐dimensional (4D) imaging techniques can be used to obtain (respiration) artifact‐free CTimages of the thorax. However, its use in radiotherapy is limited since clinical treatment planning systems are currently not able to use the full 4D data. The purpose of this study was to reconstruct a representative single 3D CT scan from the 4D data set (with tumor closest to the mean position) for use in radiotherapy planning of lungtumors to enable reduction of treatment error margins. Method and Materials: After acquisition of the 4D CT scan (10 frames), the tumor is manually segmented (roughly) in the first frame and automatically (gray‐value) registered to the tumor in the subsequent frames. This gives the motion of the tumor during the respiratory cycle in 3D. Subsequently, from the cranio‐caudal (CC) tumor motion curve, the mean tumor position and its corresponding mid‐ventilation (MV) time‐percentage are calculated. The CT scan for planning is reconstructed at this time‐percentage. As indication of the merit of this concept, its effect on margins from CTV to PTV and on the PTV volume was calculated covering respiratory motion, respiratory baseline variation and setup errors (systematic and random). Results: Based on 13 patients, the worst tumor position accuracy (with respect to the mean tumor position) in the mid‐ventilation CT scan occurred in the anterior‐posterior direction: −0.7±0.8 mm (due to hysteresis). For these patients, the errors in conventional free‐breathing CT were estimated to be 0±3.4 mm (CC) and 0±1.4 mm (AP). The mid‐ventilation concept resulted in margin reduction up to 45% and a PTV volume reduction up to 35%. Conclusion: The mid‐ventilation concept, based on tumor motion, is a simple method to obtain an artifact‐free CT scan with smaller systematic errors compared to conventional CT scans. Significant reduction of the PTV volume can be achieved.
SU‐DD‐A2‐05: Impact of Fiducial Marker Placement for the Purpose of Phase Definition of the Respiratory Cycle for 4D‐CT Image Reconstruction33(2006); http://dx.doi.org/10.1118/1.2240140View Description Hide Description
Purpose: Most 4D‐CT acquisition methods rely on an externally measurable quantity proportional to the breathing cycle (e.g. chest wall excursion), for 4D‐CT image reconstruction. Typically, the position of a single reflective marker placed on the patient's chest is monitored. The marker location is often chosen primarily to maximize the measurable motion irrespective of proximity to tumor location. We examine the behavior of motion of multiple markers, at different locations, placed on the patient surface during 4D‐CT acquisition and evaluate the impact of marker location on respiratory cycle phase definition for 4D‐CT reconstruction and its subsequent application to radiotherapy planning. Method and Materials: An infrared guided positioning system (iGPS), capable of tracking multiple reflective fiducials in 3 dimensions, has been adapted to provide respiratory phase information for 4D‐CT reconstruction. Data for 3‐5 marker positions, placed at different locations on the patients chest, from 10 patients receiving 4D‐CT was examined. Results: For most patients (9/10) motion of 3–5 markers is reasonably well synchronized suggesting no significant effect of the fiducial location. For one patient, we observed a marker on the abdomen switch from being completely in‐phase to being completely out of phase relative to a marker on the center of the chest. This dramatically illustrates that the phase of a specific external marker may not correspond to the motion, external or internal, near the volume of interest. Conclusions: The position of a fiducial marker may affect not only the amplitude of motion but also the observed phase for some patients. The importance of this phase shift depends on how the resulting 4D‐CT is ultimately applied to radiotherapy planning. In particular, if specific phases (e.g. extremes) are selected for radiotherapy target definition, special attention should be paid to the location of the fiducial marker and its role in image reconstruction.
SU‐DD‐A2‐06: The Effect of Respiratory Rate and Radiation Timing On Dose Coverage in Dynamic Breast IMRT33(2006); http://dx.doi.org/10.1118/1.2240141View Description Hide Description
Purpose:IMRT has been shown to be capable of delivering plans with desirable homogeneous dose distribution for breast cancertreatment. However, the dose distribution may be influenced by interplay between dynamic MLC and respiratory motion. The purpose of this study was to investigate the impact of respiratory rate and radiation timing on the dose distribution of breast dynamic IMRT.Method and Materials: Using similar setup configuration, a helical CT and 4DCT image sets for six breast cancer patients were collected and contoured. Dynamic IMRT plans were designed using the helical CTimages. The planned MLC sequence was segmented according to the respiratory phases with a series of respiratory rates (7.5–30/min) and radiation timing (evenly distributed in respiratory cycles). The segmented dynamic MLC sequences were applied to the radiation fields on the corresponding 4DCT phases. A program was developed to calculate the cumulative dose distribution from all the phases. Results: For normal breathing rates (15–20/min), the dose coverage didn't change significantly regardless of radiation starting time. The change of target V90 was less than 2%. However, for extremely slow respiratory rates (7.5–10/min), the dose distribution and V90 changed significantly depending on the radiation timing. The change of target V90 was more than 10%. There was no significant dose coverage change for the underlying heart regardless respiratory rate or radiation timing. Conclusions: For breast patients treated with dynamic IMRT, if the respiratory rate of the patient is within the “normal” range then the impact of such respiration on dose coverage of the target was found to be statistically insignificant. However, the dose distribution may change significantly when patient has a slow breathing rate. Respiratory gating may be required to obtain satisfactory dose coverage for such cases. There was no significant dose distribution change for heart regardless respiratory rate or radiation timing.
- Moderated Poster ‐ Area 3 (Joint): Molecular Imaging & Image Registration / Fusion
33(2006); http://dx.doi.org/10.1118/1.2240226View Description Hide Description
Purpose:Protonradiotherapy activates positron emitters in tissue, which can be imaged with PET. However, the resulting PET image is not directly proportional to the delivered dose. We are investigating the spatial relationship between the dose distribution and its PET image without reverting to Monte Carlo methods. The first goal is to validate the proton range in the patient, and ultimately to reconstruct the spatial distribution of the actually delivered dose from its PET image.Method and Materials: The relationship between the protondose and its PET image can be described mathematically as a convolution (filtering). We derive the convolution kernel analytically. This filter is unique for a given activation channel, independent of beam energy and specific absorber. The straightforward application of the method to determine the PET signal by locally filtering the planned dose distribution was validated through comparisons with Monte Carlo calculations and measured PET data in homogeneous and inhomogeneous media. The challenging inversion of the relationship, determining the dose from the PET signal, was initially explored for a simplified mono‐energetic case in a homogeneous absorber. Results: Activity depth profiles obtained with the convolution approach agreed with measured and Monte Carlo data within 1 mm in depth. In terms of absolute intensity, the agreement was within 1.5% between filtered and simulated profiles and 10% between filtered and measured data in the distal region. Attempts to recover the dose distribution from its PET image through a de‐convolution yielded promising results for idealized data but were strongly noise dependent. Conclusion: We have derived the spatial relationship between dose and positron activation and demonstrated the possibility to obtain the PET image measured after proton treatment by locally filtering the planned dose distribution. The inverse approach, i.e., direct dose quantification from the measured PET, seems possible but is very sensitive to noise.
SU‐EE‐A3‐02: The Potential of FDG‐PET in Delineating the Lumpectomy Cavity for Partial Breast Irradiation Patients33(2006); http://dx.doi.org/10.1118/1.2240227View Description Hide Description
Purpose: To investigate the potential of FDG‐PET imaging for delineating the surgical cavity in post‐operative partial breast irradiation patients. Method and Materials: A DCIS breast cancer patient was imaged with a GE Discovery ST PET‐CT scanner approximately 2 weeks post lumpectomy. Following the treatment planningCT, a single‐bed (15 cm) FDG‐PET scan was dynamically acquired in 5‐sec intervals over 15 mins. The raw PET data was combined to form bins ranging from 30 sec to 15 min. These data were reconstructed by the GEscanner through an iterative OSEM algorithm, and hardware fused to the treatment planningCT. The value of PET in visualizing the lumpectomy cavity border was investigated through visual comparisons of fused PET‐CT images, the evolution of PET intensity for various breast points, and signal‐to‐noise measurements across the lesion. Results: The PETimage showed clear signal enhancement near the lumpectomy cavity. This enhancement formed a ring in each axial slice, matching the locations of surgical clips. Enhancement was also apparent where the cavity border was difficult to evaluate by CT density or clips alone. The ring presented a significantly higher SUV than other breast tissue (2.2 vs. <1.4), while the region inside the ring had a lower SUV than the (presumably benign) glandular tissue in the same breast (1.2 vs. 1.4). The SUV values were transient below 5 min, but remained stable thereafter. Conclusion: FDG‐PET may provide useful information for delineating lumpectomy cavity borders by elucidating regions of enhanced radiotracer uptake due to inflammation. The hypointense PET volume enclosed by the high activity ring may represent fluid and non‐viable tissue stemming from post‐surgical changes, explaining its lower FDG uptake and further supporting the suggestion that the ring corresponds to the cavity border itself. On current hardware, five‐minute scans were required to achieve stable SUVs.
SU‐EE‐A3‐03: A Biomechanical Lung Deformation Model Based On MR Grid‐Tagging Using Hyperpolarized 3He33(2006); http://dx.doi.org/10.1118/1.2240228View Description Hide Description
Purpose: Deformable image registration is critical for advancements of lungtumor radiation. Commonly used mathematical deformable registrations do not provide actual biomechanical trajectories of anatomical motion. Therefore, we developed a biomechanical registration model (BRM) based on hyperpolarized 3He MRI grid‐tagging to automatically segment structures and generate BRM encoded cine CTimages.Method and materials: Normal volunteers underwent hyperpolarized 3He MR imaging on a 1.5T whole‐body MR scanner. Grid‐tagging was achieved by applying sinc‐modulated RF pulses with a composite flip angle of 90° prior to the acquisition of the images followed immediately by a multi‐slice FLASH‐based acquisition at full inspiration and exhalation. For each slice, a displacement vector was computed for each grid element. The complete lung motion was based on spatial and temporal interpolation of the displacement vectors. The motion and deformation of anatomical structures were obtained using the interpolated vector field through various phases. The displacement vectors were registered to a coronal CTimage to generate a set BRM encoded cine CTimages.Results: Tagging signals were well preserved and sufficient for quantitatively resolving the tag motion. From the displacement vector map, the lower lobes exhibit the greatest motion magnitude especially in the craniocaudal direction. The motion of the structures driven by the displacement map is continuous and smooth. No substantial artifacts in the BRM generated dynamic Ctimages were observed. Conclusions and Discussion: BRM provides an independent measure of lung motion and deformation. Compared to pure‐mathematically constructed registrations, BRM relies on fewer assumptions and avoids errors induced by image matching processes. BRM encoded dynamic images are useful to cross‐validate deformable registration by other imaging modalities and algorithms. We plan to explore the potential of this methodology for auto segmentation and treatment planning.
33(2006); http://dx.doi.org/10.1118/1.2240229View Description Hide Description
Purpose: To describe a deformable registration infrastructure to resolve the geometric discrepancies between in vivo imaging studies and histology from resected tumor specimens to reduce uncertainties in tumor definition. Method and Materials: An IRB approved prospective study investigating the correlation between in‐vivo CT and MR imaging, ex‐vivo specimen imaging and pathologic sections from colorectal cancerliver metastases treated by resection was developed to better define gross and clinical tumor volume. Triphasic liverCT scans, PET‐CT scans and MR scans were obtained in 6 patients within 4 weeks prior to liver resection. On the day of surgery, the fresh liver hepatectomy specimen was imaged using MR. The specimen was fixed and reimaged with MR prior to pathological evaluation. Axial sectioning was done at the time of pathological evaluation, with photos of each liver slice digitized. Histological evaluation was performed on the sections representing the largest tumor. Gross tumor was identified on all imagessequences. Gross tumor, microscopic tumor and vascular changes of interest were also identified on the gross and histological pathological specimens. A finite element model‐based deformable modeling algorithm, MORFEUS, was used to resolve the geometric discrepancies due to changes in the position of the liver between each imagingsequence and session through a guided surface projection and finite element analysis.Results: Deformable registration can be used to facilitate comparison of imaging to pathological specimens. In the liver, substantial specimen shrinkage and deformation were seen necessitating deformable image registration. The accuracy of MORFEUS to relate the pathology‐histology to the in vivo imaging was within the slice thickness (5 mm) of the pathology sectioning, determined via identified vessel bifurcations in the liver.Conclusion: An accurate deformable modeling infrastructure has been established to relate the geometric position of the liver and excised liver specimen on different imaging modalities and histology.
33(2006); http://dx.doi.org/10.1118/1.2240230View Description Hide Description
Purpose: To develop a dose‐based evaluation method to assess deformable image registration accuracy Method and Materials: An algorithm developed for deformable registration of MVCT to kVCT images was evaluated. The algorithm allows the generation of automatic contours on MVCT images by transferring the kVCT contours using the deformation map. The automatically generated MVCT contours can thus be used to test the deformation algorithm by comparing these contours with manual contours. Instead of a geographic contour comparison, dosimetric endpoints were evaluated after the dose distribution was calculated in the MVCT images. Three dosimetric endpoints (Dmax, Dmean, and Dose to the hottest 2 cc (Dmax (2cc)) were compared for spinal cord contours. The evaluation of geometric end‐points is directly related to the clinical information that needs to be evaluated if daily images are used for adaptive radiation therapy. A total of 93 daily megavoltage CT (MVCT) images from three patients treated for cancers in the head and neck region were evaluated. Results: Averaged over all images the calculated Dmax differed between the automatic and manual contours by 1.1 % with a standard deviation of 3.5 %. The respective values for Dmean and Dmax(2cc) are 0.1 ± 2.5 % and 1.8 ± 2.4 %. Maximum deviations between the dosimetric endpoints were 12 % for Dmax, 8% for Dmean, and 13 % for Dmax(2cc). Conclusions: Using deformable image registration,dosimetric end‐points can be generated from automatic contours in the spinal cord region that differ from manual contours by 1–2 % on average with a standard deviation of 2.5 to 3.5 %. In the spinal cord region the developed deformable image registration appears to provide sufficient accuracy to support clinical decisions. Conflict of interest: Research supported by the vendor that is commercializing the algorithm. Several co‐authors are vendor employees.
SU‐EE‐A3‐06: Deformable Image Registration in Cone‐Beam CT Images for Image‐Guided Adaptive Radiotherapy33(2006); http://dx.doi.org/10.1118/1.2240231View Description Hide Description
Purpose: With the availability of on‐board imaging devices capable of constructing cone‐beam CT(CBCT)images, it is expected that there will be great interest in using volumetric CBCT for image‐guided adaptive radiotherapy. In order to fully utilize CBCT, automatic segmentation on CBCTimages is one of key steps toward this goal. The purpose of this study is to implement a robust deformable image registration for auto‐segmentation. Method and Materials: In four head and neck cancer patients, we used our previously developed, image intensity‐based deformable image registration algorithm to register the planning CT with the 3–5 daily CBCT in three scenarios. First, the daily CBCT was directly used without modification. Second, we applied a generic look‐up‐table transformation to map the CBCTimage intensity to the conventional CT intensity using the measured electron density calibration curves for both the conventional and CBCTscanners. In the third scenario, we proposed a wavelet‐based dynamic window/level histogram matching algorithm to map the CT number from CBCTimage to the conventional CTimage. Then the deformable image registration was performed in the modified CBCTimages to map the anatomical structures from the planning CT to the corresponding CBCTimages.Results: Without pre‐processing, we found that the CT numbers in CBCTimages were inconsistent, especially in soft tissue regions and in patients with large body circumferences. The deformable image registration using the window/level histogram matching method performed the best with good consistency in delineating soft tissue structures. The algorithm is also computationally efficient. Conclusion: We implemented a wavelet‐based window/level histogram matching algorithm to pre‐process the CBCT to allow for more robust deformable image registration of with the reference planning CT. This implementation allows for volumetric CBCT‐guided adaptive radiotherapy.
- Moderated Poster ‐ Area 3 (Joint): Tomographic Imaging for Therapy Localization
33(2006); http://dx.doi.org/10.1118/1.2240143View Description Hide Description
Purpose: To develop a method that identifies an IGRTimaging session as either normal or problematic based solely on the amount of right‐left, anterior‐posterior, and superior‐inferior repositioning of the patient over the treatment session. Methods and Materials: A retrospective data set containing over 1100 anterior‐posterior, right‐left lateral, and superior‐inferior patient shift values for 29 prostate patients was examined using a non‐parametric kernel regression classification method to determine if a patient was “normal” or “problematic.” The treatment sessions were grouped as either being “normal”, or affected because they were “overweight”, or had “rectal filling”, or were both “overweight and had rectal filling”. In kernel regression, constants are fitted using a locally weighted criterion. The basis of kernel regression is to estimate a response using a weighted average of points, in a training set, which are local to the query point. A bandwidth is used to determine the definition of local. Leave one out cross validation (LOOCV) was used to select the optimal bandwidth and also evaluate the technique's performance. Results: The method correctly classified 24 of the 29 patients using their respective shift data sets, with four of the misclassifications occurring when the technique correctly identified non‐normal datasets, but assigned them to the wrong problem group. Only one patient was classified as normal incorrectly. Conclusion: Using readily accessible shift data, the kernel regression classification method was able to correctly identify the cause behind IGRT positioning problems for individual prostate patients. This technique is fully automated and can be implemented on a treatment planning computer to determine the reason a patient is having positioning errors early during treatment.
SU‐DD‐A3‐02: Evaluation of Helical Tomotherapy Megavoltage CT System for Daily Automatic Patient Setup Correction and Manual Prostate Gland Motion Correction33(2006); http://dx.doi.org/10.1118/1.2240144View Description Hide Description
Purpose: To evaluate the efficacy of helical tomotherapy megavoltage CT system (MVCT), and to study patient setup uncertainty and inter‐fractional internal organ motion for prostate cancer patients during the course of external beam treatment. Method and Materials: 34 prostate cancer patients that received definitive helical tomotherapy treatments were included in this study. MVCT images were registered with planning CTimages using automatic bone registration followed by manual registration based on soft tissue match. Patient setup corrections and internal organ motion corrections in the medial‐lateral (ML), superior‐inferior (SI), anterior‐posterior (AP) directions, and rotations around the longitudinal axis were obtained from 1345 daily MVCT image registrations.Results: The mean and standard deviation of patient setup corrections were 3.1±7.3 mm in the ML direction, −0.8±4.9 in the SI direction, −0.2±6.4 in the AP direction, and 0.8±1.3° for rotations around the longitudinal axis. The mean and standard deviation of internal organ motion corrections were −0.1±0.8 mm in the ML direction, −0.1±0.7 mm in the SI direction, and 0.0±1.9 mm in the AP direction. The fraction of manual registrations that did not have adjustment in the ML, SI, or AP direction was 84%, 95%, and 71%, respectively. The prostate motion variability did not change during the course of treatment. Conclusion: Patient setup uncertainty dominated target position uncertainty. Helical tomotherapy MVCT system was effective in correcting patient setup errors and internal organ motions in the ML and AP directions, but provided limited soft tissue resolution in the SI direction.