S Kriminski

Evaluation of respiration‐correlated digital tomosynthesis in lunga)

Digital tomosynthesis (DTS) with a linear accelerator-mounted imaging system provides a means of reconstructing tomographic images from radiographic projections over a limited gantry arc, thus requiring only a few seconds to acquire. Its application in the thorax, however, often results in blurred images from respiration-induced motion. This work evaluates the feasibility of respiration-correlated (RC) DTS for soft-tissue visualization and patient positioning. Image data acquired with a gantry-mounted kilovoltage imaging system while recording respiration were retrospectively analyzed from patients receiving radiotherapy for non-small-cell lung carcinoma. Projection images spanning an approximately 30 degrees gantry arc were sorted into four respiration phase bins prior to DTS reconstruction, which uses a backprojection, followed by a procedure to suppress structures above and below the reconstruction plane of interest. The DTS images were reconstructed in planes at different depths through the patient and normal to a user-selected angle close to the center of the arc. The localization accuracy of RC-DTS was assessed via a comparison with CBCT. Evaluation of RC-DTS in eight tumors shows visible reduction in image blur caused by the respiratory motion. It also allows the visualization of tumor motion extent. The best image quality is achieved at the end-exhalation phase of the respiratory motion. Comparison of RC-DTS with respiration-correlated cone-beam CT in determining tumor position, motion extent and displacement between treatment sessions shows agreement in most cases within 2-3 mm, comparable in magnitude to the intraobserver repeatability of the measurement. These results suggest the methods applicability for soft-tissue image guidance in lung, but must be confirmed with further studies in larger numbers of patients.

Accurate positioning for head and neck cancer patients using 2D and 3D image guidance

Our goal is to determine an optimized image‐guided setup by comparing setup errors determined by two‐dimensional (2D) and three‐dimensional (3D) image guidance for head and neck cancer (HNC) patients immobilized by customized thermoplastic masks. Nine patients received weekly imaging sessions, for a total of 54, throughout treatment. Patients were first set up by matching lasers to surface marks (initial) and then translationally corrected using manual registration of orthogonal kilovoltage (kV) radiographs with DRRs (2D‐2D) on bony anatomy. A kV cone beam CT (kVCBCT) was acquired and manually registered to the simulation CT using only translations (3D‐3D) on the same bony anatomy to determine further translational corrections. After treatment, a second set of kVCBCT was acquired to assess intrafractional motion. Averaged over all sessions, 2D‐2D registration led to translational corrections from initial setup of 3.5±2.2 (range 0–8) mm. The addition of 3D‐3D registration resulted in only small incremental adjustment (0.8±1.5mm). We retrospectively calculated patient setup rotation errors using an automatic rigid‐body algorithm with 6 degrees of freedom (DoF) on regions of interest (ROI) of in‐field bony anatomy (mainly the C2 vertebral body). Small rotations were determined for most of the imaging sessions; however, occasionally rotations >3° were observed. The calculated intrafractional motion with automatic registration was <3.5 mm for eight patients, and <2° for all patients. We conclude that daily manual 2D‐2D registration on radiographs reduces positioning errors for mask‐immobilized HNC patients in most cases, and is easily implemented. 3D‐3D registration adds little improvement over 2D‐2D registration without correcting rotational errors. We also conclude that thermoplastic masks are effective for patient immobilization. PACS number: 87.53.Kn

COMPARISON OF KILOVOLTAGE CONE-BEAM COMPUTED TOMOGRAPHY WITH MEGAVOLTAGE PROJECTION PAIRS FOR PARASPINAL RADIOSURGERY PATIENT ALIGNMENT AND POSITION VERIFICATION

PURPOSE Implanted gold markers and megavoltage (MV) portal imaging are commonly used for setup verification of paraspinal tumors treated with high-dose, single-fraction radiotherapy. We investigated whether the use of kilovoltage cone-beam computed tomography (CBCT) imaging eliminates the need for marker implantation. METHODS AND MATERIALS Patients with paraspinal disease who were eligible for single-fraction stereotactic body radiotherapy were accrued to an institutional review board-approved protocol. Each of 16 patients underwent implantation of fiducial markers near the target. The markers were visible on the MV images. Three MV image pairs were acquired for each patient (initial, verification, and final) and were registered to the reference images. Every MV pair was complemented by a CBCT scan. CBCT image registration was performed automatically by maximizing the mutual information using a region of interest that excluded the markers. The corrections, as determined from the MV images, were compared with these from CBCT and were used for actual patient setup. RESULTS The mean and standard deviation of the absolute values of the differences between the CBCT and MV corrections were 1.0 +/- 0.7, 1.0 +/- 0.6, and 1.0 +/- 0.8 mm for the left-right, anteroposterior, and superoinferior directions, respectively. The absolute differences between the corresponding pre- and post-treatment kilovoltage CBCT image registration were 0.6 +/- 0.5, 0.6 +/- 0.5, and 1.0 +/- 0.8 mm. CONCLUSION The setup corrections found using CBCT without the use of implanted markers were consistent with the marker registration on MV projections. CBCT has additional advantages, including better positioning precision and robust automatic three-dimensional registration, as well as eliminating the need for invasive marker implantation. We have adopted CBCT for the setup of all single-fraction paraspinal patients. Our data have also demonstrated that target displacements during treatment are insignificant.

Correction of image artifacts from treatment couch in cone-beam CT from kV on-board imaging

PURPOSE To investigate image artifacts caused by a standard treatment couch on cone-beam CT (CBCT) images from a kV on-board imager and to develop an algorithm based on spatial domain filtering to remove image artifacts in CBCT induced by the treatment couch. METHODS Image artifacts in CBCT induced by the treatment couch were quantified by scanning a phantom used to quantify CT image performance. This was performed by scanning the phantom setup on a regular treatment couch and in air with the kV on-board imager. An algorithm was developed to filter image artifacts from the treatment couch by processing of cone-beam radiographic projections using two scans: one scan of the phantom and treatment couch and a second scan of the treatment couch only. This algorithm is based on a pixel-by-pixel removal of beam attenuation due to the treatment couch from each projection of the phantom and couch scan. The net couch-filtered projections were then used to reconstruct CBCT. RESULTS We found that the treatment couch causes considerable image artifacts: CT number uniformity is degraded and varies as much as 15%, and noise in CBCT scans with phantom plus couch (3.5%) is higher than for the phantom in air (1.5%). The spatial domain filtering technique reduces noise by more than 1.5%, improves uniformity by a factor of 2, and removes ringing and streaking artifacts related to the standard treatment couch in CBCT reconstructed from couch-filtered projections. This filtering technique was tested successfully to filter other hardware objects such as a patient immobilization body-fix frame. CONCLUSIONS The standard treatment couch causes image artifact in CBCT from kV on-board imaging systems. The spatial domain filtering technique developed in this work improves image quality of CBCT by preprocessing the projections prior to CBCT reconstruction. This technique might be useful to filter other hardware objects from CBCT which may contribute to the degradation of image quality.

SU‐DD‐A3‐05: Evaluation of Respiration‐Correlated Digital Tomosynthesis for Soft Tissue Visualization

Purpose: To find optimal parameters for digital tomosynthesis (DTS) image acquisition, assess DTS imaging for soft tissue visualization and patient positioning, and determine if DTS can be acquired fast enough to avoid blur caused by the respiratory motion Methods and Materials: We have used Varian gantry‐mounted kV on‐board imaging system to acquire DTS images as well as reference cone‐beam CT(CBCT) scans. An external respiratory monitor system recorded patient respiration together with the x‐ray on/off signal during imaging for retrospective sorting of projections based on respiration phase. DTS reconstruction used backprojection followed by a deblur. For a lung tumor subject to the respiratory motion we also reconstructed DTS images during a short time interval (∼ 1 s = 6° arc at 1 rpm) around the end‐exhalation. Results: Phantom studies indicate that image quality increases with DTS arc length; however, longer arc lengths cause image blur and degradation. Optimal DTS arc length is 10–20.° Patient studies also indicate that at approximately 15° arc length image quality, as judged visually, is the best. For longer arcs image blur increases, while for shorter arcs out of plane objects become more pronounced. For all arc lengths tumor visualization was possible. Both manual and automatic 2D registrations of DTS and CBCT were possible in most cases. For short (6°) or long (30°) arc lengths manual registration became more challenging and automatic registration less precise, but still possible. Registration of a respiratory correlated DTS image over a 6,° non‐optimal arc, was possible. Conclusions: DTS is capable of soft tissue and bone visualization and can be an efficient imaging modality for image‐guided radiotherapy. DTS can be acquired, with some tradeoff in image quality, during a ∼1s time interval, allowing reduction of respiratory motion artefacts. Conflict of interest: Research sponsored by NCI Grant P01‐CA59017 and Varian Medical Systems.

SU-FF-J-38: Evaluation of Positioning for Head and Neck Patients Using 2D and 3D Image Guidance

Purpose: To determine and compare reduction of setup error by two‐dimensional (2D) and three‐dimensional (3D) image guidance for radiation therapy of head and neck cancer (HN a loose‐fitting mask and resulting anatomical deformation were noted. However, the impact on dose even for this patient was minimal. For all patients, rotations determined by the 6 DOF registration of in‐field bony anatomy were < 4° around any axis. Intra‐fraction motion was < 2 mm and < 2° for all sessions. Conclusions: Manual 2D/2D registration of in‐field bony anatomy reduces positioning errors for mask‐immobilized H&N patients in most cases, and is easily implemented. 3D registration adds little improvement.

TH‐D‐BRC‐05: Respiratory Motion Correction of Cone‐Beam CT in Abdomen Using a Patient‐Specific Motion Model

Purpose: Respiratory motion reduction methods to improve cone‐beam CT quality (CBCT) have focused on the thorax, but reduced tissuecontrast in abdomen poses additional challenges. We report a method to correct CBCT in abdomen, using a motion model adapted to the patient from a prior respiration‐correlated CT (RCCT) image set. Method and Materials: Model adaptation consists of nonrigid image registration that maps each RCCT image to a reference image in the set, followed by principal component analysis (PCA) to reduce noise in the resultant deformation fields and relate them to diaphragm position and motion (inhalation or exhalation). CBCT projection images are sorted into subsets according to diaphragm position in the images and reconstructed, yielding a set of low‐quality 3‐D images. Model application deforms the CBCTimages to a reference CBCT in the set; combining all images yields a high‐quality CBCTimage with reduced motion artifacts. We also investigate a simpler correction method, which does not use PCA and correlates motion state with respiration phase. Comparison of contrast‐to‐noise ratios of pixel intensities within kidneys relative to surrounding background tissue provides a quantitative assessment of relative organ visibility. Results: Evaluation of CBCT examples in upper abdomen shows that streaking artifacts and blurring of liver, kidneys, spleen, bowel and implanted fiducial markers are visibly reduced with PCA‐model‐based correction. Phase‐based motion correction without PCA reduces blurring less effectively; in addition, implanted markers appear broken up, indicating inconsistencies in the correction. Model‐based motion correction shows the highest contrast‐to‐noise ratios in the cases examined. Conclusion: Motion correction of CBCT in abdomen is feasible and yields observable improvement. The PCA‐based model is an important component: first, by removing noise; second, by relating deformation to diaphragm position rather than phase, thus accommodating breathing pattern changes between imaging sessions.

WE-E-M100F-05: Evaluation of Respiration-Correlated Digital Tomosynthesis in the Thorax and Abdomen for Soft Tissue Visualization and Patient Positioning

Purpose: To find optimal parameters for digital tomosynthesis (DTS) image reconstruction, to evaluate ability of respiration correlated DTS to reduce blur caused by respiratory motion, and to assess DTS imaging for soft tissue localization and patient positioning. Methods and Materials:Image acquisition for DTS used a gantry‐mounted kV on‐board imaging system (Varian Medical Systems). We did not acquire DTS separately, but instead simulated DTS acquisition by using projection images acquired for CBCT. DTS reconstruction consisted of backprojection followed by a deblurring operation removing out‐of‐plane objects. For tumors subject to respiratory motion we selected projection images according to an external respiratory monitor signal (Real‐time Position Management System, Varian Medical Systems). Reconstruction and registration of DTS images used vender research software. Results: Optimal DTS quality is achieved with a 6–9 cm long deblurring volume in the direction perpendicular to the image and 20–30° reconstruction arc lengths. Image blur increases with longer arc lengths, while for shorter arcs out of plane objects become more pronounced. RC DTS reconstruction from disjoint arcs containing 2–3 respiratory cycles is feasible, yields images with less motion blur, and allows visualization of tumor movement. Generally, DTS was capable to visualize lungtumors, bronchi, liver, kidneys and abdominal lymph nodes. Estimated 3D registration error of DTS to reference DTS generated from the plan CT DRRs was 3 mm, relative to cone‐beam CT as a standard. Registration of RC DTS of lungtumors was possible to within 2 mm. Conclusions: DTS is capable of soft tissue visualization and patient positioning, including tumors subject to respiratory motion. With several advantages over full rotation CBCT scan, such as much shorter acquisition time, smaller dose and relaxed clearance requirement, DTS can become an efficient imaging modality for image‐guided radiotherapy. Conflict of Interest: Research sponsored by NCI Grant P01‐CA59017 and Varian Medical Systems.

SU‐FF‐J‐49: Reducting of Motion Artifacts in Cone Beam CT Using a Patient Specific Respiratory Motion Model

Purpose: Respiratory motion degrades the quality of cone beam computed tomography(CBCT)images in the thorax and abdomen and limits its localization accuracy. We describe a method of reducing motion‐induced artifacts in CBCTimages in the thorax using a patient‐specific model to estimate 3‐dimensional respiration‐induced motion. Method and Materials: The patient‐specific model is derived from, and applied to, the same CBCT data, thus avoiding inconsistencies caused by changes in patients breathing pattern when using different image sets. The CBCT scan uses a 1‐minute gantry rotation while recording patient respiration with an external monitor. The projection images are sorted into 4 to 6 phase bins according to the respiratory signal. Each phase bin is reconstructed to produce a series of 3D images. Nonrigid image registration calculates a series of deformation fields that maps each 3D image to a reference image at end expiration. A principle component analysis (PCA) reduces noise and redundancy in the deformation fields. The resultant deformation fields correct each 3D image by morphing it to the reference motion state. The corrected 3D images are combined to obtain a high‐resolution CBCTimage. We evaluate the model by comparing CBCTimages before and after motion correction, in patients receiving radiation treatment for lungcancer.Results: Motion‐corrected CBCTimages show less respiration‐induced blurring and streaking artifacts compared to standard CBCT. Fine‐detailed features in lungtumor and airways become visible, and the liver boundary is more discernable. Conclusion: Preliminary results indicate that the proposed method is a potentially useful tool for improving CBCTimage quality and localization accuracy in the thorax. Repeating the process at other phase bins can yield a high quality respiration‐correlated (4D) CBCTimage set for evaluating respiration‐induced motion.

TH‐D‐VaIB‐01: Correction of Couch Artifacts in KV Cone‐Beam CT From An On‐Board Imaging System

Purpose: To evaluate artifacts caused by treatment couch attenuation on 3D image reconstruction for a new kV on‐board‐imager (OBI) and cone beam CT(CBCT) system and to develop an algorithm that filters couch effects from two‐dimensional radiographic projections prior to inputting to the 3D reconstruction algorithm. Material and methods: A standard quality assurance phantom was scanned in air and on couch top using both full and half fan cone‐beam scanning modes with and without bowtie filter combination. A spatial domain filter algorithm was developed to remove couch attenuation from each radiographic projection. This filter is based on a pixel‐by‐pixel subtraction technique of radiographic projections of cone‐beam scans of the couch from the corresponding radiographic projections of scans with phantom on top of the couch. The net couch‐filtered radiographic projections were used to reconstructCTimages.Results:CT numbers for scans of the phantom on couch top are less uniform than for scans of the phantom in air. The couch artifacts vary the linearity of the CT numbers by 5–15%, depending on the density of the material. Noise of the scans with phantom on couch top (3.5%) is higher than that with phantom in air (1.5%). The increased noise hinders the ability of the CBCT system to resolve low‐contrast regions when the couch is present. Pre‐reconstruction processing of the couch suppresses noise (< 1.5%) improves uniformity by a factor of 2 and removes ring and streak artifacts in the couch‐filtered reconstructedCBCTimages.Conclusion: The treatment couch produces streaking artifacts, enhances noise, and causes drifting of CT numbers in the reconstructed OBI CBCTimages. The developed couch pre‐processing algorithm suppresses noise, improves CT number uniformity by a factor of 2 and removes ring and streak artifacts in the couch‐filtered reconstructedCBCTimages. Conflict of Interest: Supported by NCI Grant P01‐CA59017.