L Jiang

An accurate algorithm to match imperfectly matched images for lung tumor detection without markers

In order to locate lung tumors on kV projection images without internal markers, digitally reconstructed radiographs (DRRs) are created and compared with projection images. However, lung tumors always move due to respiration and their locations change on projection images while they are static on DRRs. In addition, global image intensity discrepancies exist between DRRs and projections due to their different image orientations, scattering, and noises. This adversely affects comparison accuracy. A simple but efficient comparison algorithm is reported to match imperfectly matched projection images and DRRs. The kV projection images were matched with different DRRs in two steps. Preprocessing was performed in advance to generate two sets of DRRs. The tumors were removed from the planning 3D CT for a single phase of planning 4D CT images using planning contours of tumors. DRRs of background and DRRs of tumors were generated separately for every projection angle. The first step was to match projection images with DRRs of background signals. This method divided global images into a matrix of small tiles and similarities were evaluated by calculating normalized cross‐correlation (NCC) between corresponding tiles on projections and DRRs. The tile configuration (tile locations) was automatically optimized to keep the tumor within a single projection tile that had a bad matching with the corresponding DRR tile. A pixel‐based linear transformation was determined by linear interpolations of tile transformation results obtained during tile matching. The background DRRs were transformed to the projection image level and subtracted from it. The resulting subtracted image now contained only the tumor. The second step was to register DRRs of tumors to the subtracted image to locate the tumor. This method was successfully applied to kV fluoro images (about 1000 images) acquired on a Vero (BrainLAB) for dynamic tumor tracking on phantom studies. Radiation opaque markers were implanted and used as ground truth for tumor positions. Although other organs and bony structures introduced strong signals superimposed on tumors at some angles, this method accurately located tumors on every projection over 12 gantry angles. The maximum error was less than 2.2 mm, while the total average error was less than 0.9 mm. This algorithm was capable of detecting tumors without markers, despite strong background signals. PACS numbers: 87.57.cj, 87.57.cp87.57.nj, 87.57.np, 87.57.Q‐, 87.59.bf, 87.63.lm

Automated landmark-guided deformable image registration

The purpose of this work is to develop an automated landmark-guided deformable image registration (LDIR) algorithm between the planning CT and daily cone-beam CT (CBCT) with low image quality. This method uses an automated landmark generation algorithm in conjunction with a local small volume gradient matching search engine to map corresponding landmarks between the CBCT and the planning CT. The landmarks act as stabilizing control points in the following Demons deformable image registration. LDIR is implemented on graphics processing units (GPUs) for parallel computation to achieve ultra fast calculation. The accuracy of the LDIR algorithm has been evaluated on a synthetic case in the presence of different noise levels and data of six head and neck cancer patients. The results indicate that LDIR performed better than rigid registration, Demons, and intensity corrected Demons for all similarity metrics used. In conclusion, LDIR achieves high accuracy in the presence of multimodality intensity mismatch and CBCT noise contamination, while simultaneously preserving high computational efficiency.

TU-A-9A-01: A Precise Deformable Image Registration System Using Feature-Based Irregular Meshes

PURPOSE To develop a deformable image registration system using irregular meshes generated based on image features. Finer meshes are generated at organ or tissue boundaries so that higher deformation resolution is achieved with limited number of control points. METHODS This deformable image registration system consists of two parts: a finite element modeling system and a GPU-based registration system. The finite element modeling system is used to generate irregular meshes for the entire body or region of interest without segmentation. This system uses a Laplacian of Gaussion operator to extract features such as boundaries between organs and tissues. The placement of control points (vertices) of meshes is optimized based on feature intensities. The registration framework uses elastic energy as regularity and guarantees the final deformation vector field to be diffeomorphic. This system has been compared with registration algorithms based on regular meshes or voxels (Demons). The comparison is performed on XCAT digital phantom data and patient data. RESULTS Both XCAT male phantom and female phantom were used. Two respiratory phases were treated as floating and fix images for every phantom. Similarity measures were performed for image intensity and displacement vector field (DVF) for four sets of testing data with three types of deformation algorithms (irregular mesh, regular grid, and voxel). It clearly shows that the voxel-based Demons algorithm leads to slightly better images. However, the DVF results show that mesh-based algorithms behave much better. Compared with regular meshes, the irregular mesh we proposed leads to much faster convergence and converge to better results. The calculation is accelerated by GPU cards and a typical registration of patient data could be completed in about 1 minute. CONCLUSION The feature-based irregular meshing method provides a fast and accurate deformable image registration. This research is supported by CPRIT individual investigator award RP110329 and Varian research grant.

WE-D-9A-02: Automated Landmark-Guided CT to Cone-Beam CT Deformable Image Registration

PURPOSE The anatomical changes that occur between the simulation CT and daily cone-beam CT (CBCT) are investigated using an automated landmark-guided deformable image registration (LDIR) algorithm with simultaneous intensity correction. LDIR was designed to be accurate in the presence of tissue intensity mismatch and heavy noise contamination. METHOD An auto-landmark generation algorithm was used in conjunction with a local small volume (LSV) gradient matching search engine to map corresponding landmarks between the CBCT and planning CT. The LSVs offsets were used to perform an initial deformation, generate landmarks, and correct local intensity mismatch. The landmarks act as stabilizing controlpoints in the Demons objective function. The accuracy of the LDIR algorithm was evaluated on one synthetic case with ground truth and data of ten head and neck cancer patients. The deformation vector field (DVF) accuracy was accessed using a synthetic case. The Root mean square error of the 3D canny edge (RMSECE), mutual information (MI), and feature similarity index metric (FSIM) were used to access the accuracy of LDIR on the patient data. The quality of the corresponding deformed contours was verified by an attending physician. RESULTS The resulting 90 percentile DVF error for the synthetic case was within 5.63mm for the original demons algorithm, 2.84mm for intensity correction alone, 2.45mm using controlpoints without intensity correction, and 1.48 mm for the LDIR algorithm. For the five patients the mean RMSECE of the original CT, Demons deformed CT, intensity corrected Demons CT, control-point stabilized deformed CT, and LDIR CT was 0.24, 0.26, 0.20, 0.20, and 0.16 respectively. CONCLUSION LDIR is accurate in the presence of multimodal intensity mismatch and CBCT noise contamination. Since LDIR is GPU based it can be implemented with minimal additional strain on clinical resources. This project has been supported by a CPRIT individual investigator award RP11032.

WE-D-9A-01: A Novel Mesh-Based Deformable Surface-Contour Registration

PURPOSE Initial guess is vital for 3D-2D deformable image registration (DIR) while dealing with large deformations for adaptive radiation therapy. A fast procedure has been developed to deform body surface to match 2D body contour on projections. This surface-contour DIR will provide an initial deformation for further complete 3D DIR or image reconstruction. METHODS Both planning CT images and come-beam CT (CBCT) projections are preprocessed to create 0-1 binary mask. Then the body surface and CBCT projection body contours are extracted by Canny edge detector. A finite element modeling system was developed to automatically generate adaptive meshes based on the image surface. After that, the projections of the CT surface voxels are computed and compared with corresponding 2D projection contours from CBCT scans. As a result, the displacement vector field (DVF) on mesh vertices around the surface was optimized iteratively until the shortest Euclidean distance between the pixels on the projections of the deformed CT surface and the corresponding CBCT projection contour is minimized. With the help of the tetrahedral meshes, we can smoothly diffuse the deformation from the surface into the interior of the volume. Finally, the deformed CT images are obtained by the optimal DVF applied on the original planning CT images. RESULTS The accuracy of the surface-contour registration is evaluated by 3D normalized cross correlation increased from 0.9176 to 0.9957 (sphere-ellipsoid phantom) and from 0.7627 to 0.7919 (H&N cancer patient data). Under the GPU-based implementation, our surface-contour-guided method on H&N cancer patient data takes 8 seconds/iteration, about 7.5 times faster than direct 3D method (60 seconds/iteration), and it needs fewer optimization iterations (30 iterations vs 50 iterations). CONCLUSION The proposed surface-contour DIR method can substantially improve both the accuracy and the speed of reconstructing volumetric images, which is helpful for applying in adaptive radiotherapy. This research is supported by CPRIT individual investigator award RP110329.

TH-E-17A-10: Markerless Lung Tumor Tracking Based On Beams Eye View EPID Images

PURPOSE Dynamic tumor tracking or motion compensation techniques have proposed to modify beam delivery following lung tumor motion on the flight. Conventional treatment plan QA could be performed in advance since every delivery may be different. Markerless lung tumor tracking using beams eye view EPID images provides a best treatment evaluation mechanism. The purpose of this study is to improve the accuracy of the online markerless lung tumor motion tracking method. METHODS The lung tumor could be located on every frame of MV images during radiation therapy treatment by comparing with corresponding digitally reconstructed radiograph (DRR). A kV-MV CT corresponding curve is applied on planning kV CT to generate MV CT images for patients in order to enhance the similarity between DRRs and MV treatment images. This kV-MV CT corresponding curve was obtained by scanning a same CT electron density phantom by a kV CT scanner and MV scanner (Tomotherapy) or MV CBCT. Two sets of MV DRRs were then generated for tumor and anatomy without tumor as the references to tracking the tumor on beams eye view EPID images. RESULTS Phantom studies were performed on a Varian TrueBeam linac. MV treatment images were acquired continuously during each treatment beam delivery at 12 gantry angles by iTools. Markerless tumor tracking was applied with DRRs generated from simulated MVCT. Tumors were tracked on every frame of images and compared with expected positions based on programed phantom motion. It was found that the average tracking error were 2.3 mm. CONCLUSION This algorithm is capable of detecting lung tumors at complicated environment without implanting markers. It should be noted that the CT data has a slice thickness of 3 mm. This shows the statistical accuracy is better than the spatial accuracy. This project has been supported by a Varian Research Grant.

SU‐E‐T‐463: A Dosimetric Evaluation of CBCT Guided Intra‐Fractional Adaptive Radiotherapy for VMAT

Purpose: To investigate the dosimetric advantages of an image guided replanning strategy utilizing on‐board CBCTs for VMAT plans. An intensity corrected grey‐scale based CBCT to planning CT deformable image registration algorithm, combined with an intra‐fractional morphing aperture optimization algorithm (ICDIR‐MAO), is dosimtrically evaluated for head and neck cancer patients. Methods: : Head and neck VMAT plans were retrospectively adapted using the ICDIR‐MAO algorithm. Patient CBCT and CT images are deformably registered with an intensity corrected demons based deformable image registration algorithm. The deformation vector field is applied to the PTV on the structure set to obtain a deformed PTV. Digitally reconstructed radiographs of the original PTV and deformed PTV are created for each treatment angle. MLC positions are generated to contour both PTVs and a ratio of original vs. deformed MLC positions that outline the PTV beams eye view contour is applied to each angle of the planned MLC sequence. The deformed image is then sliced up and reformatted back into dicom files, with the associated metadata. The deformation vector field is added to the structure set at every point and a new dicom structure set is created. The new mlc sequences are reformatted back into dicom plans. All three dicom files are imported back into eclipse, and dose is re‐calculated on the deformed plan, deformed image set, and associated deformed structures. Results: The mean and maximum dose ratios for the OARS were lowered for all cases. The mean dose for the PTV70Gy, PTV60Gy, and PTV54Gy was the same or better for the deformed plans. The dosimetric improvement was proportional to the contour volume overlap between the deformed CT and the original CT. Conclusion: : ICDIR‐MAO for adaptive VMAT replanning improves the dosimetric quality for head and neck patients. This research is supported by CPRIT Individual Investigator Award RP110329

TH‐C‐137‐06: An Inter‐Fractional Morphing Aperture Based Reoptimization Tool for Adaptive Radiotherapy

PURPOSE To dosimetrically evaluate a streamlined adaptive radiotherapy tool to account for inter-fractional tumor shrinkage. A novel aperture-morphed simultaneously optimized finite sized pencil beam algorithm (AMSO-FSPB) was investigated over volumetric modulated arc therapy (VMAT) head and neck plans. METHODS AMSO-FSPB plans were retrospectively designed for head and neck patients in the supine position. CT image sets between fractions were rigidly registered using normalized cross correlation. A demons based deformable image registration algorithm was used to obtain a deformation vector field (DVF) between fractions. The DVFs were used to map contours between fractions. An eclipse and pinnacle compatible contour mapping algorithm was developed to apply the DVFs to the structure dicom set. DRRs at each gantry angle were used to derive the ratio of tumor shrinkage at every MLC position. The DVF and the target contour shrinkage were used to obtain a new set of fluence maps. The machine tolerance guidelines for the Varian True beam were followed for this study. A graphics processor united (GPU) based FSPB dose calculation engine with an embedded weighting factor re-optimization algorithm was developed to reassign beamlet weighting factors to every 4 degrees of arch length in the VMAT plan. RESULTS The dosimetric advantages of the AMSO-FSPB VMAT plans were directly proportional to the amount of the tumor regression between fractions. AMSO-FSPB VMAT plans were dosimetrically superior to the original plans for all cases. The ratio of the mean and maximum dose for the Paratoids, Spinal Cord, and Brain Stem were lowered for all cases. The mean for the PTV70Gy, PTV60Gy, and PTV54Gy was increased while the standard deviation was decreased for all cases. CONCLUSION Implementation of the AMSO-FSPB algorithm improved the dosimetric quality of the supine head and neck VMAT plans and was done so with minimal additional cost to clinical resources. This research is supported by CPRIT Individual Investigator Award RP110329.

SU‐E‐T‐629: Initial Results of VMAT Re‐Planning for On‐Line Adaptive Radiotherapy

PURPOSE The volumetric-modulated arc therapy (VMAT) has been widely used due to its fast delivery and good dose confirmation to target. Adaptive radiation therapy becomes essential to address the anatomic changes during a course of radiotherapy. A fast algorithm to re-plan treatment has been investigated to adapt the changes of targets and OARs. METHODS All studies were based on data of head/neck cancer patients treated by VMAT in our institution. These patients had second CT scans, contours, and treatment plans in the middle of the treatment course. CT image datasets, ROI structures, and arc beam apertures were exported from a commercial treatment planning system (Pinnacle v9.0, Philips Radiation Oncology Systems, Madison, WI). The beam apertures were adjusted based on volume change of targets and OARs volume contoured based on second CT scans. Every beam segment was loaded to Eclipse treatment plan system (Varian Medical Systems, Palo Alto, CA) to calculate dose distribution of individual beam separately. The original planning dosimetric restrictions were recalculated as individual cost functions and their weighted summation formed an overall cost function. A Limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) algorithm was applied to optimize the weighting factor of every segment by minimizing the overall cost function. RESULTS Data of three patients were used. Three plans with 2 to 4 arcs (178 ∼ 256 beams) were re-planned. Although 15 to 17 cost functions were involved, high quality of VMAT plans were obtained within a short time. CONCLUSIONS This initial research demonstrates the feasibility of on-line VMAT re-planning for adaptive radiotherapy. Though this initial work focuses on optimization of arc beams weight, it can be intergraded with on-board imaging and GPU calculating technique for online adaptive radiotherapy system which include automatic reconstruction of on-board images, deformable image registration, dose evaluation, apertures adjustment, dose calculation, re-optimization, and MLC sequencing. This research is supported by CPRIT Individual Investigator Award RP110329.

SU-C-103-04: 3D Tumor Tracking On Vero

PURPOSE The unique gimbal system of the BrainLAB Vero allows the treatment head to pivot in two dimensions for tumor tracking. In order to utilize this feature, it is crucial to directly monitor tumor motion during treatment. We present a 3D tumor motion monitoring method incorporating the orthogonal kV imaging subsystems on Vero. METHODS An anthropomorphic phantom was used in this study. A lung tumor and two BBs were driven in a regular 3D sine motion pattern using a programmable 4D motion platform. 4D CT scans were performed and the tumor was contoured at the end of inhale phase. Stereoscopic kV images were acquired at a rate of 5 fps. A spherical template was created to match the projections of BBs. Two sets of DRRs were created: one DRR set containing the tumor only and the other set containing the full anatomy without the tumor. They were combined to produce composite DRRs and their individual positions were optimized by comparing of the composite DRRs with acquired kV images. 3D tumor and BB positions were then determined from the results of stereo kV x-rays. RESULTS Dual orthogonal kV images were acquired with the MV beam at gantry angles of 0 and 45 degrees. This produced approximately 640 images with corresponding kV imaging angles of - 315, 45, 0, and 90 degrees. The tumor and the BBs were identified on every kV image. Both 2D and 3D positions were compared with the programmed positions. The maximum differences were 1 mm. CONCLUSION The dual kV imaging systems on Vero, used in conjunction with our tumor tracking algorithm, are capable of marker or marker-less lung tumor tracking. This technique may be applied to track tumors in real-time to provide direct tumor motion information for compensation of tumor motion throughout the course of radiotherapy.