S Gaudio

A novel fast helical 4D-CT acquisition technique to generate low-noise sorting artifact-free images at user-selected breathing phases.

PURPOSE To develop a novel 4-dimensional computed tomography (4D-CT) technique that exploits standard fast helical acquisition, a simultaneous breathing surrogate measurement, deformable image registration, and a breathing motion model to remove sorting artifacts. METHODS AND MATERIALS Ten patients were imaged under free-breathing conditions 25 successive times in alternating directions with a 64-slice CT scanner using a low-dose fast helical protocol. An abdominal bellows was used as a breathing surrogate. Deformable registration was used to register the first image (defined as the reference image) to the subsequent 24 segmented images. Voxel-specific motion model parameters were determined using a breathing motion model. The tissue locations predicted by the motion model in the 25 images were compared against the deformably registered tissue locations, allowing a model prediction error to be evaluated. A low-noise image was created by averaging the 25 images deformed to the first image geometry, reducing statistical image noise by a factor of 5. The motion model was used to deform the low-noise reference image to any user-selected breathing phase. A voxel-specific correction was applied to correct the Hounsfield units for lung parenchyma density as a function of lung air filling. RESULTS Images produced using the model at user-selected breathing phases did not suffer from sorting artifacts common to conventional 4D-CT protocols. The mean prediction error across all patients between the breathing motion model predictions and the measured lung tissue positions was determined to be 1.19 ± 0.37 mm. CONCLUSIONS The proposed technique can be used as a clinical 4D-CT technique. It is robust in the presence of irregular breathing and allows the entire imaging dose to contribute to the resulting image quality, providing sorting artifact-free images at a patient dose similar to or less than current 4D-CT techniques.

A comparison of amplitude-based and phase-based positron emission tomography gating algorithms for segmentation of internal target volumes of tumors subject to respiratory motion.

PURPOSE To quantitatively compare the accuracy of tumor volume segmentation in amplitude-based and phase-based respiratory gating algorithms in respiratory-correlated positron emission tomography (PET). METHODS AND MATERIALS List-mode fluorodeoxyglucose-PET data was acquired for 10 patients with a total of 12 fluorodeoxyglucose-avid tumors and 9 lymph nodes. Additionally, a phantom experiment was performed in which 4 plastic butyrate spheres with inner diameters ranging from 1 to 4 cm were imaged as they underwent 1-dimensional motion based on 2 measured patient breathing trajectories. PET list-mode data were gated into 8 bins using 2 amplitude-based (equal amplitude bins [A1] and equal counts per bin [A2]) and 2 temporal phase-based gating algorithms. Gated images were segmented using a commercially available gradient-based technique and a fixed 40% threshold of maximum uptake. Internal target volumes (ITVs) were generated by taking the union of all 8 contours per gated image. Segmented phantom ITVs were compared with their respective ground-truth ITVs, defined as the volume subtended by the tumor model positions covering 99% of breathing amplitude. Superior-inferior distances between sphere centroids in the end-inhale and end-exhale phases were also calculated. RESULTS Tumor ITVs from amplitude-based methods were significantly larger than those from temporal-based techniques (P=.002). For lymph nodes, A2 resulted in ITVs that were significantly larger than either of the temporal-based techniques (P<.0323). A1 produced the largest and most accurate ITVs for spheres with diameters of ≥2 cm (P=.002). No significant difference was shown between algorithms in the 1-cm sphere data set. For phantom spheres, amplitude-based methods recovered an average of 9.5% more motion displacement than temporal-based methods under regular breathing conditions and an average of 45.7% more in the presence of baseline drift (P<.001). CONCLUSIONS Target volumes in images generated from amplitude-based gating are larger and more accurate, at levels that are potentially clinically significant, compared with those from temporal phase-based gating.

WE‐A‐134‐08: Modeling Cardiac Induced Lung Tissue Motion for a Quantitative Breathing Motion Model

PURPOSE To improve the accuracy of a quantitative breathing motion model by developing a cardiac-induced lung tissue motion model from MRI data. METHODS 10 healthy volunteers were imaged on a 1.5T MR-scanner. A total of 24 short-axis and 18 radial views were acquired during a series of 12-15s breath-holds. The planar views were combined to create a 3D view of the anatomy. Each view contained 30 equal-partitioned frames beginning with the end-diastolic cardiac phase. A single-level 3D optical flow deformable image registration algorithm was used to measure the difference in tissue position between the end-diastolic image and the remaining phases. The maximum displacement magnitude and direction obtained in this manner was defined as g(X0 ), the cardiac-induced lung tissue motion. The motion model was assumed to be linear and the motion trajectory a product of g(X0 ) and h, where h was a phase-dependent scalar that had a value of 0 at end-diastole and 1 at the maximum tissue displacement phase. The model was evaluated by comparing the cardiac-induced lung tissue motion, using a lower motion threshold of 0.3mm, with the residual model error. RESULTS The deformable image registration algorithm was found to be highly accurate. Lung tissue near the myocardium was observed to have motion as large as 5mm. The average relative error for the model was 36.5% for sub-millimeter voxel motion. The average relative error decreased for greater voxel motion to 5.6% for >3mm voxel motion. The overall average model residual error was 0.19±0.18mm. CONCLUSION The magnitude of cardiac-induced lung tissue displacement was enough to degrade the accuracy of quantitative lung tissue motion modeling. The use of a single location-independent phase dependent term provided suitable model accuracy. Introducing a cardiac motion term has the potential to reduce the error in breathing motion models caused by uncompensated cardiac-induced lung tissue motion. This work supported in part by NIH R01CA096679 and R01CA116712.

TU‐C‐141‐06: Improving Image Quality in 4D‐CT Scans Using Deformable Registration and Selective Averaging

Purpose: To improve image quality in low‐dose 4D‐CT using a selective averaging algorithm that combines images acquired under free breathing and registered to a single breathing phase. Methods: Five patients were imaged with a low‐pitch helical protocol on a 64‐slice scanner during free patient breathing, as part of an IRB‐approved research protocol. 25 low dose scans were performed in order to image the lungs at varying breathing phases to generate a breathing motion model. The first scan was registered to the subsequent 24 using a b‐spline registration algorithm to produce 25 representations of a single patient scan geometry. These images were averaged using k‐means clustering (k=2 clusters) in conjunction with the arithmetic mean. A gradient mapping algorithm assigned the appropriate mean value to each voxel to produce a composite image. This algorithm calculates the image gradient at each voxel and assigns a cluster mean when the gradient magnitude is above a threshold, otherwise assigning the arithmetic mean. The composite image was compared with the first image in the set as well as the arithmetic mean of all 25 scans. Image noise was calculated in an axial region of the liver. Sharpness was measured as the sum of gradient magnitudes across the image. Results: By combining gradient selective k‐means clustering with the arithmetic mean, image noise was reduced by 78% while maintaining sharpness within 14.5% of the reference image. This improves on the arithmetic mean in sharpness by 14% with less than a 1% increase in noise. Conclusion: Results indicate that 4D‐CT image quality can be improved by co‐registration followed by a gradient selective averaging algorithm. This method greatly reduces image noise while maintaining sharpness that is lost in a simple averaging algorithm. This work supported in part by NIH R01CA096679

SU‐E‐J‐126: Development of a Prospective Gating Algorithm for a Novel 4DCT Technique: Retrospective Data Analysis

Purpose: To develop an algorithm for prospective CT scan selection to support a novel 4DCT acquisition techniqueMethods: The new 4DCT acquisition protocol utilizes repeated rapid helical CT scans. Unless the CT scanner is synchronized with breathing, scans may be acquired at similar breathing phases, adding little to the breathing‐motion characterization and therefore unnecessarily irradiating the patient. A retrospective patient dataset that consisted of 25 repeated CT scans and the accompanying breathing‐motion model was used to evaluate possible algorithms. A single coronal slice through the lungs was used for the preliminary analysis. The mean model discrepancy for the 25 scans was 0.63mm. The model was fit using every subset of N scans out of the first 15. The discrepancies ranged from nearly 0.68mm to 9mm. The variation was hypothesized to be due to the level of redundancy in the scans used to generate the motion models. Scans were evaluated based on how the breathing depth and breathing rate (the two independent variables in the breathing motion model) differed at each CT slice. To maximize the variety of breaths selected from the 15, the mean square root difference D of the breathing depths and breathing rates were calculated and D correlated against the model discrepancies. Results: The square‐root difference function provided excellent separation between scan combinations that had small and large model errors. As expected, scans sets with small model errors were well spaced in breathing depth and rate. An optimal combination of breathing depth and rate was identified and shown to better select good scan combinations relative to separately analyzing depth or rate. The optimal scans had residual errors of 0.68mm compared with the baseline of 0.63mm. Conclusion: The use of the mean square‐root difference in breathing depth and rate will provide guidance for prospective scanning algorithm development. Supported by NIH R01 CA096679

SU‐E‐J‐138: Breathing Motion Model Comparison Inside and Outside the Lung

Purpose: To compare the results of our deformable image registration algorithm inside and outside the lungs for our novel fast 4DCT technique in order to quantify the sliding motion between the lung and other organs in the thorax. Methods: : Data was acquired with a fast helical protocol. We used a 64‐slice CT scanner. 25 low dose whole‐lung scans were obtained in alternating directions for five patients. Images were obtained continuously under free breathing conditions. A pneumatic bellows was placed around the abdomen as a respiratory surrogate. Each image slice was assigned a unique bellows voltage signal. The lungs were segmented in each scan and, all scans were deformably registered, and were then deformed to the first scan geometry. The motion model was fit to the breathing surrogate to obtain the model parameters at each voxel inside the lung. Separately, all tissue outside the lungs was segmented, registered and fit to the motion model. Separate segmentation and registration of tissues inside and outside the lung boundary is necessary to properly treat the sliding motion that occurs at the interface of lung and chest wall. The object of this study was to determine if the separately treated inside and outside regions could be matched at the lung boundary. Results: Discontinuity in the superior‐inferior motion vectors was observed at the edges of the lungs, as expected, due to shear motion. Motion vectors in the left‐right and anterior‐posterior directions were continuous across the lung border, due to the lack of shear motion in those directions. Two masks representing the lung surface were deformed separately using the interior and the exterior motion models and found to be consistent. Conclusion: : We obtain a validation of the lung motion model and demonstrated that it can be used to reproduce accurate images for tissues outside the lung NIH grant number: ROI CA096679

SU‐D‐500‐05: Comparison of Gating Algorithms in 4D‐PET for Mobile Tumor Segmentation

PURPOSE To quantitatively compare the accuracy of tumor volume segmentation in four different gating algorithms in gated 4D-PET. METHODS Four acrylic spheres with inner diameters ranging from 1cm to 4cm were filled with a 11-C solution and affixed inside a cylindrical bath of 18-FDG. The system was attached to a robotic arm that underwent 1D motion according to large-amplitude trajectories based on measured patient breathing trajectories. Two trajectories were used: one with and one without baseline drift. List-mode data was split into two-minute images at different source-to-background ratios (SBRs), which were gated into eight bins using two amplitude-based (equal amplitude bins (A1) and equal counts per bin (A2)) and two temporal phased-based gating algorithms. All gated images were segmented using a commercially available gradient-based technique. Internal target volumes (ITVs), generated by taking the union of all eight contours per gated image, were compared to their respective ground truths. The ground-truth ITV was defined as the volume subtended by the tumor model positions covering 99% of breathing amplitude. Superior-inferior distances between sphere centroids in the end-inhale and end-exhale phases were also calculated. RESULTS Averaged over all sphere sizes, both trajectories, and high and low SBRs, A2 was the closest in accuracy of ITV segmentation, with a measured-to-expected ratio of 1.002 vs. 0.920 and 0.964 for temporal phase-based methods (p<0.05). A1 consistently recovered the greatest volume and had the highest accuracy in the presence of irregular breathing, while A2 performed the best in accurately representing the smallest (1cm) sphere. Amplitude-based methods consistently produced more distance between end-inhale and end-exhale phases than phase-based methods (27.5% more on average), with A1 recovering up to 94% of peak-to-peak separation (p<0.01). CONCLUSION Target volumes in images generated from amplitude-based gating are more accurate, potentially at clinically significant levels, than those from temporal phase-based gating. NIH R01 CA096679.

SU‐E‐J‐82: Ground‐Truth Tests of Deformable Image Registration Using Matched PET‐CT Image Pairs

PURPOSE To investigate the possibility of using tracer activity density measured in matched PET/CT scan pairs as a ground-truth test of deformable image registration and dose accumulation software. METHODS Respiratory-correlated (4D) PET/CT was used for this proof-of-concept study. Amplitude-gated 4D-CT scans of 2 lung cancer patients acquired as part of an IRB-approved protocol were studied. List-mode FDG-PET data was acquired in the same sessions and gated into 4D images (i.e., activity distributions) matched in phase to the 4D-CT. End-inhale phase 4D-CT images were deformably registered to the end-exhale phase images using a free-form intensity based algorithm provided by a commercial software package. The corresponding 4D-PET activity distributions at the same inhale breathing phase were deformed using the deformation vector fields from the 4D-CT registration. CT registration accuracy was then assessed by visually inspecting the deformed inhale-phase PET activity distributions against the exhale-phase activity distributions. Errors in registration were identified when the deformed inhale-phase activity distribution failed to match the exhale phase distribution. RESULTS To test the utility of this methodology deformed PET activity distributions were used to identify CT-CT registration failure in the region of the mediastinum. The registration failure could not be seen by inspection of the deformed CT itself, due to lack of contrast inside the mediastinum, but was clearly visible from the deformation of the activity distribution of a tracer-avid lymph node. Control cases were used for validation, consisting of correct and incorrect registrations of lung tumors near the chest wall, where the performance of the registration was visible from CT deformation alone. CONCLUSION PET-measured tracer activity in combined PET/CT scans can be used as a ground truth to qualitatively assess the accuracy of deformable image registration of CT images and deformable dose accumulation. Further work is needed to understand the limits of quantitative assessment. This work is supported in part by NIH R01CA096679 and R01CA116712.

TU‐G‐141‐01: BEST IN PHYSICS (JOINT IMAGING‐THERAPY)‐A Novel 4D CT Acquisition and Analysis Technique to Generate Low Noise Artifact‐Free Images at User Selected Breathing Phases

PURPOSE To develop a novel 4DCT technique that exploits standard fast helical acquisition, a simultaneous breathing surrogate measurement, deformable image registration, and a breathing motion model to remove motion-induced artifacts. METHODS Five patients were imaged under free breathing conditions 25 successive times in alternating directions with a 64-slice CT scanner using a low dose fast helical protocol. A pneumatic bellows around the abdomen was used to as a breathing surrogate. The lungs were segmented from each image. Deformable registration was used to register the first to the subsequent 24 segmented images. Voxel-based motion model parameters were determined using a published breathing motion model. A low-noise image was created by averaging the 25 images deformed to the first image geometry, reducing statistical noise by a factor of 5. The motion model was used to deform the low noise image to any user-selected breathing phase. Accurate HU values were assigned to each voxel in the reconstructed images. RESULTS Images produced using the model at user-selected breathing phases did not suffer from motion artifacts. The mean discrepancy between the breathing motion model results and the measured positions corresponding to each scan was determined to be 0.7mm (standard deviation of 0.4mm). In each patient, regions near to the myocardium exhibited mean discrepancies greater than 1 mm, which were likely due to uncompensated cardiac motion. CONCLUSION The proposed technique can be employed as a clinical 4DCT technique providing motion artifact free images at user-selected breathing phases. It is robust in the presence of irregular breathing, and allows the entire imaging dose to contribute to the resulting image quality, providing motion artifact free images at a patient dose similar to or less than current 4DCT techniques. We are currently modifying the protocol to work on 16-slice CT scanners. This work supported in part by NIH R01CA096679.

MO-F-WAB-07: A Novel 4D CT Acquisition and Analysis Technique to Account for the Effect of Cardiac Induced Lung Tissue Motion During Free Breathing

PURPOSE To develop a new motion modeling technique to account for internal lung tissue displacement due to cardiac motion during free breathing CT scans. METHODS Five patients were imaged 25 successive times under free breathing conditions in alternating directions with a 64-slice CT scanner using a low dose fast helical protocol. A pneumatic bellows around the abdomen was used to as a breathing surrogate and a 3-lead ECG monitor simultaneously measured heart rate. The lungs were segmented from each image and deformable registration was used to register the first image to the subsequent 24 segmented images. The registered voxel locations were fitted to a linear motion model, relating the internal lung tissue deformation to the tidal volume v, airflow f and cardiac phase h. The time dependence of cardiac induced tissue displacement was characterized by a periodic function synchronized to the cardiac cycle, optimized to reduce model error, and was patient specific. The magnitude of cardiac induced motion was evaluated by comparing the discrepancies between fitted and measured motion with and without the cardiac term applied. RESULTS The magnitude of cardiac induced lung tissue displacement was determined to be up to 2.5mm in the lung for regions close to the myocardium. The addition of the cardiac term reduced the total number of voxels with mean errors above 1mm by 33%. CONCLUSION Application of the cardiac term in the motion model reduces large errors in motion modeling in regions close to the myocardium. This work supported in part by NIH R01CA096679.