L Dawson

Accuracy of finite element model-based multi-organ deformable image registration.

As more pretreatment imaging becomes integrated into the treatment planning process and full three-dimensional image-guidance becomes part of the treatment delivery the need for a deformable image registration technique becomes more apparent. A novel finite element model-based multi-organ deformable image registration method, MORFEUS, has been developed. The basis of this method is twofold: first, individual organ deformation can be accurately modeled by deforming the surface of the organ at one instance into the surface of the organ at another instance and assigning the material properties that allow the internal structures to be accurately deformed into the secondary position and second, multi-organ deformable alignment can be achieved by explicitly defining the deformation of a subset of organs and assigning surface interfaces between organs. The feasibility and accuracy of the method was tested on MR thoracic and abdominal images of healthy volunteers at inhale and exhale. For the thoracic cases, the lungs and external surface were explicitly deformed and the breasts were implicitly deformed based on its relation to the lung and external surface. For the abdominal cases, the liver, spleen, and external surface were explicitly deformed and the stomach and kidneys were implicitly deformed. The average accuracy (average absolute error) of the lung and liver deformation, determined by tracking visible bifurcations, was 0.19 (s.d.: 0.09), 0.28 (s.d.: 0.12) and 0.17(s.d.:0.07)cm, in the LR, AP, and IS directions, respectively. The average accuracy of implicitly deformed organs was 0.11 (s.d.: 0.11), 0.13 (s.d.: 0.12), and 0.08(s.d.:0.09)cm, in the LR, AP, and IS directions, respectively. The average vector magnitude of the accuracy was 0.44(s.d.:0.20)cm for the lung and liver deformation and 0.24(s.d.:0.18)cm for the implicitly deformed organs. The two main processes, explicit deformation of the selected organs and finite element analysis calculations, require less than 120 and 495s, respectively. This platform can facilitate the integration of deformable image registration into online image guidance procedures, dose calculations, and tissue response monitoring as well as performing multi-modality image registration for purposes of treatment planning.

Adapting liver motion models using a navigator channel technique

Deformable registration can improve the accuracy of tumor targeting; however for online applications, efficiency as well as accuracy is important. A navigator channel technique has been developed to combine a biomechanical model-based deformable registration algorithm with a population motion model and patient specific motion information to perform fast deformable registration for application in image-guided radiation therapy. A respiratory population-based liver motion model was generated from breath-hold CT data sets of ten patients using a finite element model as a framework. The population model provides a biomechanical reference template of the average liver motions, which were found to be (absolute mean +/-SD) 0.12 +/- 0.10, 0.84 +/- 0.13, and 1.24 +/- 0.18 cm in the left-right (LR), anterior-posterior (AP), and superior-inferior (SI) directions, respectively. The population motion model was then adapted to the specific liver motion of 13 patients based on their exhale and inhale CT images. The patient motion was calculated using a navigator channel (a narrow region of interest window) on liver boundaries in the images. The absolute average accuracy of the navigator channel to predict the 1D SI and AP motions of the liver was less than 0.11, which is less than the out-of-plane image voxel size, 0.25 cm. This 1D information was then used to adapt the 4D population motion model in the SI and AP directions to predict the patient specific liver motion. The absolute average residual error of the navigator channel technique to adapt the population motion to the patients specific motion was verified using three verification methods: (1) vessel bifurcation, (2) tumor center of mass, and (3) MORFEUS deformable algorithm. All three verification methods showed statistically similar results where the techniques accuracy was approximately on the order of the voxel image sizes. This method has potential applications in online assessment of motion at the time of treatment to improve image-guided radiotherapy and monitoring of intrafraction motion.

SU‐EE‐A3‐04: Deformable Registration for In Vivo Imaging and Pathology Correlation

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.

MO‐F‐108‐01: Deformable Dose Reconstruction to Evaluate Image‐Guidance Strategies in Free‐Breathing Liver SBRT

PURPOSEnTo evaluate the delivered dose following image-guidance strategies for liver SBRT.nnnMETHODSnThe delivered dose was reconstructed using deformable image-registration (DIR) on the retrospectively sorted (4D) cone-beam CT (CBCT) for 30 SBRT patients (15 with multiple tumors). Plans created on exhale 4D-CT, for 27-60 Gy/6 fractions, were delivered in free-breathing after 3D-CBCT rigid liver alignment. Dose was also reconstructed after simulating two rigid alignment strategies on the exhale 4D-CBCT: the liver or the tumor directly (using DIR to predict tumor position). Delivered doses were compared to the planned 4D breathing dose, modeled with DIR of 4D-CT.nnnRESULTSnResidual mean tumor errors >3 mm occurred in nine patients (30%) following free-breathing 3D-CBCT (max: 10 mm), and two patients (6.6%) following liver or tumor 4D-CBCT alignment (maximum: 3.3 mm). With free-breathing 3D-CBCT, the delivered minimum tumor doses decreased by more than 1 Gy compared to planned dose for 4 (13%) patients. For 2 of these 4 patients, aligning the liver or tumor on 4D-CBCT reduced the dose decreases from -1.4 Gy, to within -0.5 Gy. The other 2 patients had multiple tumors and substantial liver deformation, resulting in minimum tumor doses decreases (max: -4.3 Gy) following free-breathing 3D-CBCT, that were reduced with either liver (max: -2.7 Gy) or tumor (max: -2.2 Gy) 4D-CBCT alignment. For normal gastrointestinal tissues receiving >30 Gy, the delivered maximum doses deviated by -11.5 to 2.6 Gy, with one exceeding the planning constraint by 0.7 Gy following free-breathing 3D-CBCT. Aligning the liver or tumor on 4D CBCT reduced normal tissue deviations (range: -6.3, 2.7 Gy) without exceeding planning constraints.nnnCONCLUSIONn4D-CBCT guidance for liver SBRT has been clinically implemented. The improved correlation between the 4D planning dose and delivered dose can be largely accomplished with 4D alignment of the liver, and should allow for planning margin reduction. This research is supported by the NIH, 5RO1CA124714-02, and a Canadian Institutes for Health Research Fellowship. Patient data was acquired during clinical trials supported by the National Cancer Institute of Canada, #18207, and CIHR, #202477. K.K. Brock has financial interest in the deformable registration technology through a licensing agreement with RaySearch Laboratories.

SU-E-T-640: Development of Liver SBRT Planning and Delivery at the Mean Respiratory Position Using Deformable Image Registration

PURPOSEnDeformable image registration (DIR) allows modeling of liver tumors on respiratory correlated (4D) imaging. The mid-position CT was reconstructed for liver SBRT plans using DIR, and the potential for dose-escalation was investigated.nnnMETHODSnThirty patients were planned clinically with IMRT for 27-48 Gy in 6 fractions on static exhale 4DCT with PTVs encompassing the breathing amplitude. For research, exhale 4DCT was deformed to the inhale 4DCT using biomechanical DIR. The mid-position CT was created by applying a percentage (the time-averaged normalized position between exhale and inhale calculated from daily 4D cone-beam CT) to this deformation map, assuming a linear trajectory. A probability-based PTV margin, using patient-specific breathing amplitude from DIR of 4DCT, was created around the GTV on the mid-position CT where IMRT was re-optimized. Dose was maximally escalated according to clinical protocol (e.g. liver NTCP <5%). The 4D predicted breathing dose was accumulated by interpolating the elements positions at exhale, mid-position and inhale onto the respective dose matrices (weighted by time spent nearest each matrix) then summed.nnnRESULTSnCompared the exhale plans, the GTV-to-PTV volume decreased on the mid-position plans by a mean of 31% (p<0.01, range: 24-38%). Static re-planning on the mid-position CT decreased the mean effective liver volume by 7% (p=0.032), enabling escalation of the nominal prescribed dose in 80% of patients of 6-12 Gy. Reconstruction of the 4D predicted breathing dose resulted in a mean increase of 6.7 Gy (p<0.01, maximum increase of 15.0 Gy) in mean GTV dose for the mid-position versus the exhale plan. For the mid-position plan, the minimum 0.5 cm3 GTV dose received 100% of the prescription in the 4D distribution.nnnCONCLUSIONSnLiver SBRT Planning at the mean respiratory position enables PTV reduction and a mean dose escalation of 6.7 Gy, potentially improving local control. Dr. Brock has financial interest in deformable registration technology through the licensing of Morfeus to RaySearch Laboratories. Research is funded by NIH 1R01CA124714.

TU‐C‐ValA‐05: Assessment of a Model‐Based Deformable Image Registration Approach for Radiotherapy Planning

Purpose: To assess the accuracy of a surface‐based deformable image registration strategy as a function of the elasticitymodel for the integration of multi‐modality imaging,image‐guidedradiation therapy, and quantification of geometrical change during and following therapy. Method and Materials: A surface‐model based deformable image registration system has been developed that enables quantitative description of geometrical change in multi‐modal images. Based on the deformation of organ surfaces represented by triangular surface meshes, a volumetric deformation field is derived using different volumetric elasticitymodels (Thin‐Plate Splines, Wendland functions, Elastic Body Splines) as alternatives to finite‐element modeling.Results: The system was demonstrated on five livercancer patients, ten prostate cancer patients, thorax in five healthy volunteers, and abdomen in five healthy volunteers. The accuracy of the system was assessed by tracking visible fiducials (bronchial bifurcations in the lung, vessel bifurcations in the liver, implanted gold markers in the prostate). The maximum displacements for lung,liver and prostate were 5.3 cm, 3.2 cm, and 1.8 cm respectively. The largest registration error (direction, mean ± standard deviation) for lung,liver and prostate were (inferior‐superior, −0.21 ± 0.38 cm), (anterior‐posterior, −0.09 ± 0.34 cm), and (left‐right, 0.04 ± 0.38 cm) respectively, which was within the image resolution regardless of the deformation model. The computation time (2.7 GHz Intel Xeon) was on the order of seconds (e.g. 10 seconds for two prostate data sets), and image deformation results could be viewed at interactive speed (less than 1 second for 512×512 voxels). Conclusion: Surface‐based deformable image registration enables the quantification of geometrical change in normal tissue and tumor with acceptable accuracy and speed.

Sci‐AM2 Sat ‐ 08: Impact of volume definition on prescribed dose in a liver cancer dose escalation study

The volume of irradiated liver is strongly related to target volume size in extracranial stereotactic radiation therapy (ESRT) of livertumours. Our ESRT dose escalation study treatslivercancer with prescription doses that are individualized to maintain a constant risk of radiation induced liver disease (RILD) for all patients. The impact of liver volume on normal tissue complication probability (NTCP) calculation was assessed using the whole liver,liver minus gross tumour volume (GTV), liver minus clinical target volume (CTV), and liver minus planning target volume (PTV). Assuming that liver minus GTV is the most appropriate volume to base NTCP on (since it includes all potentially functional liver), each volume was used to calculate the dose for 5% NTCP. NTCP was then recalculated using the liver minus GTV, but with doses determined from the other liver volumes. The relationship between target volume size and dose is also investigated. NTCP calculated with liver minus CTV or liver minus PTV results in extremely high risks of RILD, while using whole liver underestimates the risk and is safer. For all target volumes, the prescription dose can be increased as the target size decreases. Predicted dose for individualized dose escalation for livercancer is strongly dependent on the liver volume analyzed. We suggest that liver minus CTV and liver minus PTV volumes cannot safely be used to individualize prescription doses for dose escalation for livercancer. Based on these substantial changes in NTCP, uniform reporting of volumes and NTCP is desirable.

SU‐DD‐A4‐03: Spatial and Volumetric Comparison of Liver Tumors On CT and MR Using Finite Element Based Deformable Image Registration

Purpose: To compare the spatial and volumetric differences in livertumor definition based on triphasic IV contrastCT and Gadoliniumcontrast MR. Method and Materials: Changes in patient and liver position and differences in CT and MR imaging make an accurate comparison of tumor representation challenging, as deformation of the liver occurs. A deformable registration method has been developed to generate geometrically resolved views of the different image sets using finite element modeling to take advantage of the biomechanical relationship between the anatomical representations on each modality. The liver and livertumors are contoured on the CT and MR, using a treatment planning system. A guided surface projection method is used to identify and account for the geometric discrepancies of the liver, allowing a direct comparison of the MR and CTtumors.Results: Triphasic CT and Gadolinium MR images were obtained at end exhale. Differences in tumor volume and center of mass and in the surface area differing by greater than 3 mm in spatial position (PSA3) were measured for patients with liver metastases, cholangiocarcinoma and hepatoma. The tumor comparisons were done following registration of the liver to resolve geometric discrepancies. The average tumor volume change was 52% (19–121%). The average center of mass change was 0.4 cm (0.1–0.6 cm). The average PSA3 was 49% (28–60%). Initial results show that MR GTV was larger then CT GTV in patients with hepatoma, smaller in cholangiocarcinoma and variable in metastases. Conclusion: Deformable image registration improves the spatial correlation of the MR and CT defined liver GTVs compared to rigid body registration alone. This deformable image registration method preserves the spatial integrity of the tumor, while resolving the geometric differences between the MR and CTliver.Conflict of Interest: Research supported in part by Varian Medical Systems. LD is an ASCO career development award recipient.

SU‐FF‐J‐78: Evaluation of Setup Accuracy for Hypofractionated Radiotherapy of Liver Using Portal Imaging and On‐Line KV Fluoroscopy

Purpose: To compare image‐guided accuracy for stereotactic radiotherapy of the liver in terms of reproducibility and targeting confidence. Method and Materials: The study involved eight patients treated for inoperable liver metastases or hepatobiliary carcinoma. Each treatment fraction, these patients were imaged and repositioned using orthogonal portal images acquired in AP and lateral views. Repositioning occurred when the measured offset exceed 3mm in any of the cranio‐caudal (CC), anterior‐posterior (AP), and medio‐lateral (ML) directions. A second pair of orthogonal images assessed the residual setup accuracy. Orthogonal fluoroscopy sessions, each lasting 30 s, were acquired; some prior, but most after radiotherapy delivery. All images were acquired with the patient under active breathing control. All treatments were performed using a conventional medicallinear accelerator equipped with a kilovoltage x‐ray source and flat‐panel detector mounted at 90 degrees from the linac beam central axis. The right diaphragm was used as a surrogate for the liver position. The reproducibility and stability of the diaphragm was thus analysed for a total of 48 fractions. Results: After initial setup and repositioning, the residual setup error determined from portal images was 2.6 (CC), 3.0 (AP), and 2.6 mm (ML); conversely, the residual setup error measured from the kilovoltage fluoroscopy sessions was 3.2 (CC), 3.1 (AP), and 2.1 mm (ML). Conclusion: Daily portal imaging and repositioning based on these images improves setup accuracy. Similar accuracy is shown from kilovoltage fluoroscopy analysis. On‐line fluoroscopy can potentially assess patient position prior and after delivery of radiation therapy, and provide a baseline for cone‐beam CT analysis. Conflict of Interest:: This work is supported, in part, by Elekta Radiation Oncology. LD is supported by an ASCO career development award.