T Chen

Improvement in Interobserver Accuracy in Delineation of the Lumpectomy Cavity Using Fiducial Markers

PURPOSE To determine, whether the presence of gold fiducial markers would improve the inter- and intraphysician accuracy in the delineation of the surgical cavity compared with a matched group of patients who did not receive gold fiducial markers in the setting of accelerated partial-breast irradiation (APBI). METHODS AND MATERIALS Planning CT images of 22 lumpectomy cavities were reviewed in a cohort of 22 patients; 11 patients received four to six gold fiducial markers placed at the time of surgery. Three physicians categorized the seroma cavity according to cavity visualization score criteria and delineated each of the 22 seroma cavities and the clinical target volume. Distance between centers of mass, percentage overlap, and average surface distance for all patients were assessed. RESULTS The mean seroma volume was 36.9 cm(3) and 34.2 cm(3) for fiducial patients and non-fiducial patients, respectively (p = ns). Fiducial markers improved the mean cavity visualization score, to 3.6 ± 1.0 from 2.5 ± 1.3 (p < 0.05). The mean distance between centers of mass, average surface distance, and percentage overlap for the seroma and clinical target volume were significantly improved in the fiducial marker patients as compared with the non-fiducial marker patients (p < 0.001). CONCLUSIONS The placement of gold fiducial markers placed at the time of lumpectomy improves interphysician identification and delineation of the seroma cavity and clinical target volume. This has implications in radiotherapy treatment planning for accelerated partial-breast irradiation and for boost after whole-breast irradiation.

3D Meshless Prostate Segmentation and Registration in Image Guided Radiotherapy

Image Guided Radiation Therapy (IGRT) improves radiation therapy for prostate cancer by facilitating precise radiation dose coverage of the object of interest, and minimizing dose to adjacent normal organs. In an effort to optimize IGRT, we developed a fast segmentation-registration-segmentation framework to accurately and efficiently delineate the clinically critical objects in Cone Beam CT images obtained during radiation treatment. The proposed framework started with deformable models automatically segmenting the prostate, bladder, and rectum in planning CT images. All models were built around seed points and involved in the CT image under the influence of image features using the level set formulation. The deformable models were then converted into meshless point sets and underwent a 3D non rigid registration from the planning CT to the treatment CBCT. The motion of deformable models during the registration was constrained by the global shape prior on the target surface during the deformation. The meshless formulation provided a convenient interface between deformable models and the image feature based registration method. The final registered deformable models in the CBCT domain were further refined using the interaction between objects and other available image features. The segmentation results for 15 data sets has been included in the validation study, compared with manual segmentations by a radiation oncologist. The automatic segmentation results achieved a satisfactory convergence with manual segmentations and met the speed requirement for on line IGRT.

Parameterization of brachytherapy source phase space file for Monte Carlo-based clinical brachytherapy dose calculation

A common approach to implementing the Monte Carlo method for the calculation of brachytherapy radiation dose deposition is to use a phase space file containing information on particles emitted from a brachytherapy source. However, the loading of the phase space file during the dose calculation consumes a large amount of computer random access memory, imposing a higher requirement for computer hardware. In this study, we propose a method to parameterize the information (e.g., particle location, direction and energy) stored in the phase space file by using several probability distributions. This method was implemented for dose calculations of a commercial Ir-192 high dose rate source. Dose calculation accuracy of the parameterized source was compared to the results observed using the full phase space file in a simple water phantom and in a clinical breast cancer case. The results showed the parameterized source at a size of 200 kB was as accurate as the phase space file represented source of 1.1 GB. By using the parameterized source representation, a compact Monte Carlo job can be designed, which allows an easy setup for parallel computing in brachytherapy planning.

Objected constrained registration and manifold learning: A new patient setup approach in image guided radiation therapy of thoracic cancer

PURPOSE The management of thoracic malignancies with radiation therapy is complicated by continuous target motion. In this study, a real time motion analysis approach is proposed to improve the accuracy of patient setup. METHODS For 11 lung cancer patients a long training fluoroscopy was acquired before the first treatment, and multiple short testing fluoroscopies were acquired weekly at the pretreatment patient setup of image guided radiotherapy (IGRT). The data analysis consisted of three steps: first a 4D target motion model was constructed from 4DCT and projected to the training fluoroscopy through deformable registration. Then the manifold learning method was used to construct a 2D subspace based on the target motion (kinetic) and location (static) information in the training fluoroscopy. Thereafter the respiratory phase in the testing fluoroscopy was determined by finding its location in the subspace. Finally, the phase determined testing fluoroscopy was registered to the corresponding 4DCT to derive the pretreatment patient position adjustment for the IGRT. The method was tested on clinical image sets and numerical phantoms. RESULTS The registration successfully reconstructed the 4D motion model with over 98% volume similarity in 4DCT, and over 95% area similarity in the training fluoroscopy. The machine learning method derived the phase values in over 98% and 93% test images of the phantom and patient images, respectively, with less than 3% phase error. The setup approach achieved an average accumulated setup error less than 1.7 mm in the cranial-caudal direction and less than 1 mm in the transverse plane. All results were validated against the ground truth of manual delineations by an experienced radiation oncologist. The expected total time for the pretreatment setup analysis was less than 10 s. CONCLUSIONS By combining the registration and machine learning, the proposed approach has the potential to improve the accuracy of pretreatment setup for patients with thoracic malignancy.

TH‐C‐WAB‐07: Optimization of Heart Block in the Left‐Sided Whole Breast Irradiation

PURPOSE Blocks have been used to protect heart from potential radiation damage in left breast treatments. Since cardiac motion pattern may not be fully captured on co nventional 3DCT or 4DCT simulation scans, this study was intended to investigate the optimization of the heart block design taking the cardiac motion into co nsideration. METHODS Whole breast treatment plans using two opposed tangential fields were designed based on 4DCT simulation images for 10 left breast patients. Using an OBI system equipped to a Varian Linac, beam-eye viewed fluorosco py images were acquired for each of the treatment beams after patient treatment setup, and the MLC heart blocks were overlaid onto the fluorosco py images with an in-house software package. A non-rigid image registration and tracking algorithm was utilized to track the cardiac motion on the fluorosco py images with minimal manual delineation, and the tracked cardiac motion information was used to optimize the heart block design to minimize the radiation damage to heart while avoiding the over-shielding that may lead to underdosing breast tissues. RESULTS Heart moved under the influences of both respiratory and cardiac motion. For 4 out of 10 patients, it was observed that heart moved beyond the heart block into treatment fields. The average size of the heart portion irradiated inside the treatment fields was 0.4 mm (0.4mm-0.4 mm), 3.6 mm (0.8 mm - 5.1 mm), 4.2 mm (0.4mm - 6.3mm), and 3.9 mm (0.2 mm-5.7mm) for the four patients, respectively. The optimization process changed the positions of 1, 8, 10, and 5 MLC leaves (5 mm wide per leaf) respectively to achieve the optimal heart protection. Co nclusion: Simulation 4DCT based heart block design may not provide adequate heart protection for all the treatments. A fluorosco py-based method has been developed to adaptively optimize the heart MLC block to achieve optimal heart protection.

SU‐EE‐A3‐05: Real Time Tumor Motion Monitoring during IGRT Based on Manifold Learning and Dynamic Registration between 4DCT and Fluoroscopy

Purpose: To develop a real time tumor respiratory motion monitoring methodology for thoracic cancerIGRT by integrating manifold learning and dynamical registration of 4DCT and fluoroscopy technique during the radiation treatment.Method and Materials: 10‐phases retrospective 4DCT have been acquired from 6 patients (5 lung, 1 pancreas) with audio coaching. During their gated radiation treatment,fluoroscopy videos were collected with the same audio coaching. A radiation oncologist delineated the contours including the GTV in the 50% phase 4DCT. The methodology is divided into three steps. First an object‐constrained registration has been used to register 4DCT at other phases to the reference 50% phase in order to build a spatial‐temporal tumor motion model. Then the 4D motion model has been used as the constraint in the spatial registration and temporal matching of the 4DCT with the fluoroscopy. An Isomap based manifold learning has been used to establish a global matching scheme between the 4DCT phases and fluoroscopy frames. In the last step, the GTV in newly acquired fluoroscopy frames were monitored in real time based on the fast registration between the frame and the DRR of the corresponding 4DCT phase using the global constraint and local image information.Results: The registration between 4DCT phases is within 30 minutes. The 4DCT to fluoroscopy registration and tumor motion monitoring are close to real time. After registration and training, for all 6 patients, the tumor (GTV) in an arbitrary fluoroscopy frame can be tracked with motion error less than 2mm, based on comparison between manually delineated contours in the fluoroscopy and the deformed GTV. Conclusion: The proposed tumor monitoring approach can achieve real time efficiency as well as satisfactory accuracy and precision and has the potential to be used in IGRT to improve setup and tumor tracking accuracy to achieve desired dose coverage.

SU‐E‐T‐514: Building a Brachytherapy Parallel Monte Carlo Simulation Framework Through Source Parameterization

Purpose: This work aimed to parameterize a brachytherapy source phase space file by using multiple probability density functions. Based on this, we built a brachytherapy parallel Monte Carlo (MC) simulation framework with simple geometry input, light memory consumption, accurate dose calculation, and fast speed. Methods: A Varian VariSource HDR brachytherapy source (Ir‐192) was generated in Geant4. A phase space file with 7 indexes for each particle (3 positions, 3 directions, 1 energy) was scored at the source surface. A parameterization was performed on the phase space file with the following assumptions: (1) A particles energy is independent of its position and direction; (2) Particle locations in the transverse plane (x‐y) are uniform; (3) A particles direction depends on its position along the source axis (z). Parameterization accuracy was evaluated by comparing radial dose function and anisotropic function as defined in TG‐43 between the phase space file‐represented source (S_phase) and the parameterized source (S_para). A brachytherapy MC framework was designed by parallel computing the dose contributions from individual dwell positions using the parameterized source. Results: The phase space file was successfully parameterized. Compared to the 2 GB phase space file, the parameterized source input was 300 kB without sacrificing computation speed. The radial dose function difference between S_phase and S_para was at most 2%. The anisotropic function agreed well between the two sources. The largest difference of about 5% was only observed within 10 degrees of the source axis. This difference has no significant impact on clinical applications. Conclusion: A method is developed to parameterize the source phase space file for MC brachytherapy dose calculation at almost no expense of accuracy. With much less memory consumption, using a parameterized source allows parallel computing for clinical implementation of brachytherapy MC dose calculation.

SU‐E‐J‐149: Heart Protection in Proton Therapy Using 4D Motion Analysis Based On Registration Between SENSE MRI and 4DCT

Purpose: Proton treatment on breast cancer requires optimal block on surrounding OARs including the heart to avoid radiation complications. The radiation dose to the heart is difficult to quantify since the heart moves under the influence of the respiratory motion and the cardiac motion; and as part of the cardiac motion the myocardium around the left ventricle rotates during the cardiac cycle. This study is to investigate the actual heart dose distribution by generating a 4D heart motion model through registration between SENSE MRI and planning CT. Methods: SENSE MRI has been utilized to acquire ECG gated 4D heart images for 4 patients. The slice thickness is 8mm, and the in plane spatial resolution is 1.4mm. 20 phases of MRI have been reconstructed to cover the whole cardiac cycle. The short axis (SA) images have been used to generate a 4D motion model through non‐rigid registration between phases. Anatomic land markers such as the papillary muscles and LV‐RV junctions have been used to constrain the registration to quantify the myocardial rotation. The reference MRI frame (end of systole) was registered to the treatment planning CT through model based meshless registration to translate the 3D dose distribution into the MRI image space and then the heart dose at different cardiac phases were derived based on registration results. Plan‐optimization was conducted to achieve optimal heart block. Results: The phase to phase registration demonstrated accuracy in the reconstructed motion. The average distance error at land markers is less than 1.5mm for two patients. The MRI‐to CT registration also has a satisfactory performance but still has the potential to be improved. A 4D dose reconstruction is possible given all the registration results. Conclusion: Based on the registration, 4D heart dose distribution can be generated to enable plan optimization for proton radiotherapy on breast cancer.

SU‐E‐J‐132: Lung Dose Reduction From Patient Specific 4D Motion Based Non‐Uniform Dose Prescription in Lung IMRT Treatment

Purpose: Motion of the tumor target in lung can be significant in patients. Traditionally lung dose is uniformly prescribed to the entire Internal Tumor Volume (ITV) with certain setup margin. Such uniform dose prescription may deliver quite high dose to the normal tissue surrounding the target volume due to tumor motion. This study explores patient specific motion based non‐uniform dose prescription that is to deliver prescription dose to the target while maximizing the sparing of the surrounding normal tissue. Methods: Patient 4D CT images were collected during simulation. CT phases were deformably registered. The spatial tracks of the tumor and surrounding tissue voxels were extracted from the transform matrix. The voxel dose to the tumor target and the surrounding tissue was evaluated at each phase and then summed together. Simulated Annealing algorithm was used to search for the spatial dose prescription distribution that delivers sufficient tumor voxel dose while minimizing the surrounding normal tissue dose. Results: Simple phantom and patients studies were included. The optimization process generally converged within 60 seconds to obtain a solution for non‐uniform dose prescription. The Result has shown a reduction in the adjacent lung tissue as great as 18% while maintaining the tumor dose. Conclusion: This study proposed a methodology for a non‐uniform dose prescription that would deliver sufficient dose to the moving lung tumor target while maximizing the sparing of the surrounding lung tissue. The beam delivery and the interplay effect are being further studied.

SU‐E‐T‐238: OnedosePlus MOSFET Dosimeter Performance in Tangential Field in Vivo Dosimtery

Purpose: To investigate the incident angle dependence of the OnedosePlus MOSFETdosimeter and to evaluate the dosimeters performance in vivo dosimetry for tangential fields treatments. Methods: Two types of external beam radiation plan were generated based on a solid water phantom: one with a group of oblique beams directed toward the anterior surface of the phantom and other with two opposing half blocked beams directed parallel to the anterior surface. The second plan represented an extreme case of tangential field treatments. In the deliveries of the plans, OnedosePlus dosimeters were placed at various locations on the phantom anterior surface. Calibrated Radiochromic EBT2 film dosimeters were placed at the same locations of the phantom anterior surface and at the corresponding locations 1cm away from the anterior surface. The measureddoses from the OnedosePlus dosimeters were compared against the corresponding doses in the plans and the film measurements. Results: As the incident beam angle increased from 0 to 90 degree to the CAX, the OnedosePlus reading increased 15%. The dosesmeasured using the OnedosePlus dosimeters in the extreme tangential field plan did not have a clear correspondence to either the surface dose or the doses at 1 cm depth. The film readings were in close agreement with the TPS calculated doses with an average dose difference less than 5%. The OnedosePlus measureddoses were higher than the TPS calculated doses at the 1cm depth by a constant (average: 31.45+/− 2.71cGy), and much higher than the surface film measurements by an average of 246%. Conclusions: OnedosePlus dosimeter readings have a strong dependence on the incident beam angle, especially when the beam angle is larger than 45 degree. The point of measurement for OnedosePlus dosimeter is unclear for the treatments involving tangential beams and it overestimates doses for the assumed point of measurement (1cm depth).