Juan Carlos Celi

Constrained surface evolutions for prostate and bladder segmentation in CT images

We propose a Bayesian formulation for coupled surface evolutions and apply it to the segmentation of the prostate and the bladder in CT images. This is of great interest to the radiotherapy treatment process, where an accurate contouring of the prostate and its neighboring organs is needed. A purely data based approach fails, because the prostate boundary is only partially visible. To resolve this issue, we define a Bayesian framework to impose a shape constraint on the prostate, while coupling its extraction with that of the bladder. Constraining the segmentation process makes the extraction of both organs’ shapes more stable and more accurate. We present some qualitative and quantitative results on a few data sets, validating the performance of the approach.

DOSE RECALCULATION AND THE DOSE-GUIDED RADIATION THERAPY (DGRT) PROCESS USING MEGAVOLTAGE CONE-BEAM CT

PURPOSE At the University of California San Francisco, daily or weekly three-dimensional images of patients in treatment position are acquired for image-guided radiation therapy. These images can be used for calculating the actual dose delivered to the patient during treatment. In this article, we present the process of performing dose recalculation on megavoltage cone-beam computed tomography images and discuss possible strategies for dose-guided radiation therapy (DGRT). MATERIALS AND METHODS A dedicated workstation has been developed to incorporate the necessary elements of DGRT. Patient image correction (cupping, missing data artifacts), calibration, completion, recontouring, and dose recalculation are all implemented in the workstation. Tools for dose comparison are also included. Examples of image correction and dose analysis using 6 head-and-neck and 2 prostate patient datasets are presented to show possible tracking of interfraction dosimetric endpoint variation over the course of treatment. RESULTS Analysis of the head-and-neck datasets shows that interfraction treatment doses vary compared with the planning dose for the organs at risk, with the mean parotid dose and spinal cord D(1) increasing by as much as 52% and 10%, respectively. Variation of the coverage to the target volumes was small, with an average D(5) dose difference of 1%. The prostate patient datasets revealed accurate dose coverage to the targeted prostate and varying interfraction dose distributions to the organs at risk. CONCLUSIONS An effective workflow for the clinical implementation of DGRT has been established. With these techniques in place, future clinical developments in adaptive radiation therapy through daily or weekly dosimetric measurements of treatment day images are possible.

Radiation therapy planning and simulation with magnetic resonance images

We present a system which allows for use of magnetic resonance (MR) images as primary RT workflow modality alone and no longer limits the user to computed tomography data for radiation therapy (RT) planning, simulation and patient localization. The single steps for achieving this goal are explained in detail. For planning two MR data sets, MR1 and MR2 are acquired sequentially. For MR1 a standardized Ultrashort TE (UTE) sequence is used enhancing bony anatomy. The sequence for MR2 is chosen to get optimal contrast for the target and the organs at risk for each individual patient. Both images are naturally in registration, neglecting elastic soft tissue deformations. The planning software first automatically extracts skin and bony anatomy from MR1. The user can semi-automatically delineate target structures and organs at risk based on MR1 or MR2, associate all segmentations with MR1 and create a plan in the coordinate system of MR1. Projections similar to digitally reconstructed radiographs (DRR) enhancing bony anatomy are calculated from the MR1 directly and can be used for iso-center definition and setup verification. Furthermore we present a method for creating a Pseudo-CT data set which assigns electron densities to the voxels of MR1 based on the skin and bone segmentations. The Pseudo-CT is then used for dose calculation. Results from first tests under clinical conditions show the feasibility of the completely MR based workflow in RT for necessary clinical cases. It needs to be investigated in how far geometrical distortions influence accuracy of MR-based RT planning.

MO‐E‐M100F‐04: Optimizing Low‐Dose Megavoltage Fluoroscopy for 4D Radiotherapy

Purpose: This study explores the feasibility of using megavoltage fluoroscopy (MVF) for on‐line real‐time verification and guidance for gated/4D treatment delivery. Initial experiences in implementing MVF and methods developed to minimize required dose and optimize the signal‐to‐noise ratio will be presented. Methods and Materials: A Siemens LINAC retrofit with a flat panel imager in the beam direction was used to acquire MVF images with 10242 pixel matrix at 7 frames‐per‐second (fps) using 6 MV photons.Image processing software tools were developed to remove artifacts caused by beam pulsing, dead pixels and stuck bits and to average pixels and video frames. The beam delivery rate was adjusted during MVF acquisition in an attempt to minimize the dose per frame. A phantom designed to test the contrast resolution of kilovoltage fluoroscopy systems was used to assess the MVF image quality at different beam delivery rates. Results: Good quality MVF images with sequences longer than a typical respiratory cycle (5 sec) were obtained with an estimated dose as low as 0.6 cGy. Matching the beam pulse frequency to a multiple of the video frame rate minimized striping artifacts. Image sequences acquired at a frame rate of 3.5 fps and beam delivery rate of 12 MU/min had an SNR of 125; the minimum object resolvable during cine display was 0.063 inch in diameter, and the 0.31 inch diameter object was detectable at 2% contrast.Contrast detectability was improved by averaging video frames, reducing spatial resolution and displaying video in high‐speed cine mode. Conclusions: It is feasible to acquire good quality megavoltage fluoroscopyimages using a low dose‐rate treatment beam, indicating that MVF can be used to verify or guide gated/4D treatment delivery in real time (e.g. prior or during treatment delivery). Conflict of Interest: This work was supported in part by Siemens OCS.

SU‐FF‐J‐28: A Software Tool for On‐Line Real‐Time Verification of Gated Delivery Using Megavoltage Fluoroscopy

Purpose: To test and to explore the application of a software tool developed for on‐line real‐time treatment verification for 4D/gated radiation delivery using megavoltage fluoroscopy (MVF) acquired with treatment beams. Method and Materials: MVF images were acquired with a flat panel detector installed on a Siemens Linac using 6MV photon beams for phantoms with moving structures and for thoracic cancer patients. A previously developed software, RTReg4D, was used to register acquired MVF with digitally reconstructed radiographs(DRRs) from 4DCT. The MVF frame grabber can be synchronized with the gating signal from Anzai gating device to capture target motion in selected gating window. The isocenter and target contours were transferred from a planning system into RTReg4D and were overlaid on the MVF images, verifying whether the target with residual motion is adequately covered during the gated delivery. Appropriate actions can be initiated should the verification fail. Results: The software tool can effectively and accurately register MVF images with DRRs. The MVF images acquired with a dose as low as 1 cGy were adequate for the registration. In addition, the registration can be performed between the averaged MVF frame of all frames acquired within gating window and the corresponding DRR, further reducing dose required. The accuracy of registration can be better than 3 mm. The entire process is fast. Conclusion: MVF achieves adequate image quality at a low dose. It is practically feasible to use the software tool with MVF for real‐time (prior or during) treatment verification of 4D/gated radiation delivery. Conflict of Interest: This work is sponsored in part by Siemens OCS.

SU‐FF‐J‐50: Dose‐Guided Radiation Therapy Strategies with Megavoltage Cone‐Beam CT

Dose Guided Radiation Therapy (DGRT) is an extension of adaptive radiation therapy where dosimetric considerations constitute the basis of treatment modification and validation. The objective of this work is to elaborate two DGRT scenarios, treatment‐time and progressive, and gather all the required functionalities within the same workstation to use it in a clinical environment. The proposed DGRT strategies use a density‐calibrated Megavoltage Cone‐Beam CT (MVCBCT) image acquired at treatment time, the reference dose plan imported from the planning system and an independent dose calculation engine. A DGRT workstation was created to compare the initial plan with the dose distribution of the day. The beam arrangement is copied to the pre‐registered MVCBCT and dose distribution is calculated. Side by side comparison, dose difference maps and DVH differences are used to assess the clinical situation. Phantom experiments and patient data were used to test the DGRT scenarios. A Pinnacle dose calculation was performed to validate the local dose calculation engine. The DGRT workstation integrates the following steps; image registration, correction and conversion of MVCBCT images into density, dose re‐calculation, dose comparison with the reference plan and a data extraction module for statistical analysis.Dose re‐calculation on MVCBCT images is performed with a precision better than 3%, enabling the progressive DGRT scenario. When performed on a patient treated for a base of tongue tumor with IMRT, the DGRT scenario showed a mean dose increase from 26 to 42 Gy in the left parotid. DGRT brings together the knowledge of 3D imaging and the 3D dose distribution to assess the clinical situation. By making dose re‐calculation fast and accurate, the DGRT workstation allows dosimetric studies on a large scale. Results from on‐going clinical studies will determine how frequently significant dose differences occur and lead to site‐specific adaptation strategies. Research sponsored by Siemens OCS.

Radial subsampling for fast cost function computation in intensity-based 3D image registration

Image registration is always a trade-off between accuracy and speed. Looking towards clinical scenarios the time for bringing two or more images into registration should be around a few seconds only. We present a new scheme for subsampling 3D-image data to allow for efficient computation of cost functions in intensity-based image registration. Starting from an arbitrary center point voxels are sampled along scan lines which do radially extend from the center point. We analyzed the characteristics of different cost functions computed on the sub-sampled data and compared them to known cost functions with respect to local optima. Results show the cost functions are smooth and give high peaks at the expected optima. Furthermore we investigated capture range of cost functions computed under the new subsampling scheme. Capture range was remarkably better for the new scheme compared to metrics using all voxels or different subsampling schemes and high registration accuracy was achieved as well. The most important result is the improvement in terms of speed making this scheme very interesting for clinical scenarios. We conclude using the new subsampling scheme intensity-based 3D image registration can be performed much faster than using other approaches while maintaining high accuracy. A variety of different extensions of the new approach is conceivable, e.g. non-regular distribution of the scan lines or not to let the scan lines start from a center point only, but from the surface of an organ model for example.

SU‐FF‐J‐48: Developing In‐Line KV Fluoroscopic Verification for 4D Adaptive Radiotherapy

Purpose: To develop a system for on‐line real‐time treatment verification for 4D‐ART using in‐line kV fluoroscopy that will be included in a new generation of accelerator. Method and Materials: We have developed a software tool, as a component of the 4D‐ART verification system, to register fluoroscopic with dynamic (time‐sequenced) DRR (DDRR) images. The fluoroscopic images are obtained using kV x‐rays in‐line with the treatment‐beam direction. The DDRR images(DRRs at different phases during respiratory cycle) are generated from 4DCT. The image registration of DDRR and fluoroscopy is based on pre‐defined structures or points of interest and needs to be performed on‐line in real time, allowing the treatment parameters to be modified in real time if a discrepancy is observed. To approve the principle, we have employed a simulator (Siemens/Mevasim) to acquire the fluoroscopic images. Both hardware and software tools were developed to synchronize the acquisition of fluoroscopy with respiratory signal using a pressuresensor (Anzai). This synchronization, in turn, harmonizes fluoroscopic images with DDRR. The verification system was tested on a motion phantom and on lung cancer cases. Results: The system developed can effectively register respiration‐synchronized fluoroscopic and DDRR images for both phantom and patient data. The registration is able to detect discrepancies between planning images (DDRR) and verification images (in‐line fluoroscopy) for a 4D‐ART delivery. The system is found to be effective for validating respiratory gating. Conclusion: We have developed a treatment verification system for 4D‐ART. The system, employing in‐line kV fluoroscopy, may be used for validating respiratory gating and for 4D‐ART with the new generation of image‐guideddelivery machine capable of in‐line dynamic imaging. The system can be also potentially useful for 4D real‐time tumor tracking based on fluoroscopy.Conflict of Interest: This work is supported in part by Siemens OCS.

High-resolution video-based EPID for cone beam CT

A high-resolution video (HRV) based EPID that is capable of matching the high spatial resolution and SNR (signal to noise ratio) of a-Si flat panel devices was developed at Siemens OCS (Oncology Care Systems) for cone beam CT. This system is using a high resolution CCD camera (1300 x 1030). The optical components and scintillator screen were modified to generate high-resolution images. A Pips-Pro QC3 phantom was used to compare the spatial resolution and the contrast to noise ratio (CNR) of a-Si flat panel (1024x1024) and HRV system. The QC phantom was placed at the linear accelerator iso-center, and the detectors were placed 40cm below iso-center. The measured f50 for the Siemens a-Si flat panel and HRV are 0.49 lp/mm, and 0.43 lp/mm; respectively. The image of the flat panel had already been corrected for Offset, Gain and Defective pixels; however, no correction was performed on the HRV system. Due to the fast readout of HRV, small pixel size, and adjustable camera lens aperture, a thicker scintillator screen would have resulted in an increase in SNR without sensor saturation during radiation treatment imaging. This is one of the main advantages of this system compared to the flat panels, and it makes the system ideal for cone beam reconstructions as well as regular therapy imaging. A new electronic readout is implemented in HRV control circuit that will synchronize the image acquisition with the linear accelerator, and thereby increases the SNR.