Index of content:
Volume 34, Issue 6, June 2007
- Joint Imaging/Therapy Symposium: Auditorium
- Young Investigators Symposium
34(2007); http://dx.doi.org/10.1118/1.2761177View Description Hide Description
Purpose: The purpose of this work was to develop a novel technique for automatically evaluating exit dosimetry on tomotherapy systems using auto‐associative modeling that is robust and has the capability to learn complex detector data relationships, even with detector data with a low temporal resolution and beam attenuation from the patient. Method and Materials: Delivery sequences from 3 patients were used in this study. Each delivery sequence was modified by reducing the opening time for random individual MLC leaves by random amounts. The error and error‐free treatments were delivered with different phantoms in the path of the beam. Multiple auto‐associative kernel regression (AAKR) models were developed and tested by the investigators using combinations of the sinogramdata sets.AAKR is a non‐parametric model that is used to predict correct values when supplied a group of sensor values that is corrupted. Models were tested using the data containing errors. However, models were never developed with data which had the same object in the path of the beam as the dataset it was testing. This allowed the testing of the model's error detection capabilities in the presence of attenuation. Results: The results show that the model correctly distinguished the MLC positional error from changes in attenuation. The model identified errors in compressed detector data that had been summed over 94 frames. Generally, errors greater than 7 milliseconds were visually discernable. Some smaller errors could be detected, but it depended on the position of the erroneous leaf in the projection and the actual projection shape. Conclusion: The results presented suggest that AAKR modeling could be used to monitor and eventually improve the reliability of radiation delivery. This method has the potential to play a noteworthy role in determining and possibly correcting for the types of machine‐related errors that occur during actual patient treatments.
SU‐GG‐AUD‐02: Dose Convolution Filter: Incorporating Spatial Dose Information Into Tissue Response Modeling34(2007); http://dx.doi.org/10.1118/1.2761178View Description Hide Description
Purpose:Radiotherapytreatment planning commonly involves analysis of static dose distributions and corresponding Dose Volume Histograms (DVHs). However, such analysis does not account for biological effects of spatial variations in the physical dose distribution. We introduce the Dose Convolution Filter (DCF) model capable of incorporating spatialdoseinformation in plan analysis and optimization, and integrating biological factors such as cell migration and bystander effects into physical dose distributions. The DCF model should allow more accurate prediction of tissue response from complex radiotherapydose distributions, and can facilitate modeling of the effects of patient motion. Method and Materials: We use a Gaussian convolution filter with standard deviation, σ, determining the degree of dose washout. To test this model, filtered dose distributions are applied to a NTCP model to calculate tissue response. As an illustration, we determine σ from existing rat spinal cord data, and compare model‐predicted NTCP with published data. We also simulate the GRID technique, in which an open field is collimated into many pencil beams. Results: After applying DCF, an NTCP model can predict dependence of tissue response on variations in spatialdose distribution. The model successfully fits the rat spinal cord data with a predicted value of σ=2.6±0.5mm, consistent with 2mm migration distances of remyelinating cells. Moreover, it enables the correct prediction of a high relative seriality for spinal cord. Finally, this model also predicts the sparing of normal tissues by the GRID technique when the size of each pencil beam becomes comparable to σ. Conclusion: The DCF model incorporates spatialdoseinformation and offers an improved way to estimate tissue response from complex radiotherapydose distributions. It does not alter the prediction of tissue response in large homogenous fields, but successfully predicts increased tissue tolerance in small or highly non‐uniform fields.
Partially supported by Varian Medical Systems.
SU‐GG‐AUD‐03: The Development and Validation of An Image‐Based Dosimetry System for 90Y Microspheres Used to Treat Hepatic Tumors34(2007); http://dx.doi.org/10.1118/1.2761179View Description Hide Description
Purpose: To develop and experimentally validate an image‐baseddosimetry system for determining the three‐dimensional (3D) dose distribution from microspheres used to treat hepatic tumors.Method and Materials: A rapid, efficient, and stable batch technique was used to label yttrium‐loaded microspheres with . These ‐labeled microspheres served as surrogates for ‐labeled microspheres. and microspheres were coinjected into a gel‐based phantom and the activity distribution was determined using a GE Discovery LS PET/CT scanner. The activity distribution was converted from to by applying a precise activity ratio, which was determined using germanium detection and a low uncertainty positron branching ratio. To calculate the dose, the image data was convolved with a dose point kernel using 3D‐ID software. This dose was compared to the dose measured in the central plane using HD‐810 radiochromic film and a new film protocol. The film protocol and the gel‐based phantom were validated using a single source seed. The film was calibrated using two NIST‐traceable ophthalmic applicators and was analyzed using a flatbed scanner in reflective mode. Additionally, the image‐based dose to the entire gel phantom was compared to a Monte Carlo‐derived dose. Results: The image‐based (3D‐ID) dose in the central plane was 90.20 Gy ± 6% and the film measured dose was 90.64 Gy ± 5%. A mean phantom dose of 74.30 Gy ± 6% and 74.70 Gy ± 2% was determined using 3D‐ID and Monte Carlo, respectively. Overall, these results agreed to within 0.5%. The image‐basedin vivo dose volume histogram (DVH) for this study was in excellent agreement with the film measured DVH. Conclusion: Through the implementation of ‐labeled microspheres, a precise non‐destructive assay of , and a validated film protocol, a new image‐baseddosimetry system for microspheres was experimentally validated.
SU‐GG‐AUD‐04: Localizing Through Optimization of Image Acquisition Rate and Tube Current in X‐Ray Fluoroscopy‐Guided Therapy34(2007); http://dx.doi.org/10.1118/1.2761180View Description Hide Description
Purpose: Numerous models exist that relate the parameters of an imaging system to the image quality. Unfortunately, very little is known about the relation between the parameters of an imaging modality and the corresponding targeting precision which is of key importance in many image‐guided procedures. The purpose of this study is to explore the relation between imaging parameters and the geometric precision that can be achieved, and to show that such relations can be exploited in a framework in which the geometric objectives of a therapy can be used as feedback to drive the parameters of an imaging system and the associated therapy within constraints. Method and Materials: The task of localizing a sphere that is subject to respiratory motion and imaged under X‐ray fluoroscopy is considered. Two cases are considered: (i)a system that modulates the temporal sampling frequency using feedback of uncertainty while tracking the sphere in order to maintain a specified maximum spatial uncertainty while minimizing exposure, and (ii)a system that modulates the tube current in order to maintain the localization uncertainty within the specified level as the sphere traverses a noise field. The performance of the proposed system is contrasted with a similar system that does not employ feedback of uncertainty. Results: It is observed that a relation exists between the localization uncertainty and the imaging parameters considered, namely, the temporal sampling frequency and the tube current. The use of uncertainty as feedback allows tradeoffs between targeting precision and imaging parameters to be controlled. Conclusion: The proposed framework can: (i)automatically maintain the geometric objectives of the therapy whenever possible and report to the operator when these objectives cannot be achieved, (ii)optimize the system parameters by dynamically assigning them based on feedback of performance, and (iii)reduce the level of human intervention required to carry out the therapy.
SU‐GG‐AUD‐05: Segmented Crystalline Scintillating Detectors for Radiotherapy Imaging: A Monte Carlo Investigation of Swank Factor34(2007); http://dx.doi.org/10.1118/1.2761181View Description Hide Description
Purpose: A systematic theoretical investigation of Swank factor for thick, segmented crystalline scintillating detectors is reported. The Swank noise, originating from variation in the x‐ray‐to‐photon conversion gain, degrades the detective quantum efficiency (DQE) performance of radiotherapyimagers. Therefore, it is of great interest to examine the Swank factor, in particular the optical Swank factor, as a function of various detector design parameters. Method and Materials: Swank factor was investigated using a recently implemented Monte Carlo package (MANTIS) that simulates both radiation and optical transport. The radiation and optical Swank factors were examined using a 6 MV photon beam as a function of various parameters, such as the material and thickness of the scintillator, the element‐to‐element pitch of the detector, as well as the material and thickness of the septal wall separating detector elements. Moreover, the optical Swank factor was examined as a function of the optical properties of the scintillator, top reflector and septal wall. Results:Radiation Swank factor improved when thicker scintillators or higher density septal walls were employed. The optical Swank factor dropped dramatically when the scintillator surfaces facing the walls were not polished. Moreover, the results indicate that septal walls must be highly reflective, as the optical Swank factor was found to drop sharply with decreasing septal wall reflectivity, even for a relatively clear scintillator. Furthermore, the optical Swank factor was also strongly affected by the thickness, absorption and scattering of the scintillator, and modestly affected by the reflectivity of the top reflector. Conclusion: Simulation of radiation and optical transport in segmented scintillating detectors provides the means to examine properties that are critically important to performance, such as optical Swank factor. It is strongly anticipated that such studies will greatly assist in the design of detectors exhibiting very high DQE.
SU‐GG‐AUD‐06: Stray Radiation Exposure During Proton Radiotherapy of the Prostate: The Influence of the Patient On Scatter and Production34(2007); http://dx.doi.org/10.1118/1.2761182View Description Hide Description
Purpose: To characterize the scatter, production, and attenuation of secondary radiation in patients receiving passively‐scattered protonradiotherapy for prostate cancer.Methods and Materials: A proton therapytreatment was simulated using a Monte Carlo model of a double scattering treatment machine. Whole body effective dose (E) from secondary radiation was estimated from a weighted sum of doses to the major organs in an anthropomorphic phantom. The effect of the patient on secondary dose was quantified by comparing E with ambient dose equivalent, H*(10), which was based on free‐in‐air spectral fluence calculations at isocenter. Various treatment parameters (proton beam energy, range modulation width, field size, and snout position) were varied in order to study their influence on E and H*(10). Results: The calculated E for the simulated treatment was 7.8 mSv/Gy, while the calculated H*(10) at isocenter was 16 mSv/Gy. Both E and H*(10) approximately doubled over the range of modulation widths and energies studied. As field size increased from 0×0 to 15×15, E doubled, while H*(10) decreased by 30%. When the snout position was changed from 30 cm to 48 cm, E decreased by less than 20%, while H*(10) decreased by 44% over the same interval. Simulations revealed that, while E is predominated by neutrons generated in the nozzle, neutrons produced in the patient contributed significantly (up to 40%) to dose equivalent in near‐field organs. In most cases, H*(10) provided a conservative estimate of E. However — because H*(10) does not account for neutrons created in the patient — it did not conservatively estimate E for large field sizes, where neutron production in the patient becomes significant. Conclusions:Neutrons generated in the patient contribute significantly to exposures to organs near the irradiated volume. When evaluating stray radiation exposure, production, scatter, and attenuation in the patient should be taken into consideration.
34(2007); http://dx.doi.org/10.1118/1.2761183View Description Hide Description
Purpose: The purpose of this study was to design an instrument for measuring protonenergy spectra, and present energy deposition results that compare calibration runs with two state of the art Monte Carlo codes, HETC‐HEDS, and GEANT. Method and Materials: The protonenergyspectrum analyzer developed in this study is based on a telescope called the Cosmic Ray telescope of the Effects of radiation (CRaTER), originally designed to achieve characterization of the global lunar radiation environment and its biological impacts. It employs a stack of silicon detectors and tissue‐equivalent plastic (TEP) to establish the linear energy transfer (LET) spectra of cosmicradiation relevant for human and electronic parts considerations. Measured and calculated spectra were obtained using dE/dX= (E0−Ef)/L, where E0 and Ef are the energies of the particles as measured by the silicon detectors at either end of each TEP section and L is the length of the material.Results: Results show a very good agreement between experimental and computer models. A comparison of the peak position shows that HETC‐HEDS simulation of the instrument in a clinical proton beam yield a peak at channel 50. In comparison, the peak channel was 55 with Geant and 57 from the MGH data. The reason for the discrepancy is that the HETC‐HEDS analysis was performed for energies down to 0.2 MeV, while the analyses with GEANT and the MGH data included around 20 more energy channels. Conclusion: An LET energy spectrum was designed and computational testing for proton therapy. Comparisons between calibration runs and two simulation runs using GEANT and HETC‐HEDS show promising results in being able to use the instrument for proton therapycalibration purposes. Future investigations will include analyses for different geometrical configurations, as well as investigating the detector's response to other charged particles.
SU‐GG‐AUD‐08: Real‐Time Three‐Dimensional Position and Orientation Data of a Brachytherapy Robot Using Magnetic Tracking34(2007); http://dx.doi.org/10.1118/1.2761184View Description Hide Description
Purpose: To incorporate a magnetic tracking system with a brachytherapyrobot to provide real‐time information on position and orientation of anatomical structures. Methods and Materials: A custom‐built six‐degree‐of‐ freedom robot was engineered for highly accurate prostate brachytherapy implantations. The robot was tested and optimized in phantom to provide sub‐millimeter error in seed placement. However, an implantation's accuracy may be limited by the ability to transfer 3‐D anatomical data and convert it to the robot's coordinate frame. This transformation is difficult if the imager is not fixed, because the transformation from anatomical coordinates to robot coordinates changes dynamically. We tested the characteristics of a Minibird II tracking system (Ascension Technology, Burlington, VT) which consists of a transmitter that emits a magnetic field and a sensor to provide feedback of position and orientation. Multiple sensors can be added to determine the relationship between the robot and an ultrasound probe, for example. An analysis of the Minibird's accuracy in relationship to the highly precise movements of the robot in three dimensions was performed. Results: The robot's movement compared to the magnetic tracker's positional information over 24 cm of travel was within 1.21%, 0.77% and 1.18% in the x, y, and z directions respectively. Conclusion: Using magnetic tracking, a robot designed for brachytherapy implantation may use nearly any imaging modality to acquire anatomical data. A sensor may attach to a free‐hand ultrasound probe or fluoroscopy machine, for example. This opens the possibility of using the robot for needle insertions other than prostate LDR procedures with a rectal probe. The robot can be modified easily for accurate needle targeting of many soft tissue structures. Lastly, the cost of magnetic tracking is small (in equipment and upkeep cost) when compared to 3‐D infrared tracking, while also eliminating the “line‐of‐sight” issues that may occur with those systems.
SU‐GG‐AUD‐09: On the Impact of Functional Imaging Accuracy (Sensitivity and Specificity) On Selective Boosting IMRT34(2007); http://dx.doi.org/10.1118/1.2761185View Description Hide Description
Purpose: To quantify the impact of loss in functional imaging sensitivity and specificity on tumor control and normal tissue toxicity for selective boosting IMRT.Methods and materials: Four selective boosting scenarios were designed: SB91‐81 (EUD=91Gy for the high risk tumor subvolume (nodule) and EUD=81Gy for a remaining low risk PTV (rPTV)), SB80‐74, SB90‐70, and risk‐adaptive optimization. For each sensitivity loss level, the loss in tumor control probability (ΔTCP) was calculated. For each specificity loss level, the increase in rectal and the bladder toxicity was quantified using the radiobiological indices (equivalent uniform dose (EUD) and normal tissue complication probability (NTCP)) and physical indices (%‐volumes). Results: The impact of loss in sensitivity on local tumor control was maximized as the dose level for rPTV had a lower value. The SB90‐70 plan had a ΔTCP= 93.8 %, for the SB91‐81 plan ΔTCP = 26.8 %, while for risk‐adaptive optimization ΔTCP= 8.0 %. Independent of planning technique, loss in functional imaging specificity appears to have a minimal impact on the expected normal tissue toxicity since an increase in rectal or bladder toxicity as a function of loss in specificity was not observed. Additionally, all plans fulfilled the rectum and the bladder sparing criteria found in the literature for late rectal bleeding and genitourinary complications. Conclusions: Our study shows that the choice of a low‐risk classification for the rPTV in selective boosting IMRT may lead to a significant loss in TCP. Furthermore, it appears that in order to improve the therapeutic ratio a functional imaging technique with a high sensitivity, rather than specificity, is needed.
34(2007); http://dx.doi.org/10.1118/1.2761186View Description Hide Description
Purpose: At present the most promising method for an in vivo and noninvasive monitoring of radiation treatments with protonbeams is positron emission tomography(PET). This study investigates the sensitivity and accuracy of the PET/CT treatment verification method in the presence of highly inhomogeneous tissue regions as well as in the presence of metallic implants. Method and Materials: A circular SOBP proton field has been delivered to a sophisticated in house designed phantom consisting of polymethyl methacrylate (PMMA), lung and bone equivalent slabs. The bone material contained authentic dental gold implants as well as tin lead alloy implants in different shapes. PET data were acquired in listmode starting within 15 min after irradiation at a commercial PET/CT scanner. The measured PET distributions were compared to full‐blown simulations of the PET signal based on Geant4 and FLUKA Monte Carlo(MC) codes. Results: Activation profiles were analyzed behind air‐lung, air‐bone and lung‐bone interfaces parallel to the beam as well as downstream interfaces angled at 6°. In general, a good agreement between measured and simulated PET distributions was found. Measured PET images reflected even small characteristic changes in the dose distribution. This showed the potential of PET/CT treatment verification in the presence of highly inhomogeneous tissue regions. CT artifacts due to metal implants can trick all dose calculation algorithms and lead to the prediction of a proton range overshoot behind metal implants, whereas in reality a range undershoot occurs. The PET/CT treatment verification method can detect such dose calculation errors. Conclusion: This preliminary study indicates the feasibility of PET/CT treatment verification to detect the full characteristic of the delivered dose distribution even in the presence of complex tissue inhomogeneities and metal implants.