Volume 35, Issue 6, June 2008
Index of content:
- Joint Imaging/Therapy Scientific Session: Auditorium C
- Applications of Functional Imaging to Therapy
35(2008); http://dx.doi.org/10.1118/1.2962390View Description Hide Description
Purpose: One of the most important properties for biologically‐based radiotherapy(dose painting) is accurate definition of the treatment target. Here we investigate consequences of non‐uniform dose escalation based on proliferative response, when the treatment target is defined on FLT‐PET standardized uptake values (SUV) as opposed to the parametric image parameters KFLT obtained through compartmental modeling.Method and Materials: Patients were imaged with FLT‐PET before the start and after one to two weeks of radiation therapy. The SUV images and KFLT parametric images were calculated from dynamically acquired PET data. Mid‐therapy images were co‐registered with pre‐treatment images, and ratios of mid‐treatment to pre‐treatment SUV or KFLT images were calculated. These ratios were connected to spatially‐dependant radiosensitivity according to the linear‐quadratic survival model. The two continuous voxel‐based dose painting treatment plans were compared to each other and to the uniform dose escalation treatment plan. Results: Calculated SUV and KFLT images show similar, although distinctively different visual patterns. Ratios of SUV or KFLT images show increased spatial heterogeneity compared to the original images. Correlation of the ratio images is comparable to the correlation of individual SUV or KFLT images.Dosimetric evaluation of the plans revealed that approximately 25% of the target volume received more than 10% different dose for the SUV ratio prescription compared to the KFLT ratio prescription. On the other hand, the SUV ratio dose painting plan was clearly superior to the uniform dose escalation plan where over 75% of the target volume received more than 10% different dose.Conclusion:Dose painting using the prescriptions based on SUV or KFLT leads to significantly different treatment plans; however, their differences are smaller than the differences to the uniform dose escalation. These differences warrant careful clinical investigation to establish superiority of either prescription.
MO‐E‐AUD C‐02: Incorporating SPECT Functional Lung Images in Routine Treatment Planning for Lung Cancer35(2008); http://dx.doi.org/10.1118/1.2962391View Description Hide Description
Purpose:SPECTlung perfusion images has been used to derive functional DVH for planning a decade ago (Marks et al 1994), however, it is still not widely used. The current study describes a method to incorporate functional images in routine planning based only on commercial TPS. Method and Materials: 12 patients were immobilized with the same Alpha‐Cradle on SPECT/CT combo unit and CT simulator. The SPECT was registered to the planning CT by registering the low‐resolution CT to the planning CT. The whole lung was divided into several sub‐volumes according to the pixel values of the perfusion image (PVPI) above the background. The lung functionality was assumed to be linear with the PVPI for up to 80% of its maximum value, as used by other investigators. The mean PVPI in each lung sub‐volume was used to defined the functionality weight factor. The V20function was determined based on the dose‐volume and weight factor for each sub‐volume. Results: The priority score was set for each sub‐volume according to the weight factor for IMRT optimization. For non‐IMRT plan, sub‐volumes with higher weight factors serve as avoidance structures for field optimization. The 22% (range 11–35%) lung volume had a weight factor ⩾ mean weight factor. The difference between perfusion weighted lung volume and actual lung volume was − 1565 cc (mean, range −491 to −3270 cc). V20function was smaller than V20 in 11 of 12 patient and the difference ranged from −0.4% to −12%. Conclusion: Because the exact relationship between lung functionality and PVPI is still uncertain, large bin number may not be required. Our method for reducing the number of functional bins to several/less than a dozen may still provide useful prediction for toxicity and significantly reduces the amount of effort in planning and can be routinely implemented in a clinical setting.
35(2008); http://dx.doi.org/10.1118/1.2962392View Description Hide Description
Purpose: In radiation therapy, use of 2‐Fluoro‐2‐deoxy‐D‐glucose (FDG) PET imaging to assess treatment response in tumors and normal tissues is suboptimal because of the confounding effects of radiation‐induced inflammation. We have investigated the feasibility of 3′‐Deoxy‐3′‐fluorothymidine (FLT) PET imaging, as a surrogate of cell proliferation, to assess treatment response in tumors and as well as normal tissues.Method and Materials: Patients receiving radiation therapy were imaged twice with FLT‐PET/CT: prior to therapy, and after receiving approximately 10–20 Gy of radiationdose. A 90 minute dynamic FLT‐PET image acquisition was initiated after the injection of FLT. The average standardized uptake values (SUV) between 60–90 minutes were used in the analysis. The CT data between the imaging sessions was co‐registered and the corresponding PET data compared and analyzed. Both, the tumor region as well as the surrounding normal tissues within the radiation field were analyzed for their response. Results: The FLT‐PET SUVs decreased approximately 20–50% within the first two weeks of radiation therapy, which provides an optimal timing window for treatment response assessment. Of the normal tissues,radiation effects were most notable in bone marrow, because of the high normal FLT uptake. Radiationdoses in excess of 10 Gy lead to complete bone marrow ablation, as assessed with FLT‐PET imaging. For doses below 10 Gy, an exponential cell‐kill relation correlates well with the observed decrease in FLT bone marrow uptake. Radiation effects were observed in other normal tissues as well. However, due to the low baseline FLT uptake, the changes were not readily detectable. Conclusion: FLT‐PET imaging was demonstrated as a powerful tool for early assessment of proliferative response to radiation therapy.Treatment assessment is possible as early as one week after the initiation of therapy. In addition to tumor response, FLT‐PET imaging also provides means to assess normal tissue damage.
MO‐E‐AUD C‐04: MR Guided Focused Ultrasound (MRgFU) for Treatment of Prostate Cancer: Feasibility Study of Incresing Intratumoral Uptake of Docetaxel in Vivo35(2008); http://dx.doi.org/10.1118/1.2962393View Description Hide Description
Purpose: Docetaxel has been used for the treatment of advanced hormonerefractory prostate cancer and it is also a potent radio‐sensitizer. Our goal is to determine if pulsed MR guided focused ultrasound (MRgFU) will enhance the intratumoral concentration of docetaxel. Method and Materials: This study was performed on an InSightec ExAblate 2000 HIFU system together with a 1.5T GE MR scanner. Human prostate cancer cells LNCaP 105, were grown orthotopically in the prostates of nude mice. A radioactive tritiated Docetaxel was used to determine the uptake enhancement into prostate tumor. Fifteen mice were randomly divided into 3 groups (Group 1, ; Group 2, only; Group 3, control). The tumors (163 ± 9.0mm3) were treated using pulsed ultrasound with an acoustic power of 4W, pulse width 100msec and 300 pulses in one sonication. The focal peak was set within the target using MR guidance. Eight to ten sonications were used to cover the whole tumor. Immediately after the treatment , dissolved in PBS, was given by tail vein injection at doses of 15 mg/kg and a tracer amount of 1.25 uCi/25g. After 0.5 hr, the animals were euthanized and tumors removed. Tumortissues were digested in solubilizing reagent for 2hr at 55°C and decolorized by hydrogen peroxide. The digested samples were added to liquid scintillation cocktail and counted using a liquid scintillator. Results: Our preliminary results showed that the animals tolerated well the MRgFU treatment. The average of radioactive cpm counts in the MRgFU treated group is 2 folds more than that without the HIFU treatment. The variation is large between individual animals and further experiments are being conducted to reduce the experimental uncertainty. Conclusion: MRgFU may have a great potential as a safe, noninvasive treatment modality for the enhancement of docetaxel for prostate cancer therapy.
MO‐E‐AUD C‐05: Can Voxel Based Prescription Dose Be Determined From Multiple Physiological MR Modalities for Brain Tumors?35(2008); http://dx.doi.org/10.1118/1.2962394View Description Hide Description
Purpose: To determine spatially non‐uniform dose prescription for braintumors based on physiological MRI data, we have retrospectively analyzed the correlation between pre‐RT MR imaging, delivered dose, and post‐RT MR imaging using various multivariate analysis methods. Method and Materials: Non‐uniform dose prescription using the pre‐RT functional imaging can be meaningful only if a correlation between imaging signal to required minimum dose exists. This correlation can be indicated by a separation in the imaging — dose space between clusters of voxels where tumor recurred and those where tumor is sterilized. To determine the correlation, we performed various multivariate analyses including k‐means clustering, decision tree, principle component analysis and visual scattering plots on a voxel‐by‐voxel basis. A separation in the clusters of responsive and non‐responsive voxels is sought in the N+1 dimensional space (N=number of physiological imaging modalities + dose). Analyses were performed on co‐registered pre‐ and post‐RT multiple physiological MR data (MRSI, rCBV (spin‐echo (GE), gradient‐echo (SE) and ratio parameters), DTI‐derived ADC (apperent diffusion coefficient), and FA (fractional anisotropy)) with delivered dose distributions and tumor recurrence maps based on post‐RT imaging for selected braintumor patients. Results: Among various combinations of MRI modalities tried, no imaging modality or combination provided strong separation. A weak separation was observable only when multiple MR modalities were considered. Visual separation happened mainly with ADC map and rCBV‐GE. Different clustering methods yielded a weak separation (mean within‐cluster distance to among cluster distance ratio < 0.5). A decision tree built by the data provided good sensitivity/ specificity (both >0.7), for the high risk region voxels. Conclusion: The multivariate analyses show a weak correlation for the studied patient cases, which may be attributed to complex tumor dynamics. This indirectly indicates that temporal (update during course of treatment) and spatial non‐uniform dose prescription may be helpful for braintumors.
35(2008); http://dx.doi.org/10.1118/1.2962395View Description Hide Description
Purpose: To present a novel method for monitoring radiation damage using the parameters of our 5D lung motion model. Method and Materials: The 5D breathing motion model describes breathing motion as a function of tidal volume ν and airflow f and is parameterized as: , where r is the position of a piece of tissue located at reference position r 0. α and β are each functions of r 0 and relate tissue motion with tidal volume and airflow. The continuity equation , where ρ is the local density, t is time, and U is the velocity field, is modified to replace tidal volume as the independent variable. At inhalation or exhalation, the f=0 and under these conditions, the continuity equation leads to in the tissue reference frame. This equation provides the relationship between the divergence of α and the relative change in local tissue density. The powerful aspect of this result is that the α parameter is determined using free‐breathing scan registrations which inherently include the complex hysteresis interplay, and yet its divergence indicates the local relative density change as a function of tidal volume. The values of α were determined for repeat 5D CT scans for both irradiated and unirradiated lungs.Results: The unirradiated patient had little change in the value of α between two scan sessions. A lungcancer patient had 5D CT scans acquired after 16 Gy and 70 Gy tumor dose. The value of α changed dramatically in the irradiated regions. This result indicates that α is sensitive to radiation dose damage. Conclusion: The divergent of α is shown to be related to the relative local density variation as measured using free‐breathing 5DCT. Repeat scans indicate that variations in the α distribution, corresponding to changes in local density variations, may be sensitive to local radiation damage.
35(2008); http://dx.doi.org/10.1118/1.2962396View Description Hide Description
Purpose: To model and compute spectral output, intensity, dose fall off, dose rate, and penumbra that can be achieved using industrial orthovoltage x‐rays sources used in small animal micro irradiators. Method and Materials: Using data from commercial sources as input parameters we investigated the performance of orthovoltage sources that can be used in small animal micro irradiators. We developed a pencil beam propagation code which was combined with a mouse digital phantom (MOBY‐John Hopkins Univ.) to simulate a complete small animal delivery system (source + animalmodel). We computed the source spectral output, beam filters, beam penumbra, soft tissue to bone dose ratio, dose fall off, and total dose delivered to the animalmodel. Sources covering the range of 150 to 450kVp with submillimeter focal spot were simulated. Results: We determined that a source of nominal maximum potential output of 320kVp and focal spot of 0.4×0.4 mm2 outperformed other available sources. We designed an optimum Thoraeus‐like filter to obtain a bremsstrahlung spectrum energy greater than 2 mm of Cu to increase skin spare and reduce bone dose. An average beam penumbra of 0.25mm and a dose rate of 40Gy/min were possible using this filtered beam. Higher energy sources would increase cost and shielding thickness. Lower energies sources showed limited intensities when they were aggressively filtered. Conclusion: We developed a numerical model to evaluate the radiation dose delivered by orthovoltage sources typically used in small animal irradiators. We concluded that when radiation quality, skin dose, bone to tissue dose ratio and animal throughput were considered, commercial orthovoltage sources of nominal energy of 320 keV were the best fit for conformal small animal micro irradiators. To obtain high conformality a submillimeter focal spot of 0.5×0.5 mm2 or less must be used.
This work supported in part by NIH grant R21CA108677
- Imaging for Therapy Assessment
TH‐D‐AUD C‐01: Early Blood‐Brain‐Barrier Disruption in Response to RT as a Biomarker for Neurotoxicity35(2008); http://dx.doi.org/10.1118/1.2962913View Description Hide Description
Purpose: Studies of neurocognitive dysfunction after radiation in animals suggest that vascular injury plays a key role. We hypothesized that blood‐brain‐barrier (BBB) disruption in normal appearing cerebral tissue of patients early in the course of fractionated radiation therapy (RT) is a biomarker for delayed neurocognitive dysfunction. Method and Materials: Ten patients with low‐grade glioma, or suprasellar lesion and underwent 3D conformal cranial RT (28–33 fx of 1.8 Gy) participated in a prospective MRI study. Dynamic‐contrast enhanced (DEC) MRI was acquired before, at week 3 and week 6 during the course of, and at 1, 6 and 18 months after the completion of RT. Using the modified Toft model, the contrast transfer constant (K) from the intravascular space to the extravascular extracellular space was estimated. A battery of standardized neuropsychological tests was performed at the same times as the pre‐ and post‐RT MRI. The relationship between the temporal changes in K and the dosimetric parameters was analyzed by a linear mixed model. Correlations between the changes in K and early delayed changes in the neurocognitve functions were analyzed by linear regression.Results: The K values increased significantly in normal appearing tissue regions that received >40 Gy at week 6 during RT (p<0.05), suggesting BBB opening. The elevated K values decreased gradually after RT. The changes in K both during and after RT were significantly correlated with the doses received at the time but the significance decreased from p = 0.0001 at week 3 during RT to 0.03 at 6 months after RT. The changes in K of left frontal lobe at week 3 during RT were significantly negatively correlated with the changes in verbal learning scores at 6 months after RT (p<0.02). Conclusion: Our data suggest early BBB disruption could be a biomarker for delayed neurocognitve function deterioration.
Supported by NIHP01CA59827 and R21CA11369901.
TH‐D‐AUD C‐02: Reliability Study of Ultrasound Tissue Characterization in Quantitative Measurement of Radiation‐Induced Breast Tissue Toxicity35(2008); http://dx.doi.org/10.1118/1.2962914View Description Hide Description
Purpose: To evaluate the reliability of the ultrasoundtissue characterization (UTC) as a measure for breast‐tissue radiation toxicity. Method and Materials: We have recently reported that UTC midband‐fit could be used to assess radiation‐induced tissue toxicity. The reliability of this method is presented in this report. Twenty‐three breast patients previously treated with radiation were recruited. Both treated and untreated breasts were scanned by one radiation oncologist. The untreated breasts were used for reliability study, in which two ultrasound radio‐frequency (RF) images were acquired at the same position. A region‐of‐interest (ROI) was selected either manually by a physician or a computer program, in which a fixed ROI was used. Between‐scan repeatability and the correlation between physician and computer program were assessed using intraclass correlation coefficient (ICC). To evaluate the ability of UTC to scale tissue toxicity, the patients were divided into four groups according to patient self‐assessment of breast hardening and the medians of the midband‐fit differences between treated and untreated breast in each group were investigated. Results: The repeatability using single measurement is 0.860 and 0.804 for physician and computer program, respectively. When the average of two measurements is used, the repeatability for computer program is 0.891, which suggests computer program can contest with the physician at the expense of double measurements. The correlation between physician and computer program is very good (ICC = 0.897), which indicates a substantial agreement between the physician and the computer program. For both method, the UTC midband‐fit increases with higher tissue toxicity of the patients. Conclusion: Both physician and computer program can assess toxicity reliably. The physician may prefer the computer program for automatic evaluation of tissue toxicity. There is a clear concordance between the UTC evaluations and the patients' self‐assessment, which proves the reliability of using UTC in quantitative measurement of radiation toxicity.
35(2008); http://dx.doi.org/10.1118/1.2962915View Description Hide Description
Purpose: To investigate, at what time point in therapy the patient hemoglobin (Hgb) levels and the blood perfusion in cervical cancer have the greatest impact on outcome prediction of radiation/chemotherapy (RT/CT). Method and Materials: Eighty‐eight patients with cervical cancer stages IB2‐IVA were treated with standard RT/CT. Serial weekly blood tests, including Hgb levels, were collected and all patients underwent 4 serial DCE‐MRI: pre‐RT, at 2–2.5 weeks, at 4–5 weeks during RT and 1–2 months post‐RT. Mean follow‐up was 4.7 (range 0.1–9.0) years. Hgb level, representing systemic oxygenation, and the lowest 10th percentile of signal intensity within tumor (SI10), representing local tumor blood supply, were combined and evaluated for various time points to predict the effectiveness of RT/CT. Outcome analyses were carried out with Mann‐Whitney ranksum test and Kaplan‐Meier method. Results: In separate analyses, the best time for outcome prediction for either Hgb or SI10 was at 2–2.5 weeks into treatment and a dose of 20–25 Gy. The p‐value for local tumor control was <0.001 and 0.013 for Hgb2wk and SI102wk respectively, significantly better than at the other time points, ranging 0.02–0.4 for Hgb and 0.06–0.95 for SI10. Low tumor oxygenation, reflected by simultaneously low Hgb and low SI10, significantly correlated with local tumor recurrence. The 5‐year local recurrence rates are 45% vs. 7% at 2–2.5 weeks (p<0.001) respectively, compared to 36% vs. 9% at pre‐therapy (p=0.003), and 37% vs. 9% at 4–5 weeks (p=0.003). Conclusion: Combining information of patient Hgb levels and tumor perfusion provides a good indirect measure of the tumor oxygenation in cervical cancer. This study indicated that examinations performed early in therapy at a dose of 20–25 Gy, significantly correlated with therapy outcome, and could be used to identify early those patients at risk of local tumor recurrence.
TH‐D‐AUD C‐04: Four‐Dimensional Motion Correction to Distinguish Respiratory From Biological Changes in Treatment Response Assessment Using Molecular Imaging35(2008); http://dx.doi.org/10.1118/1.2962916View Description Hide Description
Purpose: The in‐vivo imaging of tumor biology through positron emission tomography(PET) scans has become an invaluable tool for therapy response assessment since it provides tumor response data at early treatment stages. However, imaging in‐vivo small molecular changes in tissues subject to respiratory motion is infeasible as the PET tracer is smeared during the acquisition process introducing errors up to 75% associated with degraded tracer activity. The purpose of this study was to develop a motion detectionsystem that quantifies respiratory changes using spatio‐temporal Bspline deformable image registration for analysis of motion‐degraded PET datasets. Method and Materials: A clinical software tool implemented the Bspline deformable image registration method. Algorithm output is a motion‐corrected 4D PETimage containing only treatment‐induced changes. The motion‐free dataset can be summed to provide a simple 3D dataset that is used to for therapy evaluation using the standard tools developed for motion‐free treatment sites. Results: Capabilities of motion correction algorithm were established in studies on tumor motion in dynamic phantoms using checkerboard and surface distance tools and in patients using the convergence analysis method. The BSpline model was able to reproduce respiratory motion with a maximum error of 3 mm on five clinical cases of lung lesions. Conclusion: The integrated motion‐correction 4D PETsystem will provide a standardized protocol for characterization of tumor response using a combination of conventional and motion‐sensitive measures. The algorithm discerns between respiratory and treatment induced changes and thus, enables individual tumor response and evaluation of anti‐angiogenic therapies in images of thoracic tumors affected by respiration.
35(2008); http://dx.doi.org/10.1118/1.2962917View Description Hide Description
Purpose:Tumors are known to be heterogeneous but it is unclear whether and how heterogeneity changes throughout therapy. The purpose of this work was to quantitatively assess proliferative heterogeneity over time course of treatment.Method and Materials:Tumor heterogeneity and its temporal development were investigated for patients undergoing either radiotherapy or chemotherapy. Six radiotherapy patients were imaged prior to radiation therapy and 2–3 weeks after the first scan while receiving treatment. Four chemotherapy patients were imaged with FLT‐PET prior to treatment, during Sunitinib malate therapy (week 4) and during withdrawal (week 6). Spatial statistics were used to assess proliferative heterogeneity over tumor volume. Global Moran I statistics with inverse distance weighting were used to estimate the overall degree of spatial autocorrelation in cell proliferation. Proliferation clusters were visualized using local G statistics, which identified local regions of strong autocorrelation. Results: No significant changes in tumor heterogeneity during radiation therapy were seen, with the mean change in Moran I throughout treatment of only −2.5±6.9 %. On the contrary, chemotherapy patients showed changes in tumor heterogeneity over the time course of treatment. Moran I changed by −14.5±3.7 % from pre‐treatment to week 4 and 13.0±8.1 % from week 4 to week 6. The correlation coefficient between chemotherapy response and change in Moran I from pre‐treatment to week 4 was 0.75. On the other hand, no strong correlation between the initial heterogeneity and response to treatment was observed. Conclusion: Results showed that spatial statistics successfully provide a measure of tumor heterogeneity and its changes over the course of treatment. The heterogeneity changes depend upon treatment type and response to treatment.
TH‐D‐AUD C‐06: Early Assessment of Treatment Response in Hematopoietic Disease Using [18F]FLT PET Imaging35(2008); http://dx.doi.org/10.1118/1.2962918View Description Hide Description
Purpose: Assessment of treatment response in hematopoietic diseases like leukemia is essential for disease management and routinely performed after therapy via bone marrow biopsy, an invasive and ineffective predictor of treatment response. We developed a molecular imaging based methodology for bone marrow assessment and applied it to leukemia patients during their chemotherapy for early assessment of treatment response. Method and Materials: Six adult leukemia patients and ten adult subjects with normal bone marrow were injected with 5mCi 18F‐FLT, a marker of cellularproliferation, and received whole body PET/CT scans. Leukemia patients were treated with standard induction chemotherapy and imaged at progressively earlier time points during therapy. Normal bone marrow was used to establish baseline assessment parameters including bone marrow mean SUV, SUV distribution, and heterogeneity of the axial distribution of bone marrow uptake from the pelvis to neck. Leukemic bone marrow was assessed and therapy responders were compared with non‐responders. Responders were distinguished from non‐responders based upon clinical outcome of therapy. Results: Mean bone marrow SUV of responders was lower than that of non‐responders (0.76±0.05 vs. 1.60±0.14, p=0.0054). For responders, the SUV distribution dramatically shifted toward lower SUVs than in normal patients. This shift was considerably smaller in nonresponders. Axial distribution of bone marrow uptake was more heterogeneous in non‐responders than responders, and this heterogeneity may explain the poor predictive power of the bone marrow biopsy, which is a point measurement. Measurement of treatment response was not significantly affected by the time of assessment. Conclusion: FLT PETimaging was used for early assessment of treatment response in leukemia patients during chemotherapy.Treatment response was characterized by bone marrow mean SUV, SUV distribution, and heterogeneity of the axial distribution. Given the limitations of bone marrow biopsy, FLT PETimaging provides a superior tool for assessment of treatment response in hematopoietic diseases.
TH‐D‐AUD C‐07: Theoretical Analysis of a New Generation Portal Imaging Sensor Based On Thin‐Film CdTe: A Feasibility Study For Clinical High Energy X‐Ray Detection35(2008); http://dx.doi.org/10.1118/1.2962919View Description Hide Description
Purpose: Currently most popular electronic portal imaging devices (EPIDs) are manufactured from hydrogenated amorphous silicon, a material with low atomic number and electron density, exhibiting low quality images and poor radiation hardness when used for megavoltage imaging. We propose a new generation of portal imagers based on thin‐film Cadmium Telluride (CdTe), offering device improvement in both imaging and dosimetric properties. In this paper the result of our optimization studies in material/thickness combinations with an estimate of its output signal under typical conditions are presented. Method and Materials: Due to very small thickness (∼100 microns) the sensor has to be combined with a metal plate facilitating conversion of high‐energy photons to charge carriers directly, maximizing the deposited dose in the sensor layer. Monte Carlo (MC) package MCNP5 was utilized to optimize type and thickness of the material used for this purpose. We modeled an ELEKTA SL‐25 Linac head matching characteristics of a 6MV X‐ray beam. The effect of CdTe layer thickness on frequency‐dependent detective quantum efficiency DQE(f) of the device was also analyzed. Based on MC generated profiles we evaluated the voltage output signal of the CdTedetector for different thickness of sensor layers, as well as the effect of front and back exposure, using the software package SCAPS‐1D. Results: Based on calculations of DQE(f) we proved CdTe‐based detector system to have higher performance than those using amorphous silicon or selenium. We established the optimal material/thickness combinations for thin‐film CdTe/metal plate detector and found that resultant charge carrier generation leads to the voltage output of 0.2 – 0.3 Volts, warranting pulse mode operation without biasing and cooling. Conclusion: Following a successful mathematical modeling of an optimized detector and measured output voltage, we believe the thin‐film CdTe‐based detector is well suited for imaging with high energy x‐rays used in clinical radiation therapy.
35(2008); http://dx.doi.org/10.1118/1.2962920View Description Hide Description
Purpose: A closed‐loop MR‐guided laser induced thermal therapy (LITT) system has been developed and validated in large animals, obtained 510(k) clearance and undergone initial safety studies in humans. In this work we outline the capabilities and potential applications of the system, recent advances and current challenges in MR temperatureimaging for guidance of these procedures, initial results in human subjects and current work in integrating the system with an interventional MR suite. Method and Materials: All imaging has been performed on 1.5T MRI scanners. Temperature was measured in real‐time using MR thermography based on the proton resonance frequency shift and used to control a high‐power 980‐nm diode actively‐cooled laser for interstitial LITT. An MR compatible targeting template, similar to that used for brachytherapy delivery, was developed and used to investigate the feasibility of performing MR‐guided LITT in prostate using a canine model. Treatments in spine used a canine sarcoma model. Intracranial treatment in a canine model were performed to test the safety of the equipment and evaluate the lesions over time prior to commencing treatment of intracranial metastasis in patients. MR‐guided LITT using MRTI feedback was also performed in patients with liver metastases. The system was integrated into a multi‐modality interventional suite featuring a large bore 1.5T magnet and floor mounted fluoroscopy unit for image guided procedures. Results: Actively‐cooled laser applicators facilitate much more rapid therapy delivery than previous laser technology allowing 2‐cm treatments to be completed in <90 seconds in many cases. MR temperatureimaging provided valuable feedback during therapy allowing a more aggressive therapy delivery. Procedures were well tolerated in both animal and human subjects. Conclusion: MRI‐guided LITT is feasible and safe. MR temperatureimaging provides a means to either qualitatively monitor the therapy or apply quantitative dosimetry which can accurately predict the region of ablation.
TH‐D‐AUD C‐09: Towards On‐Line Treatment Verification Using Cine EPID for Hypofractionated Lung Radiotherapy35(2008); http://dx.doi.org/10.1118/1.2962921View Description Hide Description
Purpose: To develop a computational algorithm based on an artificial neural network (ANN) that allows treatment verification using an EPID in cine mode for hypofractionated lungradiotherapy.Method and Materials: We developed a novel ANN based technique using cineEPIDimages to verify that the target was within the beam aperture when the beam was on. The first step of using the ANN involved training from a training dataset. We simulated training images, i.e., cineEPIDimages with different tumor locations, by shifting DRRs relative to the beam aperture. With a pre‐defined threshold p%, we associated category 1 to the training image if more than p% of the tumor projection in the beam eye view was within the aperture and category −1 otherwise. The trained network could therefore analyze the cineEPIDimages obtained during the treatment and classify them into the corresponding category 1 or −1. Results: Two patients, each treated with 5 fractions, were included in our feasibility study. A radiation oncologist read the cineEPIDimages and classified them into category 1 or −1; this served as our ground truth. The ANN was applied to the training images to build the neural network. We set p%=95% for this study. For each treatment field, one neural network needs to be built. Averaging over both patients and all fields, the trained network successfully classified 97.5% of the cineEPIDimages overall. Conclusion: The proposed ANN based technique can successfully analyzecineEPIDimages to verify whether or not the tumor is within the beam aperture, and it can do so with high accuracy. This technique provides an important clinical safeguard—whenever the tumor moves out of the irradiation field, the treatment beam can be interrupted, so that radiation won't be unnecessarily delivered to normal tissues.