Volume 34, Issue 6, June 2007
Index of content:
- Imaging Moderated Poster Session: Exhibit Hall C
- Moderated Poster — Area 4 (Imaging): Image Processing, Dosimetry, QC and PACS
SU‐DD‐A4‐01: Nonrigid Registration of Mesorectal Region for PET Signal Follow‐Up During Radiation Therapy34(2007); http://dx.doi.org/10.1118/1.2760326View Description Hide Description
Purpose: We wish to achieve maximum registration performance in the region around the tumor for patients with rectal cancer to assess the changes in PET signal due to radiotherapy.Method and Materials: A nonrigid registration was performed on nine PET‐CT images of three patients with rectal cancer. We use a B‐spline transformation model with mutual information as similarity criterion. The registration is performed in a multiresolution framework. To exclude the influence of differences in bladder filling, registration was limited to the mesorectum, delineated at planning time, by only calculating the similarity criterion for the control points inside the mesorectum. In the last multiresolution stages of the registration, a small, local volume constraint was used to regularize the deformation field inside the tumor region. Validation was performed by comparing volume overlap (Dice Similarity Coefficient or DSC), centroid distance and volume change of the manually delineated rectum contour for all slices where tumor was actually seen on the FDG‐PET. Results: When using the region of interest, volume overlap increases while centroid distance decreases for all registrations. The mean DSC when using the entire image for registrations was 0.68 compared to 0.79 when only registering inside the mesorectum. The mean centroid distance decreased from 4.40mm to 3.30mm. The mean volume difference decreased from 28.11% to 14.09%. Conclusion: Results show that by limiting the registration to the mesorectum, a much higher volume overlap for the rectum can be seen. Also, by regularizing the deformation field around the tumor in the last stages of the registration, a slight increase in rectum correspondence is found. The results of these registrations can be used to evaluate the PET signal around the tumor over time.
SU‐DD‐A4‐02: Autosegmentation and Internal Target Volume (ITV) Generation in 4DCT Lung Imaging Using Deformable Image Registration34(2007); http://dx.doi.org/10.1118/1.2760327View Description Hide Description
Purpose: To present a methodology using an improved intensity based deformable image registration algorithm (i.e. Juggler algorithm) for autosegmentation of normal anatomical structures, and algorithm generation of internal target volume (ITV). Method and Materials: The Juggler algorithm separately registers high intensity gradient and low intensity gradient features, and is particularly well suited for maintaining the topological properties of individual structures. Simulated CTimaging consisted of algorithm deformed clinical CTlungimaging. Clinical imaging, acquired from 4DCT, consisted of free breathing (FB) CT and 10 phased CT sets associated with the respiratory cycle. Manual image segmentation of normal structures was carried out on a reference CT (i.e. FB or end of exhalation phase), and the registration map between the reference and target CTs was used to autosegment the structures onto the target CT. This map was also used for automatic generation of ITV based either on “union target volume” (UTV) (i.e. the union of all the separate 3D segmentations), or “probability density function target volume” (PDFTV). Results: Based on known displacement vectors for simulated data, and difference imaging and cross‐correlation values for clinical data, Juggler yielded superior and faster registration compared to current deformable registration algorithms used in radiotherapy: demons, accelerated demons, free‐form deformation. Based on 4DCTs for 5 patients, normal anatomical structures (lung, skin, trachea, esophagus, heart) were automatically autosegmented onto target 3D CT sets within ⩽2mm accuracy based on visual agreement. The UTV computed ITV increased the FB GTV by up to 5mm. Total computation time was <3min (including deformable registration). Conclusions: The proposed methodology using an improved registration algorithm allowed for accurate autosegmentation of normal structures from initial manual segmentations, and the computation of ITV. It provides an effective treatment planning tool for 4DCT. New computer hardware could potentially reduce computation time to <30sec.
34(2007); http://dx.doi.org/10.1118/1.2760328View Description Hide Description
Purpose: The acquisition of multiple images of varied yet complementary information is rapidly becoming the norm in radiotherapytreatment planning. Beside CTdata sets, PET and/or MRI or MRS images are also being used to aid in the definition of the target volume (or normal structures) for treatment optimization. We have been developing new methods to integrate available imaginginformation (anatomical and/or physiological) for concurrent target registration and segmentation. Towards this goal, we have investigated several clustering and active contour methods for simultaneous 2D/3D segmentation/registration of multi‐modality images consisting of combinations of PET, CT, or MRI datasets. In this work, we present a phantom validation study of this approach. Method and Materials: A commercial anthropomorphic head phantom of average human size was used. Targets consisting of plastic spheres and rods were placed throughout the cranium section of the phantom. Tap water was used for CTimaging. For MRI and PET imaging, the water inside the phantom was doped with CuNO3 and 18F‐FDG, respectively. The cold spots spheres were considered as the segmentation targets (four spheres each is 25.4 mm in diameter). The rods were used as landmarks to assist in alignment. A geometric deformable model known as the multi‐valued level set (MVLS) method was applied as our computational vehicle. Results: The MVLS algorithm achieved 90% segmentation accuracy and less than 2% volume error when integrating all of the three modalities. This is contrasted with 74% segmentation accuracy and 4.4% volume error when using CT‐based systems only. Conclusion: We have validated a semi‐automatic method for integrating information from different imaging modalities. Our phantom study demonstrates the feasibility of the proposed image integration approach. This approach could potentially provide radiologists/oncologists with reliable and efficient tools to analyze simultaneously different modality data in different cancer sites.
This work was supported by ACS‐IRG‐58‐010‐50 grant.
SU‐DD‐A4‐04: Dosimetry and Image Quality Evaluation of a Dedicated Cone‐Beam CT System for Sinus and Temporal Bone Applications34(2007); http://dx.doi.org/10.1118/1.2760329View Description Hide Description
Purpose: To evaluate the dose and image quality performance of a dedicated cone‐beam CT scanner for sinus and temporal bone applications. Method and Materials: A low‐dose cone‐beam CTsystem with flat‐panel detector has recently been introduced for high‐contrast applications in the head. Because of the non‐uniform dose distribution throughout the volumetric field of view, the dose in the central plane may not accurately represent the overall dose. In this work, we introduce a novel metric, volumetric average dose, to incorporate the spatial variation of radiation dose in a cone‐beam scan. The definition and measurement of volumetric average dose are analogous to conventional Weighted CTDI, though conceptually different. Using this metric, we evaluated the dose performance of a cone‐beam CT scanner (MiniCAT, Xoran Technologies). Two methods were employed for measurement. One was with a small solid‐state detector (RTI CT‐SD16), the other was with a conventional CT pencil chamber. Both were measured with a standard CTDI head phantom. The low‐ and high‐contrast spatial resolution, and cone‐beam and truncation effects, were also evaluated. Results: The volumetric average doses for sinus and temporal bone studies measured with CT pencil chamber were 5.02 mGy and 4.33 mGy, respectively. With the RTI CT‐SD16 detector, 4.41 mGy and 3.93 mGy were obtained for the two studies. Both are substantially lower than those of conventional adult CT protocols in our practice (58 mGy for sinus and 84 mGy for temporal bone). Isotropic high‐contrast spatial resolution of 16 1p/cm was measured for the temporal bone mode. Low contrast resolution, as anticipated, was inferior to conventional CT.Conclusion: A novel dose metric, volumetric average dose, was used to characterize the dose performance of a cone‐beam CTsystem. The dose and image quality of the dedicated cone‐beam CTsystem appear appropriate for sinus and temporal bone applications.
34(2007); http://dx.doi.org/10.1118/1.2760330View Description Hide Description
Purpose: The diagnostic x‐ray shielding requirements for Radiographic and R&F rooms presented in Figs. 4.5 through 4.8 in NCRP Report No. 147, and for cardiac angiography labs, have been fit to a three‐parameter equation that relates the barrier thickness x to the value of NT/(Pd2). The locations of the source in the imaging room appropriate to the determination of distance d are also explicitly presented. Methods and Materials: Section 4.2.4 of NCRP‐147 presents lead and concrete shielding requirements for barriers around “representative” Radiographic and R&F rooms that include contributions from all clinical beam locations and directions. The required shielding thickness, x, for the various barriers around each room is presented graphically as a function of NT/(Pd2), where N is the weekly number of patients, T is the occupancy, P is the permitted weekly air kerma, and d is the distance (in m) from an x‐ray source to the occupied area. This method has been applied previously to cardiac angiography labs. Letting η0 be the maximum value of NT/(Pd2) for which no shielding is required, barrier thickness x depends on (NT)/(Pd2) following the equation of Archer et al. (1983): . Results: The values of NT/(Pd2) (mGy−1 m−2) have been fit to Eq. 1 as a function of x for the curves in Figs. 4.5 through 4.8 of NCRP‐147, and for cardiac angiography labs. The resultant values of η0, α, β, and γ for lead and concrete barriers are presented. Conclusions: The use of Eq. 1 with the fitting parameters facilitates the use of the NT/(Pd2) methodology from NCRP‐147 in computer applications. The agreement of the fit and the thicknesses read from the NCRP report is better than 0.026 mm lead and 1.7 mm concrete.
34(2007); http://dx.doi.org/10.1118/1.2760331View Description Hide Description
Purpose: In the last a few years PACS has been increasingly available and used in clinical care centers. However, there was no standardized quality control for PACS display units, which is the crucial last step in an imaging chain for all diagnostic imaging modalities. In 2005 the final draft of AAPM TG‐18 report was released. The purpose of this work was to implement QA for PACS display utilizing AAPM TG18 report in a tertiary care center. Method and Materials: There are total 66 diagnostic and clinical PACS display monitors located in various areas of the main hospital and 2 satellite clinics. QA procedures were developed using the 2005 final version of AAPM TG‐18 report. VeriLUM (IMAGE Smiths, Inc.) software was installed on all display monitors and a VeriLUM luminance/illuminance spot meter was used for all measurements. Various AAPM TG‐18 report test patterns were used for these evaluations. Results: Quarterly QA including geometric distortion, reflection, luminance response, luminance dependencies, and resolution was performed by a PACS technologist/coordinator. Annual QA including geometric distortion, reflection, luminance response, luminance dependencies, resolution, noise, and veiling glare was performed by a medical physicist. PACS display quality related issues were quantitatively determined using the QA procedures developed from AAPM TG‐18 report. Conclusion: QA for PACS display has been successfully implemented for a year and half in a clinical health care center using AAPM TG‐18 report. It is going to be used for the enterprise wide PACS in the health care system in the near future.
- Moderated Poster — Area 4 (Imaging): Magnetic Resonance, Ultrasound, and Micro‐CT Imaging
SU‐EE‐A4‐01: Development of Cardiac‐Gated 3‐Dimensional Ultrasound Imaging of Carotid Atherosclerosis34(2007); http://dx.doi.org/10.1118/1.2760351View Description Hide Description
Purpose: To determine the time interval in the cardiac cycle for prospectively gated ultrasound imaging of the carotid artery that will reduce change in cross‐sectional lumen area to 5%. Method and Materials: Three female subjects (ages 26–46) were used for this pilot study. The ultrasound transducer was placed over the left common carotid artery in the transverse orientation. Images and electrocardiogram values were recorded by in‐house 3‐dimensional ultrasound acquisition software. The lumen of the common carotid artery was outlined by one observer using manual segmentation at 66 ms intervals and the area of each outline was calculated by in‐house software; this was repeated 3 times. Results: The mean change in lumen area was (15 ± 2)%. A change in lumen area of 5% was selected as the desired change after prospective gating. At systole, the interval of the cardiac cycle for 5% change was 0.19 ± 0.05 to 0.54 ± 0.08 for subject 1, 0.26 ± 0.06 to 0.67 ± 0.07 for subject 2, and 0.6 ± 0.4 to 1.0 ± 0.4 for subject 3. At diastole, the interval was 0.77 ± 0.05 to 1.15 ± 0.06 for subject 1, 0.77 ± 0.09 to 1.20 ± 0.08 for subject 2, and 0.6 ± 0.2 to 0.9 ± 0.2 for subject 3. For both diastolic and systolic intervals, there was no significant (p<.05) difference between the start and end of the intervals for subjects 1 and 2, but the interval for subject 3 was significantly different from both. Conclusion: We have combined ultrasound imaging and data from an ECG to measure the effect of cardiac cycle on lumen cross‐sectional area. The lumen area of the common carotid artery changed by 15% over the cardiac cycle. No consistent interval was found to reduce this area change to 5% for prospective gating.
SU‐EE‐A4‐02: Evaluation of Ultrasound Localization Versus MV Portal Images of Fiducial Markers in Prostates34(2007); http://dx.doi.org/10.1118/1.2760352View Description Hide Description
Purpose: This pilot study evaluated prostate localization by comparing ultrasoundimages to orthogonal MV portal images of fiducial markers implanted into the prostate. Method and Materials: Each prostate patient had gold fiducial markers implanted into his prostate prior to simulation. The Restitu™ ultrasound system (Resonant Medical, Inc.) was used to acquire the ultrasoundimages. The first ultrasoundimage was acquired immediately prior to acquiring the CT simulation image. These images were fused using the CT isocenter, and the prostate reference volume was contoured for each patient. This contour included a portion of the inferior bladder wall near the trigone to assist with daily localization. Each day, therapists acquired an ultrasoundimage and overlaid the prostate reference volume contour onto the current image. Orthogonal MV portal images were then acquired. Displacements of imaged fiducials from their expected locations as observed on the DRRs were removed by shifting the couch if the displacements exceeded 5 mm. Following treatment, the location of each fiducial in all 3 directions was measured on the final portal images for each day and averaged to measure the final prostate location. Differences in where the ultrasoundimage and where the portal images of fiducials would locate the prostate were compared. Results: For 11 patients, the differences are less than 6.1 mm within the 95% confidence interval. For 1 patient, ultrasound imaging did not consistently reproduce the prostate location to within 10 mm as compared to fiducials via our technique. Sources of deviation include slight discrepancies in calibrating the ultrasound systems, slice spacing, different users, fusion discrepancies, image quality, and random uncertainties. Conclusion: For most patients, ultrasound and ports of fiducials provide comparable localization information for prostates. However, sources of disagreement still exist. Anatomical landmarks can be useful in most cases but can also be misleading if improperly used.
34(2007); http://dx.doi.org/10.1118/1.2760353View Description Hide Description
Purpose: To develop an in vivoimaging method with small animal computed tomography(CT) and clinical water soluble iodinated contrast agent for investigations of aneurysm formations in rat brain that could lead to hemorrhagic stroke.Method and Materials: Small animal cone‐beam CT (60 kVp/0.530 mA) was acquired for a normal Sprague‐Dawley rat (300 g) anesthetized with isoflurane. The rat was positioned with a nonmetallic stereotaxic immobilization device to minimize the motion during CT acquisition. Noncontrast CT was immediately followed by contrast‐enhanced CT with continuous administration of iopromide (300 mgI/ml) at the rate of 2.5 ml/min with 512 projections over approximately 2 min. Noncontrast CT was obtained a) to visualize the effect of contrast agent in contrast‐enhanced CT, and b) to examine the effectiveness of the immobilization device between two time points of CT acquisitions. The accuracy of registration between the noncontrast and contrast‐enhanced CTimages with the immobilization device was compared against the images registered with normalized mutual information using FMRIB's Linear Image Registration Tool (FLIRT) using 12‐parameter affine transformations. Translations and rotations were examined between the FLIRT coregistered noncontrast CTimage and the uncoregistered noncontrast CTimage.Results: Continuous administration of iopromide during the CT acquisition provided consistent vascular contrast in the CT reconstructions. Carotid arteries and major cerebral blood vessels were visible with contrast‐enhanced CT, but not with noncontrast CT. The translational and rotational differences between FLIRT‐coregistered and uncoregistered images acquired with the stereotaxic immobilization were less than 0.5 voxel (∼85 μm) and 0.5°, respectively. Conclusion: A CT‐compatible immobilization device was useful for microCT imaging vascular structures in and around rodent brain with continuous administration of a standard iodinated contrast agent (e.g., iopromide), which provides a small animal CT technique that does not require specialized small animal specific contrast agent that has comparatively long in vivo residence time.
34(2007); http://dx.doi.org/10.1118/1.2760355View Description Hide Description
Purpose: To use Monte Carlo simulations to characterize the design and performance of a nanotechnology‐based high spatial and temporal resolution small animal micro‐RT system. Method and Materials: The proposed micro‐RT system uses carbon nanotubefield emissiontechnology to produce arrays of individually and electronically controllable x‐ray pixels that can form spatially and temporally high resolution modulated irradiation. Combined with the existing carbon‐nanotube field emission micro‐CT, the future micro‐CT‐RT system promises a high spatial and temporal resolution image‐guided irradiation system ideal for small animal research. Prospective designs for the micro‐RT system were evaluated using EGSnrc‐based Monte Carlo simulations. Design aspects studied included: 1) x‐ray anode design, 2) collimator design, and 3) dosimetric considerations, including dose rate, inhomogeneity corrections, and the overlap effect of adjacent pixel beams. Results:Anode studies indicate that reflective and transmission target designs produce similar dosimetric properties, finding dose rates of 40 cGy/min/mA/pixel are achievable at isocenter (10 cm SAD) for 2 mm × 2 mm pixels in a 3 cm diameter mouse model. For beam energies of 80–100 kV, photons below 20 keV must be filtered out to eliminate excessive surface dose for the mouse model. This filtration can be achieved using a tungsten transmission target 26 μm thick or a 2.5 mm Al filter with the reflective target. Dose overlap between adjacent x‐ray pixels produces high or low dose spots for a single beam array direction. When opposing beam pair configurations are used, the dose inhomogeneity is reduced to negligible levels near isocenter and 20% near the surface. Conclusion:Monte Carlo simulation studies are instrumental to the development of a novel micro‐RT system. They are used to evaluate performance on different system designs without the cost of prototype fabrication. The Monte Carlo study verified basic dosimetric features of the intended micro‐RT design.
34(2007); http://dx.doi.org/10.1118/1.2760356View Description Hide Description
Purpose: Both cerebral tumor and cardiovascular disease may compromise cerebrovascular reactivity due to abnormal angiogenesis or blood supply. Functional MRI during hypercapnia stress, such as breath‐holding (BH), has been used to assess the cerebrovascular reactivity. However, in clinical settings, patients may not be able to hold their breath well, which can reduce the sensitivity for detection when an fMRI model‐based analysis is applied. In view of the global effect induced by breath‐holding, this study proposed a model‐independent analysis, using the whole‐brain‐averaged time curve as a reference function(WBRF), and compared with the results using the canonical hrf(CHRF). Methods: T2*‐weighted images were acquired using a single‐shot GE EPI sequence on 1.5T MR scanner, with TR/TE=3000ms/60ms. Four experiments were performed on two normal volunteers: (1) BH after deep inspiration, (2) after an instruction, BH after the end of the natural expiration, (3) self‐paced BH by button pressing and (4) take in a short breath during the BH duration. For model‐independent analysis, the WBRF was the averaged time curve over the whole braintissue. For model‐dependent analysis, the CHRF was adopted from the SPM2. Voxels with statistically increased signals were determined (p<0.001), and then the fractional activation volume (activated volume/whole volume) was calculated. Results: In all cases, the WBRF was able to show the real breath‐holding paradigm performed by the subjects. From the four experiments, the fractional activation volumes by model‐independent analysis were 62, 53, 57, 54% and 58, 4, 5, 12% by model‐dependent analysis, respectively. Conclusion: The results illustrated model‐independent analysis is a more reliable approach. This is caused by the fact that WBRF better reflects the true beath‐holding situation then CHRF. Due to the impaired patient performance in clinical applications, the method proposed by this study can help increasing the sensitivity of cerebovascular reactivity MRI.
SU‐EE‐A4‐06: Calorimetric Measurements of the Efficiency of a Transmit‐Receive Head Coil for the Optimization of MR Head Examinations Within SAR Safety Levels34(2007); http://dx.doi.org/10.1118/1.2760357View Description Hide Description
Purpose: To determine the efficiency (i.e. the ratio of the actual SAR over scanner reported SAR) of a transmit‐receive (TR) head coil, over clinically‐relevant ranges of transmitted RF power and patient weight, aiding in the optimization of parameters for head examinations requiring careful SAR monitoring. Methods and Materials:Temperature increases due to the deposition of RF energy during MRI scanning were measured on a clinical 1.5T Siemens MAGNETOM Avanto (revision B13 software). The calorimetric tests employed a 4‐liter cylindrical head phantom placed inside a standard TR head coil. The phantom and the ambient temperatures,T and TA , were recorded simultaneously using MR‐compatible temperature probes. A spin echoimaging sequence with 4‐echoes, TR/TE1/TE2/TE3/TE4 100/20/40/60/80ms, 256×256 matrix, a single 3mm‐slice, and 70 signal averages to prolong the scan time to 30min was used. The measurements were repeated for five transmitter reference voltages, 117, 105, 93, 84, and 58.5V. The scanner's SAR estimates S were recorded for patient weights between 50 and 150kg. The temperature measurement continued after scanning to allow determination of the cooling constant k. Data were analyzed using Newton's Law of Cooling: , where e is the coil efficiency, c is the heat capacity of water. First k and subsequently e were determined by fitting the measuredtemperature history to the corresponding solutions of that equation. Results: For higher reference voltages, 117 and 105V, the efficiencies were found to increase with patient weight, from 0.6 to 0.88 for weights 50–150kg. Conversely, for lower reference voltages, the efficiencies become weight‐independent and are about 0.5, 0.4, and 0.2 for 95, 84, and 58.5V, correspondingly. Conclusions: The observed efficiencies indicate somewhat conservative SAR estimates, especially for the low range of transmitted RF power. Therefore these measurements provide a useful refinement to the model‐dependent SAR estimate displayed by the MRI system.
- Moderated Poster — Area 4 (Imaging): Cone beam CT and X‐Ray Imaging
TU‐EE‐A4‐01: Dose Saving and Scatter Reduction in Volume‐Of‐Interest (VOI) Cone Beam CT — a Monte Carlo Simulation Study with Geant434(2007); http://dx.doi.org/10.1118/1.2761404View Description Hide Description
Purpose: To estimate and study the dose saving and scatter reduction properties of the VOI cone beam CT technique. Method and Materials: To implement the VOI scanning technique, a filter with a circular or rectangular opening is inserted between the x‐ray source and the patient to deliver a higher level x‐ray exposure inside the VOI and a lower exposure level outside the VOI in acquiring the projection images. This technique is expected to result in higher image quality inside the VOI with lower entire patient dose and significantly lowered scatter‐to‐primary ratios (SPRs) within the VOI. Modeling the patient as a cylinder of soft tissue, we have developed Monte Carlo simulation programs based on the Geant4 package to estimate the dose to the patient and the SPRs at the detector input. Simulation was performed at selected locations to estimate the variation of the dose level and SPRs from inside to the outside of the VOI. Simulation was also performed to estimate the reduction of the dose levels and SPRs with the VOI scanning technique as compared to full field cone beam CT.Results: Our simulation results show that with the exposure level reduced by the VOI filter, dose levels for the entire scan were significantly reduced both inside and outside the VOI. Scatter intensities at the detector input were also shown to decrease significantly both inside and outside the VOI in the projection images.Conclusion: Our simulation study successively demonstrated the ability of the VOI scanning technique to reduce patient dose and x‐ray scatter for improved image quality. Acknowledgement: This work was supported in part by a research grant EB000117 from the NIH‐NIBIB and a research grant CA104759 from the NIH‐NCI. The author also would like to thank Dr. Ioannis Sechopoulos and Yu Chen for consulting assistance in Geant4 programming.
TU‐EE‐A4‐02: Exposure Index Calibration for a Portable DR Detector — A Practical Guide for the Clinical Medical Physicist34(2007); http://dx.doi.org/10.1118/1.2761405View Description Hide Description
Purpose: Establish a method for calibrating and utilizing the exposure indices (EXP, REX) of a Canon portable DR imagingsystem.Method and Materials: The Canon model CXDI‐50 was investigated in order to understand the nature of the reported EI's, to establish a calibration procedure, and to verify the EI adequacy as an indicator of the detector exposure for clinical procedures. Canon reports two EI's: “REX”, which changes with image processing settings and “EXP” which is independent of image processing changes. However, there was very little documentation of the EXP parameter and no associated calibration procedure. The effect of an undocumented adjustment parameter (“Constant for Exposure Index”) was investigated and ultimately served to calibrate the EXP value for a standard RQA6 beam (80kVp, 27 mmAl). Using an ionization chamber the exposure at the input to the detector was measured as a function of the reported values of EXP and REX for the RQA6 beam, for 70 kVp, 0.5 mmCu+2 mmAl (TG116), and for several combinations of kVp and acrylic attenuator. Results: Both REX and EXP linearly increase with exposure (different slope). REX is associated with the density on a printed film and thus changes with window and level settings. This makes REX a poor indicator of exposure adequacy. The “constant for exposure index” has no affect on REX. To achieve the same EXP value as obtained with RQA6, the detector dose must be increased by 20% for 80kVp, 8″ acrylic, 31% for 120kVp, 6″ acrylic, and 43% for 60kVp, 6″ acrylic. Conclusion: The information presented will be particularly useful to a medical physicist desiring to establish the calibration of EXP, to determine the appropriate target EXP for a particular body part and view, and to calibrate a phototimer to be utilized with the Canon DR system.
TU‐EE‐A4‐03: A Comparison of Anatomical Noise Properties Between Breast CT and Projection Breast Imaging34(2007); http://dx.doi.org/10.1118/1.2761406View Description Hide Description
Purpose: To compare the statistical properties of the anatomical noise present in breast images.Images from a dedicated breast CT scanner are compared with mammographic projection images.Method and Materials: A dedicated breast CT (bCT) scanner was used to image the breasts of volunteers and patients. Twenty‐five of the available 105 patient datasets were selected for analysis.Noise power spectra (NPS) calculations were performed on the datasets for the right breast of the 25 patients in order to examine the statistical nature of the anatomical background of the breast. The projection images acquired at 80 kVp during the cone beam breast CT acquisition were analyzed using NPS calculations. Approximately 50 projection images and 50 bCT slice images were analyzed per patient. Three regions of interest (ROIs ) were used per image. The ROIs were randomly distributed within the boundary of the breast. The 2D NPS was calculated for each ROI and the average 2D NPS was computed and averaged radially yielding a 1D NPS. A power law expression of the form αf−β was computed from the results of the projection and bCT images, and the β values were compared. Theoretical development suggests that the β for bCT should be one less than that of the projection images, i.e. β [bCT] = β [proj] − 1. Results: The β values for the bCT slice images were consistently less than those for the projection images. The difference ranged from 0.90 to 2.15, with the average difference being 1.34. The β values for projection images were consistent with the 2.8 value described by Burgess (2001). Conclusion: Lower β values from the NPS results of the bCT images could quantitatively indicate potential improvement in detection ability. More work is needed to conclusively say if this is the case.
34(2007); http://dx.doi.org/10.1118/1.2761407View Description Hide Description
Purpose: A new custom‐designed antiscatter grid for high‐resolution angiographic detectors is presented that would improve the image quality without introducing substantial grid‐line artifacts. Method and Materials: The new antiscatter grid is a custom‐made, parallel‐focus, crisscross cellular grid that employs gold septa material (CREATV MicroTech, Potomac, MD). The prototype 4.25 cm × 4.25 cm field‐of‐view grid has 20 μm thick septa with interspace distance of 380 μm, height 1.95 mm, and grid‐ratio of 5. This study was performed with the Microangiographic detector (43 μm pixel, 1024×1024 pixel matrix, 4.5 cm × 4.5 cm field‐of‐view, 250‐μm‐thick structured CsI(Tl) scintillator coupled to a CCDcamera via minifying taper) in simulated neurovascular angiographic conditions, where a uniform head‐equivalent phantom was used as scattering media. The air‐gap between the phantom and the detector‐with‐grid was kept at 2.5 cm as used in clinical conditions for minimal blurring, but increased scatter. The standard lead‐beam‐stop technique was employed to determine the scatter‐fraction with and without the grid. In order to evaluate the low‐contrast imaging performance of the grid, phantoms with three different bone patterns of varying thickness and three different simulated iodinated vessel inserts in acrylic were imaged with and without‐grid at 70 kVp.
Results: The grid demonstrated approximately 59% scatter reduction at 70 kVp for the uniform head‐equivalent phantom without introducing substantial grid‐line artifacts following flat‐field correction. The average contrast‐improvement‐factor for the low‐contrast vessel‐phantom was found to be 1.75, whereas for the relatively higher contrast bone‐phantom it was 1.5. The primary transmission factor was measured to be 66%. Conclusion: The grid demonstrated significant scatter reduction when used in simulated neurovascular angiographic conditions even for the small FOV. Use of this grid with the Micro‐angiographic detector with reduced air‐gap can provide substantially improved image quality.
(Partial support: NIH R01‐NS43924, R01‐EB002873, Toshiba MedicalSystem Corp.)
TU‐EE‐A4‐05: Comparison of Backprojection and a Penalized Maximum Likelihood Algorithm for Detection of Microcalcifications in Breast Tomosynthesis.34(2007); http://dx.doi.org/10.1118/1.2761408View Description Hide Description
Tomosynthesis is proving to be a valuable tool for generating three dimensional images of the breast with only a limited number of projection angles. The image quality of the resulting tomosynthesis slices can potentially be improved by using more optimal acquisition geometries and/or improved reconstruction techniques. In this paper we compare the performance of a standard back‐projection (BP) algorithm with an algorithm that maximizes a penalized likelihood (PML) objective function using an optimization transfer principle leading to a simultaneous update algorithm. To compare reconstruction algorithms, a computer simulation is used to model the breast tomosynthesis geometry and human observer performance in detecting micro‐calcification clusters is evaluated. The simulation modeled the compressed breast using a structured breast phantom, with randomly inserted clusters of spherical “micro‐calcifications”. Clinically realistic x‐ray spectra were generated and x‐ray transport through the breast phantom was modeled using ray‐tracing combined with the focal spot and the detector blur of a 200 micron CsI scintillator. The iterative reconstruction used a simultaneous update algorithm where the non‐quadratic penalized likelihood objective function (which is difficult to maximize) is replaced by a surrogate paraboloidal function. A Huber prior was used as the potential function with the ability to control edge preserving factors and smoothness. A standard back‐projection technique that is computationally faster (in comparison to the iterative method) was used as a comparison. To evaluate microcalcification detection accuracy of the methods, an N‐alternative forced choice (NAFC) based observer study was performed. The results of NAFC studies performed with 3 observers and 63 pairs of images for each method shows that the PML based technique gave a percent correct improvement by 0.32 over the FB technique. The observers read the PML based images about 4 times faster than the FB technique indicating higher visibility of calcifications in the former.
TU‐EE‐A4‐06: Experimental Evaluation of Effective Detective Quantum Efficiency for Digital Radiographic Imaging Systems34(2007); http://dx.doi.org/10.1118/1.2761409View Description Hide Description
Purpose: To develop and evaluate an experimental methodology for measuring the effective detective quantum efficiency (eDQE) of digital radiographicsystems which reflects the actual signal‐to‐noise performance of the system per unit exposure. Method and Materials: A NEXT phantom, simulating the scatter and attenuation properties of an adult human thorax was used to measure the resolution, noise, and scatter performance of a digital radiographicsystem(GE xQi) under conditions approximating those seen in clinical chest radiography. The resolution was measured in terms of the modulation transfer function(MTF) using an edge device placed at the phantom surface closest to the x‐ray tube. The noise was measured in terms of the noise power spectrum (NPS) of the region corresponding to the phantom center, acquired at three exposure levels. The scatter fraction (SF) was evaluated using a beam‐stop technique. These measurements along with measures of phantom attenuation and estimates of x‐ray flux and exposure were incorporated in the computation of the effective Detective Quantum Efficiency (eDQE). Results: The phantom exhibited a broad‐beam transmission fraction of 18.65%. The measuredscatter fraction in the presence of grid and phantom was 33%. The MTF of the system dropped by 25% at 1.0 cycles/mm when the edge was placed at the phantom surface due to scatter and focal spot blurring. The computed eDQE was assessed to be 0.038 and 0.028 at 0.5 and 1.0 cycles/mm, respectively (for E= 5.6 mR). Conclusion: Conventional DQE measurements performed under relatively idealized conditions do not accurately represent the relative performance of digital radiographicimagingsystems in routine clinical use. A more appropriate metric, the eDQE, measured under conditions that reasonably approximate those encountered clinically reflects the additional contributions from scatter, grid, and focal spot blurring, and provides a better estimate of the relative clinical performance of digital radiographicimagingsystems.
- Moderated Poster — Area 4 (Imaging): Computed Tomography — New Developments, Dosimetry and Applications
TU‐FF‐A4‐01: Combined Effects of Respiratory Motion and Object Size On 3‐D and 4‐D PET/CT Images: Dynamic Phantom Study34(2007); http://dx.doi.org/10.1118/1.2761437View Description Hide Description
Introduction: 4‐D PET/CT has potential to greatly improve the accuracy of radiotherapy target definition for treatment sites where internal organ motion is significant. While PET has an inherently greater capacity to detect cancer than CT, the best methodology for applying 4‐D PET to target definition is not currently well understood. In our study, targets of different sizes in a dynamic phantom were imaged using 3‐D and 4‐D PET/CT with the goal of better understanding how to best apply these images to radiotherapy target definition. Materials and Methods: Using a PET/CT scanner with 4‐D capability, 3‐D/4‐D image studies were acquired using a dynamic phantom. Hollow spheres filled with ‐FDG were inserted into a cavity within the phantom made of material of a density similar to lung. Recovery coefficients (RCs) were determined using 3‐D and 4‐D PETimages acquired with the phantom in static mode and a dynamic mode set to simulate respiratory motion. Results: The activity concentration in the sphere from the 3‐D PETimages can be underestimated by 40% (23%) in 2 cm (1 cm) motion. The 4‐D PET successfully recovers most of the loss of activity concentration resulting from the respiratory motion with 1% (4%) loss in 1 cm (2 cm) motion. We found that the percent recovery slightly varies with minimum 82% (93%) at the middle of expiration and inhalation and maximum 98% (99%) at the end of expiration and inhalation in 2 cm (1 cm) motion in the 4‐D PETimages. Conclusions: We demonstrate that the speed of respiratory motion at phase‐sampling position affects the measured activities of 4‐D PET due to spatial mismatch between the 4‐D PET and 3‐D CTimages for attenuation correction. Therefore, RC loss caused by respiratory motion may be minimized using 4‐D PET with 4‐D CT attenuation correction.
TU‐FF‐A4‐02: Impact of Sinogram Modeling Inaccuracies On Image Quality in X‐Ray CT Imaging Using the Alternating Minimization Algorithm34(2007); http://dx.doi.org/10.1118/1.2761438View Description Hide Description
Purpose: To provide a scientific basis for setting sinogram modeling accuracy targets based on impact of such errors on image quality. Modeling inaccuracies in photon spectrum and scatter distribution assumed by statistical image reconstruction (SIR) algorithms lead to systematic image artifacts. Methods and Materials: A synthetic two‐dimensional phantom (25×35 cm) was used to generate both noiseless and noisy sinogram data, based upon a 120 kVp spectrum filtered by 12 mm Al (66.6 keV mean energy)and variable scatter levels (4%, 20%, and 100% of the minimum primary transmission through the phantom). A third generation Siemens Somatom Plus 4 scanner geometry was assumed. The SIR algorithm was the alternating minimization (AM) algorithm [IEEE TMI 26:283]. 500 AM iterations using 22 ordered subsets were applied to the data. Various mismatches between the assumptions in the algorithm and the truth were studied, including erroneous spectra (110kVp to 130kVp, filtration from 6 mm to 18 mm Al, or 62.2 to 69.7 keV mean energy) and erroneous scatter levels (0.25 to 4.0 times the actual sinogramscatter).Result: AM image quality was evaluated in terms of bias, noise, contrast ratio, etc. To assure +/−2% accuracy in the reconstructed attenuation image,photon spectrum uncertainties corresponding to 2 keV shifts in mean energy can be tolerated. For a 30 cm thick subject, this corresponds to errors in primary transmission of 6%–8%. For 20% scatter levels, the maximum tolerated discrepancy in scatter‐to‐primary ratio (SPR) is about 5% to 8%and 30%–50% for typical MSCT scatter levels. Conclusions: This work indicates AM and other SIR algorithm image estimates are sensitive to errors in the detector response models assumed by the algorithms. For thick patients, a sinogram modeling accuracy of 6% is needed to support reconstructed images of 2% accuracy. Supported in part by NIH grant R01 CA 075371.