Improved adaptive reconstruction of multichannel MR images

PURPOSE

To improve adaptive reconstruction of multichannel MR images by simultaneously removing nonsmooth phase and signal-loss imaging artifacts.

METHODS

The improved adaptive reconstruction consists of three steps: (1) modified multichannel images are first derived by dividing raw multichannel images by a reference image (i.e., a normalized single-channel image); (2) the modified multichannel images are smoothed by a low-pass filter; (3) adaptive spatial matched filters determined from the smoothed multichannel images are utilized to obtain multichannel combined images. Numerical simulations, as well as MRI experiments, on phantoms and human subjects are performed to evaluate and compare the effectiveness of this improved adaptive reconstruction approach against traditional coil combination methods.

RESULTS

Both simulation and MRI experimental results demonstrated that the proposed improved adaptive reconstruction method is able to obtain combined images with reduced nonsmooth phase and signal-loss imaging artifacts.

CONCLUSIONS

A novel multichannel image reconstruction method is developed that produces high quality multichannel combined images.

Sci-Thur PM: YIS - 08: The effect of copper conversion plates on low-Z target image quality

Current generation electronic portal imaging devices (EPID) contain a 1.0 mm copper conversion plate to increase detection efficiency of a therapeutic megavoltage spectrum. When using these EPIDs for low-Z target imaging, the conversion plate largely attenuates the large populations of diagnostic energy photons, thereby decreasing the benefits of low-Z target imaging. In this work we measure directly the effect the variation in thickness of a copper conversion plate has on image quality in planar and cone beam computed tomography imaging. Monte Carlo modeling was used to quantify changes to the diagnostic spectrum and detector response for low-Z target beams generated with 2.35 and 7.00 MeV electrons incident on a carbon target. Planar contrast-to-noise ratio (CNR) measurements were made as a function of copper thickness. Cone beam computed tomography (CBCT) image CNR measurements were made as a function of dose both with and without the copper plate present in the EPID. The presence of copper in the EPID decreased the diagnostic photon population by up to 20% and suppressed the peak detector response at 60 kV by a factor of 6.4. Planar CNR was increased by a factor ranging from 1.4 to 4.0 with no copper present compared to 1.0 mm thickness. Increases in CBCT image CNR ranged from a factor of 1.3 to 2.1 with the copper plate removed. As a result of this we suggest that the copper conversion plate be removed from the EPID when used for low-Z target planar or CBCT imaging.

TU-E-217BCD-03: Characterization of Anatomical Noise in Mammography, Tomosynthesis and Breast CT

PURPOSE

The role of breast density in cancer detection has been well characterized, and newer modalities such as breast tomosynthesis and breast CT (bCT) were developed to improve cancer detection in the dense breast by reducing anatomical complexity. Anatomical noise was characterized on a small cohort of patient images and compared across digital mammography, tomosynthesis, and bCT images.

METHODS AND MATERIALS

An IRB-approved and HIPPA-compliant clinical study was performed on women undergoing breast biopsy, and mammography, tomosynthesis, and bCT were performed on both breasts immediately prior to biopsy. A total of 23 women participated in this study, and the unaffected breast (no lesion) was evaluated. A total of 1000 regions of interest were sampled on each image data set, and the 2D noise power spectrum (NPS) was evaluated. This was radially averaged to produce a 1D NPS, and the NPS was fit to a power law: ln{NPS(f)} = alpha+betaxln(f), over an anatomically-relevant range of spatial frequencies. The slope, beta, was averaged across patients and compared between modalities and projections.

RESULTS

The value of beta was determined for bCT data sets, and they were 1.75 (0.424), 1.83 (0.352), and 1.79 (0.397), for the coronal, sagittal and axial views, respectively. For tomosynthesis, beta was 3.06 (0.361) and 3.10 (0.315) for the CC and MLO views, respectively. For mammography, these values were 3.17 (0.226) and 3.30 (0.236), for the CC and MLO views, respectively. The values of beta for breast CT were significantly different than those for tomosynthesis and mammography (p<0.001, all 12 comparisons).

CONCLUSIONS

The results of this investigation demonstrate that the anatomical complexity of the breast, as characterized by the parameter beta, is statistically similar between mammography and tomosynthesis, a somewhat surprising finding. The breast CT image data, however, demonstrate a statistically-significant reduction in beta across all projections. Funded in part by Hologic Corporation and by a grant from the National Institute of Biomedical Imaging and Bioengineering, EB002138.

SU-E-J-21: An Intercomparison of Imaging Performance of Two Linac-Mounted Imaging Systems Used in Radiation Therapy: TrueBeam and Trilogy

PURPOSE

To evaluate and compare the performance of the imaging systems of two linear accelerators, used in radiation therapy. The study includes the following imaging components: electronic portal imaging device (EPID), kilovoltage projection imaging and kilovoltage cone-beam CT.

METHOD AND MATERIALS

The imaging systems mounted on the Varian Trilogy (Varian Medical Systems) and Varian TrueBeam, were evaluated. Image quality of two EPID systems (ASI-1000) and the two kV flat panel imagers (PaxScan 4030CB) was evaluated in terms of spatial resolution and contrast-to-noise ratio (CNR) using the QC-3 and QCkV-1 phantoms (Standard Imaging, Inc.). Cone-beam CT image sets of the CatPhan phantom (The Phantom Lab.) were obtained for standard dose head (100kVp, 0.4mAs per projection) and body (125kVp, 1.04mAs) protocols. Imaging parameters of the default clinical settings were used. The end points of the comparison were spatial resolution, CT number linearity, low contrast detectability and image uniformity. Analysis of all types of images was performed by the PIPSpro software (Standard Imaging).

RESULTS

The critical frequency (f50 in units of lp/mm) of 0.446 and 0.403 were obtained for TrueBeam and Trilogy MV detectors, respectively. The CNR was found double for Trilogy. For kilo-voltage detectors the f50 was 1.337 and 1.363, while the CNR was better by 6% in Trilogy machine. The CBCT comparison showed a 30% higher uniformity index for the TrueBeam system for pelvis protocol and 50% higher head. No significant difference was found in low contrast detectability and CT number linearity and resolution, 5 lp/mm. The Trilogy image was noisier by 35% and 30% for pelvis and standard head protocol, respectively.

CONCLUSIONS

The critical frequencies of both kV and MV detectors were found better in TrueBeam, while CNRs were found better in Trilogy. TrueBeam preformed superiorly in CBCT in terms of image uniformity and noise level.

SU-D-213CD-05: Identifying Prostate Brachytherapy Seeds at MRI: A Study in Phantom

PURPOSE

Conventionally, post-implant CT scans are used for identification of prostate brachytherapy implant seed locations. The dosimetric quality of the seed implant in our institution is evaluated based on CT/MRI fusion and contouring of prostate and rectal wall on MRI. Post-implant evaluation of prostate brachytherapy using MRI alone is generally not feasible due to the uncertainty associated with seed localization despite its excellent anatomical delineation. The fusion of CT and MRI has some variability and may be time consuming. The goal of our current work was to use SWI phase images for identification of prostate brachytherapy seeds. Using MRI alone to identify seeds will eliminate the need for CT scan of the patient post-implant and eliminate the variability of the CT/MRI fusion.

METHODS

A prostate gel phantom containing five inactive brachy seeds (Advantage I-125™, Biocompatibles, Oxford, CT) each has longitudinal cross section area of 3.6 mm2. It was assessed using CT, and MRI. Imaging was done using a GE Signa 3T HD MRI system (GE Heathcare, Millwaukee, WI). Imaging parameters for SWI were: 512×384 (zero filled to 512×512), FOV=10 cm, ASSET factor=2, TE/TR=20/42 ms, FA=15°, RBW= 80 Hz/pixel, spatial resolution=0.3 × 0.3 × 2.0 mm.

RESULTS

Brachytherapy seed, as confirmed on CT images, were easily identified in the phantoms on the filtered SWI phase images. The mean area for the 5 seeds, as measured on CT and SWI filtered phase images, was 3.5±0.5 mm² and 3.8±0.6 mm² , respectively. There appeared to be linear relationship in seed area as determined by SWI filtered phase compared to CT (R2=0.8).

CONCLUSIONS

With the improved resolution, SNR and proper filtering on high field MRI systems, SWI phase images can be used to identify prostate brachytherapy seeds on conventional MRI without using CT.

TU-E-BRA-07: Post-Operative Eye Plaque Imaging Using Tomotherapy MVCT

PURPOSE

Intra-operative ultrasound is used to verify the positioning of episcleral eye plaques used to treat ocular melanoma. Ultrasound can be ambiguous because of image artifacts, and plaques may shift position after surgery. Ultrasound verification is particularly challenging for anterior tumors. Post-operative imaging could be used to trigger interventions that would prevent local treatment failure. We investigated if, and under what conditions, the Tomotherapy megavoltage computed tomography (MVCT) system could be used to perform post-implantation verification of eye plaque positioning.

METHODS

Plaques were placed on a preserved cow's eye, and imaged with the megavoltage CT of a Tomotherapy linear accelerator (Accuray, Sunnyvale, CA). The images were visually and quantitatively assessed to determine if they were of sufficient quality to verify tumor coverage and plaque tilt with respect to the sclera. We used the visibility of the lens as a proxy for visibility of a tumor. To test the utility of hypothetical higher beam current Tomotherapy images, we averaged sequential images of the same setup.

RESULTS

The plaque, the lens of the eye, and the globe are visible in the images. The CNR of the lens with respect to the vitreous was 5.6 for a single image. For 10 images averaged, the CNR was 9.2. Estimated dose from a single image was 1.3 cGy (body CTDIvol); even 10 times this dose would be an acceptable image-guidance dose for radiotherapy patients. One limitation of the imaging procedure is the long scan time (up to 240 seconds), during which time any significant patient motion would lead to image artifacts. Human trials on eye plaque patients are planned.

CONCLUSIONS

Tomotherapy MVCT imaging could be used to verify tumor coverage and plaque tilt after episcleral plaque implantation. Tumors should be visible in standard Tomotherapy images but higher beam current images would be preferred if available.

SU-E-I-53: Optimal KVp for Image Quality and Noise in Iodine Contrast in Head CTA

PURPOSE

For brain CT perfusion it is well established that 80 kVp is optimal. Although neuro-CT angiography is somewhat similar, emphasis is on the detection of aneurysms and related vascular pathologies throughout the brain. Thus it is necessary to visualize small and large blood vessels with contrast material, as well as form multi-planar views and 3D images, so image quality and noise in addition to contrast are important for thin slices. A study was initiated to determine the optimal kVp for neuro-CTA.

METHODS

A customized version of a commercial head phantom (CIRS 007TE-27 medium adult head CT dose phantom) was purchased to facilitate quantitative measurements with iodinated contrast material, contrast for white and gray matter, and to maintain the ability to perform dosimetry. The customization consisted of adding four 25 mm holes, 35 mm from the center arranged at 45 degree angles from the center, with solid rods equivalent with brain, white, and gray matter, as well as four fillable vials were included for study of contrast agents. Dosimetry measurements were carried out with standard pencil chamber and with 0.6 cc ionization chamber. For study of the optimal kVp for a head CTA, the vials were filled with four different concentrations of contrast, approximating low to medium concentrations that would be expected in such a study. The standard CTA protocol was followed, 64 × 0.625, pitch 0.53, rotation speed 0.5 second, and CTDIvol was kept constant for each kVp.

RESULTS

The best contrast was observed at 80 kVp; however, in order to achieve noise in CTA low enough to be clinically useful there may be issues with tube current capability for a clinical technique. Clinical investigation is underway.

CONCLUSIONS

The best balance of contrast and noise currently possible will be achieved at 100 kVp in a clinical scan.

MO-D-213CD-04: 4D X-Ray DSA and 4D Fluoroscopy

During the past decade the use of undersampled acquisition and constrained reconstruction have led to significant increases in data acquisition speed, SNR, spatial resolution and temporal resolution in MR imaging. When a separately acquired constraining image is combined with an angiographic time series the traditional tradeoff between spatial and temporal resolution is greatly reduced. Artifacts and limited resolution that would normally be associated with a rapid highly undersampled temporal image series are mitigated by the constrained reconstruction process which transfers the SNR and spatial resolution of the constraining image to the individual time frames. In rotational C-Arm DSA a 3D image volume is formed from all the projections acquired during the C-Arm rotation. Although the individual projections contain temporal information, the reconstructed 3D image has no temporal information and represents a composite of the vascular filling that has occurred during the iodine injection. However, the 3D cone beam CT reconstruction can be used to constrain the reconstruction of one 3D volume for each of the rotational projections. This extends the traditional DSA time series of 2D images to a series of 3D volumes at rates up to 30 per second. Similar techniques can be used to provide fluoroscopy that can be embedded in the 3D space of the constraining volume and viewed from arbitrary angles without gantry motion. This overcomes the problem of forbidden views and guarantees that an intervention can be done without having to send patients to surgery. Unlike 4D DSA which requires only one source and receptor, 4D Fluoroscopy requires a bi-plane fluoroscopy system.

LEARNING OBJECTIVES

1. To understand the application of under sampling and constrained reconstruction to 4D DSA and Fluoroscopy.

WE-G-217BCD-07: Implementation and Evaluation of Helical On-Board CBCT and Exact Image Reconstruction

PURPOSE

Longitudinal coverage of CBCT, which is 17 cm for head scan and 15.5 cm for body scan, is not enough to cover the entire PTV for over 90% of head/neck and gastrointestinal/genitourinary/gynecologic patients if lymph nodes are involved. Helical CBCT, which was accomplished using external beam LINAC in its research mode, is one promising way to extend the CBCT longitudinal coverage. Aim of this study is to compare Katsevich's exact algorithm with traditional FDK algorithm for helical CBCT image reconstruction.

METHODS

CBCT projection raw data were acquired on a TrueBeam LINAC machine (Varian Medical Systems) in the research mode in helical trajectories that encompass a 360 degree rotation, 25 cm pitch, 100 kVp, 80 mA, and 25 ms, with a Catphan 600. Reconstruction was done with Katsevich's exact and FDK approximate algorithms. Scatter correction, beam-hardening correction, and non-uniform gantry angle correction, are performed on projection data to reduce artifacts and noise. Image qualities (CT number accuracy, uniformity, SNR) were evaluated and compared between the reconstruction algorithms.

RESULTS

Images reconstructed by Katsevich's algorithm show better qualities, compared to ones by FDK algorithm and HU numbers have higher uniformity and accuracy. The HU-density calibration curve closely conforms to the manufacturer recommended values. The level of noise computed as the standard deviation in the phantom uniform region is 28.07 for the Katsevich algorithm, compared to 44.64 for the FDK algorithm.

CONCLUSIONS

Katsevich's exact reconstruction algorithm provided better image qualities than FDK for helical CBCT scans. This result will very useful for our ongoing investigation of helical CBCT, which would lead to improvement of CBCT longitudinal coverage of PTV and would be essential for future image-guided adaptive radiation therapy applications. Varian Research Agreement with Washington University in St. Louis.

SU-E-I-39: Intensive Monitoring of System Performance and Protocols on CT Systems

PURPOSE

The regulatory response to the recent CT overdoses is still in the process of being implemented. The more action taken before the regulations take effect reduces the exposure of the institution to medical and legal problems. Therefore, eighteen months ago we implemented an intensive QC program to ensure CT safety. The first component was to monitor the dose and SNR produced by a single CT examination weekly to identify changes. The second was to monitor changes in the protocols.

METHODS

We measured the CTDI and SNR for the routine head examination. We developed a script to identify and log changes in any of the 234 protocols.

RESULTS

Surprisingly, there were many changes to the protocols: on one unit 388 protocols were changed on 26 dates over the 18 month period. The reasons included: protocol optimization, scanner service, software reloads, mistakes, mistakenly loading the protocols from another section's scanner. The standard deviation of the CTDI was 2% to 3%. The dominant source of variation was probably positioning. The standard deviation of the SNR was 4% to 9%. The entire procedure requires 30min to 50min to test five CT scanners each week.

CONCLUSIONS

The standard deviations of the CTDI values and of the SNR values are sufficiently small to allow anomalies to be identified using these relatively simple methods. Frequent protocol checking is an important component to any QC program.

SU-E-I-12: Analysis of Image Noise in 3D Scatter Corrected Cone Beam CT Images as a Function of Object Size

PURPOSE

To characterize the spatial correlation of quantum noise of a cone beam CT (CBCT) system for breast imaging under scatter and non-scatter corrected conditions as a function of object radius.

METHODS

Experiments were conducted using a conical phantom under scatter and non-scatter corrected conditions. The cone phantom filled with 700ml of water was positioned near the isocenter of the quasi-monochromatic CBCT imaging system. 360 projections were acquired and beam-stop measured scatter correction was performed. Both scatter and non-scatter corrected projections were reconstructed with an iterative ordered subsets convex (OSC) algorithm. NPS measurements were made at several cone radii (6-12cm) using 4 overlapping 50×50 pixel ROI's with a 50% overlap between adjacent ROIs. An ensemble average of NPS measurements was taken over 5 slices at each radius.

RESULTS

For both scatter and non-scatter corrected data overall noise is higher with increasing object radius. Each 1-D NPS area under the curve was found to be within the standard deviation of the respective ROI variance. The variance of scatter corrected images is higher at each radius. The 1-D NPS indicates a decrease in peak value for both scatter and non-scatter corrected images as radius increases. Scatter corrected NPS shows a slight shift in the peak towards higher spatial frequencies, and a larger spread of noise over all spatial frequencies as radius increases.

CONCLUSIONS

This work demonstrates that scatter correction increases noise at each spatial frequency at various object radii. owever as radius increases, lower frequency noise decreases and concomitantly higher frequency noise increases. As a Result of this equalization of noise across spatial frequencies, the noise texture appears more uniform at larger radii under scatter corrected conditions. These results and methodology can be used to determine optimal detection of low contrast signals in phantoms of different radii and with clinical breast data.

WE-G-217BCD-11: A New Phantom to Study Combined Effects of In-Plane (x,y) and Z Axis Resolution for 3-D Imaging

PURPOSE

The aim of this work is to develop phantoms that can be used to sample the radial and 3D properties of a CT image, including in-plane (x,y) and z-axis information. The Phantom is amenable to mathematical analysis of the x, y, and z axis resolution properties separately and combined.

METHODS

A periodic pattern of a pair of opposed (30°) angled ramps is configured to produce a waveform profile across the CT image. A perfect CT image (with no loss of resolution) of the test object would produce a consistent geometric pattern of the intersection of a line with the pair of angled ramps. However, due to the finite resolution (x, y and z), the CT waveform profile will not yield the perfect profile; rather it will be influenced by slice thickness, and in-plane resolution (PSF, MTF), as well as noise limitations, and other sources of non-uniformity such as beam hardening etc. Various characteristics of the waveform profile including, amplitude, frequency, and slope (rate of climb) of the peaks, can be studied using mathematical analysis such as the Fourier transform. It will be shown how these performance characteristics are encoded in the wave pattern.

RESULTS

The waveform profiles are visually examined and mathematically analyzed, to demonstrate the effect of Slice Thickness (z axis) and changes of In-Plane (x,y) Resolution and non-uniformity across the image field; moreover, the harmonic analysis of the waveform is used to predict, either the in-plane resolution (MTF), or the z-axis MTF when one of the two is already known.

CONCLUSIONS

The Wave pattern phantom offers a way to consider 3-D imaging characteristics of a CT scanner by scanning a single repetitive test object that encodes both in-plane resolution and z-axis resolution and also offers a way to study non-uniformity effects throughout the CT plane (volume). DJG is a consultant to The Phantom Laboratory and Image OWL, Salem, NY. Funding of other authors is supplied by Image OWL Salem, NY.

SU-E-I-24: Determining the Optimal B-Values to Use in Diffusion Weighted Imaging for Differentiating Benign and Malignant Breast Lesions

PURPOSE

To investigate and determine the optimal b-values for Diffusion Weighted Imaging(DWI) for Apparent Diffusion Coefficient(ADC) maps in differentiating normal, benign and malignant breast tissue.

METHODS

Twenty-five patients underwent diffusion-weighted magnetic resonance imaging (DWI-MRI) and conventional breast MRI with suspicious breast fmdings(BIRADS >4). Breast lesions were defined by pathology. The DWI was acquired with different b-values ranging from 0,500, 600,750 and 1000s/mm² . Apparent Diffusion Coefficient(ADC) maps of breast tissue were constructed using different b-values, e.g., using either 2 b-values(0- 1000) or combinations of 3 or more (0,500,1000 or 0,600,750,1000). Quantitative analyses of the ADC maps of glandular, fatty and lesion tissue were obtained. Ratios of lesion to glandular tissue(L/GT) and signal to noise(SNR) were assessed. Paired t-tests were performed for statistical significance.

RESULTS

Eighteen patients had invasive ductal carcinoma and 7 had benign breast lesions. The mean ADC value for malignant lesions using all b values was 1.17±0.16×10-3mm² /s with a lesion to glandular(L/GT) ratio=0.65. The benign lesions, ADC map value was 1.86±0.03×10-3mm² /s with L/GT=0.98. There was a significant difference(P<0.05) between benign and malignant lesions ADC map value.The lowest SNR(12±6) was with single b-values. There was a significant difference(P<0.05) in SNR with multiple b-values(34±6) compared to single b-values. The highest SNR was given by using two b values greater than 500. Finally, the background noise for all combinations was surprising stable and ranged between 60±20%.

CONCLUSIONS

This is the first study to investigate the effect of changing different b values in DWI breast imaging. There were significant differences in the SNR between single and multiple b values. Our data show suggest that the recommended b-values for DWI in breast are 0, and two that are 500 or greater. Therefore use of at least 3 b-values in DWI/ADC mapping of breast lesions are needed for better characterization of benign and malignant breast tissue. P50CA88843, Avon Foundation for Women:01-2008-012, U01CA070095, andU01CA140204.

SU-E-I-46: Quantitative Effects of Clinical Practice on the Efficiency of CT Tube Current Modulation for Radiation Dose Reduction

PURPOSE

CT tube current modulation (TCM) represents one of the most important and efficient methods for radiation dose reduction and has been well accepted. However, the possible influence of clinical practice is likely to be ignored although venders may provide clinical protocols and instructions. This study is designed to further investigate the quantitative effects of clinical operation, including radiograph direction and patient positioning, on the efficiency of TCM by measuring patient radiation dose and image quality.

METHODS

An anthromorphologic chest phantom was scanned in a Sensation 40 and a lightSpeed 16 CT scanner, respectively, using routine chest protocols with TCM. We first investigated the effects of radiograph direction. Anterior-posterior (AP), lateral, and posterior-anterior (PA) directions were chosen. CTDIvol and dose-length-product (DLP) were recorded for analyses. Our second experiment studied the influence of patient position. First the phantom was positioned at the iso-center then scanned with AP direction. CTDIvols and DLPs were recorded as reference. Then the phantom was moved out of iso-center, up or down 2 and 4 cm respectively then scanned. CTDIvols and DLPs were recorded for comparison. For each setting, image noise was measured.

RESULTS

CTDIvol increases approximately 20% for PA direction, compared to AP or lateral direction which generates similar CTDIvol and DLP with TCM. Image noise for the PA direction is less than those for the AP or lateral directions. CTDIvol increases approximately 9% for LightSpeed 16 and 13% for Sensation 40 when the phantom was moved up 4 cm, while CTDIvol decreases approximately 5% for LightSpeed 16 and 8% for Sensation 40 when the phantom was moved down 4 cm.

CONCLUSIONS

Our quantitative study can direct clinical practice to improve the efficiency of CT tube current modulation and reduce patient radiation dose.

WE-G-217A-09: Phase Imaging Measurement of Static Magnetic Field Homogeneity

PURPOSE

The ACR MRI accreditation program requires measurement of the magnetic field homogeneity (MFH) in the annual QA. Full implementation of vendor methods can be very time consuming and incorporates techniques not available to clinical physicists. Conversely, many of the less involved techniques proposed tend to be less inaccurate and/or precise. Here, we propose a robust approach to MFH analysis using a simple phase mapping acquisition which is a time efficient compromise.

METHODS

The root mean square (r.m.s.) for MFH measurement is commonly used over multiple slices (∼24). In regions with high SNR, the MR signal can be reasonably assumed Gaussian distributed. Therefore, the standard deviation of phase values in ROI is equivalent to the r.m.s. of those. The standard deviations of phases in x, y and z are assumed uncorrelated. Thus, only axial, sagittal and coronal planes need to be acquired to approximate the MFH as opposed to a full 3D acquisition. To investigate this concept, phase images were acquired on four 1.5T clinical scanners and one 3.0 clinical scanner (MAGNETOM Espree, Siemens Medical Systems, Signa Excite HD 1.5T, and 3.0T GE Healthcare Technologies). The manufacturer phantoms were scanned using two echo times (delta_TE=TE2-TE1<5ms) using a simple 2D gradient echo acquisition to produce phase images. After acquiring phase difference images in three orthogonal planes, the standard deviation was calculated in three circular ROIs (Diameter=10, 20 and 30cm) in each plane, respectively, to estimate the MFH for the effective DSV.

RESULTS

The MFH values in five scanners using this method were within vendor specifications for the DSV. Additionally, the measured MFH values compared favorably with vendor planned maintenance records with <0.1 ppm discrepancy.

CONCLUSIONS

This proposed method may be a reliable and practical for regular MFH measurement in QA programs and providing an independent check of the vendor measurement.

WE-E-213CD-08: A Novel Level Set Active Contour Algorithm Using the Jensen-Renyi Divergence for Tumor Segmentation in PET

PURPOSE

Positron emission tomography (PET) presents a valuable resource for delineating the biological tumor volume (BTV) for image-guided radiotherapy. However, accurate and consistent image segmentation is a significant challenge within the context of PET, owing to its low spatial resolution and high levels of noise. Active contour methods based on the level set methods can be sensitive to noise and susceptible to failing in low contrast regions. Therefore, this work evaluates a novel active contour algorithm applied to the task of PET tumor segmentation.

METHODS

A novel active contour segmentation algorithm based on maximizing the Jensen-Renyi Divergence between regions of interest was applied to the task of segmenting lesions in 7 patients with T3-T4 pharyngolaryngeal squamous cell carcinoma. The algorithm was implemented on an NVidia GEFORCE GTV 560M GPU. The cases were taken from the Louvain database, which includes contours of the macroscopically defined BTV drawn using histology of resected tissue. The images were pre-processed using denoising/deconvolution.

RESULTS

The segmented volumes agreed well with the macroscopic contours, with an average concordance index and classification error of 0.6 ± 0.09 and 55 ± 16.5%, respectively. The algorithm in its present implementation requires approximately 0.5-1.3 sec per iteration and can reach convergence within 10-30 iterations.

CONCLUSIONS

The Jensen-Renyi active contour method was shown to come close to and in terms of concordance, outperforms a variety of PET segmentation methods that have been previously evaluated using the same data. Further evaluation on a larger dataset along with performance optimization is necessary before clinical deployment.

SU-E-J-151: Evaluation of a Real Time Tumour Autocontouring Algorithm Using In-Vivo Lung MR Images with Various Contrast to Noise Ratios

PURPOSE

To quantitatively evaluate a lung tumour autocontouring algorithm using in-vivo lung cancer patient MR images with varying contrast to noise ratios (CNR) simulating images acquired at various MR field strengths.

METHODS

A non small cell lung cancer patient with posterior lung tumour is imaged (sagittal plane) in a 3T MRI using a dynamic bSSFP sequence (FOV: 40×40cm² , voxel size: 3.1×3.1×20mm³ , TE = 1.1ms. TR = 2.2ms, 275ms per image) under free breathing for approximately 3 minutes (650 images). Gaussian random noise is added to the 3T images to approximately simulate the equivalent CNR in images acquired at 1.5T, 1.0T, 0.5T, 0.3T and 0.2T. The moving tumour in all 3T images is contoured by a physician for reference. The first 20 of these manual contours are used for the parameters optimization of auto-contouring algorithm. The automatic contours from the remaining images are quantitatively compared with the physician's contours using the centroid's displacement and the Dice's coefficient (DC).

RESULTS

The oncologist's contours of the 3T images show a maximum S-I motion of 26mm. Compared to the oncologist's contours, automatic contours have an average centroid displacement of 1.37mm, and an average DC of 0.881. The autocontouring algorithm's performance with images in the range of 1.5T to 0.5T equivalent CNRs is similar to that of the 3T data. However, for the lowest CNR datasets (0.2, 0.3T) an increase in centroid displacement and decrease in DC is observed, with mean displacements of 1.56mm, 1.71mm and DCs of 0.870, 0.836 for the 0.3T and 0.2T dataset, respectivelyConclusions: With in-vivo MR images, the autocontouring algorithm generated lung tumour contours similar to ones drawn by a physician (DC 〉 0.83). In this patient, additional CNR from 〉0.5T MRIs does not provide statistically significant improvement in the accuracy of our autocontouring software. E.Yip is supported by the Canadian Institutes of Health Research as well as Alberta Innovates - Health Solutions.

TU-E-217BCD-02: An X-Ray Scatter Correction Method for Dedicated Breast Computed Tomography

PURPOSE

To reduce the impact of x-ray scatter in dedicated breast computed tomography (BCT) images.

METHODS

The inclusion of x-ray scatter in BCT projections results in cupping artifacts, loss of contrast, and quantitative inaccuracies. To correct for this, an additional set of BCT projections is acquired with a tungsten plate placed between the x-ray source and the patient breast. The tungsten plate includes a two-dimensional grid of perforations to generate an array of pencil beams. Due to the limited area illuminated by the x-ray pencil beams, an array of signals of primary x-rays only is obtained. At the pencil beam locations, the difference between the plate projections and the standard projections is an estimate of the scatter present in the latter. These estimates are interpolated to obtain scatter-only estimates of the whole images, which are subtracted from the standard projections, resulting in BCT projections with primary signal only, which are then reconstructed. To reduce the impact of the quantum noise of the scatter signal, the resulting reconstructions are noise filtered. Monte Carlo simulations were performed to estimate the amount of scatter included in the pencil beams and the dose from these additional projections. The algorithm was tested using breast phantoms on a BCT clinical prototype system.

RESULTS

The maximum scatter signal in the pencil beams is 2.2% (mean of 0.7%) of the total signal, so the pencil beams are an excellent estimate of the primary-only signal. The additional projections Result in only 0.4% of the glandular dose of the standard projections. The homogeneity of the resulting phantom images, the signal difference between adipose and glandular tissue, reconstruction accuracy, and contrast-to-noise ratios were improved with this algorithm.

CONCLUSIONS

The proposed algorithm has the potential to substantially improve BCT image quality with practically no additional dose to the patient breast.

TU-E-217BCD-01: X-Ray Spectra for Breast CT

PURPOSE

To study a range of x-ray spectra with regard to their suitability for dedicated breast CT, taking into account realistic tube loading limitations.

METHODS

A simple theoretical model was used to get a preliminary indication of the best mono-energetic x-ray energy as a function of object size. That model was extended to poly-energetic spectra and used to study a range of object sizes, tube voltages, filter materials, and filter thicknesses. Spectra from IPEM Report 78 were used as input to the model. Initially we have looked at lanthanide filters (Z = 57-60) and tube voltages from 30 to 80 kVp. Outputs from the model included contrast, contrast-to-noise ratio (CNR), dose, dose efficiency (CNR̂2/dose), and tube loading. Dose was estimated as incident minus transmitted energy fluence. We have also started experiments to verify the computational results. Experiments are done using a bench-top cone beam CT and a 14 cm diameter cylindrical PMMA phantom filled with liquid mixtures equivalent to breast tissue of varying glandularity. We use reconstructed data to determine the CNR of a test block representing 100% glandular breast tissue, immersed in the liquid. Air kerma measurements are obtained at the center and periphery of the phantom, and used in the calculation of dose efficiency. Monte Carlo calculations are being done to get a more accurate dose estimate.

RESULTS

For a constant CNR, computational results indicate that as filter thickness increases above 0.4 mm dose efficiency plateaus. However, the power required to maintain a constant CNR becomes prohibitively large for filter thicknesses greater than 0.3 mm.

CONCLUSIONS

Spectra generated at tube voltages above 40 kVp are likely to provide the best trade-off between dose efficiency and tube loading. For lanthanide filters, there is little benefit from using thicknesses greater than 0.3 mm.

WE-G-217A-05: Automatic Method for RF Coil Assessment in Clinical MRI: A Three-Dimensional Approach

PURPOSE

MRI RF coil assessment is usually evaluated with region-of-interest (ROI) analysis from a single 2D phantom image. This simple approach has worked well for large volume coils or phased-array coil with large receivers, but not the high density phased-array coils characterized by 3D array arrangement of their multiple receivers. This abstract proposes a novel approach for quantitative coil assessment based on 3D imaging and 3D ROI analysis.

METHODS

To characterize all receivers of the coil of interest, a large uniform phantom (preferably a corresponding anthropometric phantom) and a large 3D geometric coverage fully includes the coil sensitivity volume was applied during MR imaging. After imaging, data from all receivers were used to reconstruct a composite 3D image, and to reconstruct 3D images from each individual receiver, leading to a total of N+1 3D image datasets (where N is the number of coil channels). IDL programs were developed to automatically perform ROI analysis on the composite image and on the individual receiver images. Instead of choosing one single 2D slice out of each 3D dataset, the whole 3D dataset was treated as a 3D image, and 3D ROIs were automatically generated for coil assessment.

RESULTS

This 3D coil evaluation approach could be applied to all clinical coils including quadrature body/head coils, and phased-array coils with 2 to 32 channels. 3D sensitivity map could be generated to check receiver function visually. 3D mean SNR, max SNR, and uniformity could be obtained from composite and individual channel 3D images fully automatically. Coil/receiver performance assessment was very fast and straightforward, regardless of the number of receivers of the coil.

CONCLUSIONS

3D imaging in combination with 3D automatic ROI analysis is a fast, convenient, and less subjective approach for quantitative coil assessment, particularly for high density phased-array coils.

Poster - Thurs Eve-18: Performance evaluation of MV CT imaging on the HI ART II tomotherapy unit

The HI-ART II unit (TomoTherapy Inc., Madison WI) is a modality used by the London Regional Cancer Program (LRCP) for radiation therapy. This machine uses the same source of Megavoltage energy radiation to image (3.5 MV) and to treat (6MV) patients, combining the functionality of a traditional linear accelerator and CT simulator into one unit. Thus, it is possible to assess patient positioning and adjust for anatomy changes just prior to radiation therapy. Unfortunately, at MV energy levels, the physics of radiation interaction limits image quality, and gives rise to an inherent dose limitation concern that enhances noise levels. Therefore, we propose to quantify the image quality produced by the HI-ART II unit using techniques established for kVCT scanner technology. Our study involved the use of three standard phantoms to test image resolution, noise, uniformity, and linearity for a 512 × 512 reconstruction matrix and three scan pitch settings (0.8, 1.6, and 2.4). Results follow: linearity between MV CT number versus relative electron density was observed, noise calculations ranged from 2.15-2.51%, and a distinct central artifact was revealed during uniformity testing. The linearity between MV CT number versus relative electron density implies that MV CT images are highly suitable for dose calculations. MV CT image quality of uniform phantoms were acceptable and demonstrated noise levels higher than those produced by kVCT simulators. Further study is necessary to correct for the central artifact in MV CT images.

Sci-Sat AM(1): Imaging-05: Analytical scatter estimation for cone-beam computed tomography

A significant challenge to the implementation of cone-beam computed tomography (CBCT) for high-resolution imaging is the high scatter to primary ratio. Scatter causes cupping and shading artifacts, increased noise and decreased contrast in reconstructed images. Methods to reduce the impact of scatter in CBCT are thus very desirable. We are investigating methods for computational scatter estimation and compensation for CBCT, with the goal of incorporating a scatter estimator within a statistical reconstruction algorithm. We have developed an analytical method for estimating single scatter, based on Klein-Nishina cross-sections. We have compared scatter estimates generated with this method with the results of high-count EGSnrc Monte Carlo simulations. The analytical estimates compare favorably with the Monte Carlo estimates. The paper will discuss our method for analytical estimation of single scatter, including the assumptions and simplifications required to render it computationally tractable, along with the results of the comparison between the analytical method and Monte Carlo simulations. The paper will extend previous results obtained with small (40 × 40 × 40 voxel) homogeneous computational phantoms to include results for larger, more clinically relevant phantoms (128 × 128 × 128 voxels, simulated 50/50 breast tissue with inserts of varying contrast). The paper will also discuss computational acceleration obtained through the use of parallel processing via the WestGrid High-Performance Computing network.