Parinaz Massoumzadeh

Properties of preprocessed sinogram data in x-ray computed tomography

The accurate determination of x-ray signal properties is important to several computed tomography (CT) research and development areas, notably for statistical reconstruction algorithms and dose-reduction simulation. The most commonly used model of CT signal formation, assuming monoenergetic x-ray sources with quantum counting detectors obeying simple Poisson statistics, does not reflect the actual physics of CT acquisition. This paper describes a more accurate model, taking into account the energy-integrating detection process, nonuniform flux profiles, and data-conditioning processes. Methods are developed to experimentally measure and theoretically calculate statistical distributions, as well as techniques to analyze CT signal properties. Results indicate the limitations of current models and suggest improvements for the description of CT signal properties.

Noise simulation in x-ray CT

A method is presented for accurately simulating the effects of dose reduction in x-ray computed tomography (CT) by adding synthetic noise to raw projection data. A model for realistic noise in projection data was generated, incorporating the mechanisms of stochastic noise in energy-integrating x-ray detectors, electronic system noise, and bowtie beam filtering (used for patient dose reduction). Parameters for the model were extracted from phantom measurements on a variety of clinical CT scanners (helical single row, four-row, and 16-row). Dose reduction simulations were performed by adding synthetic noise based on the noise model to raw data acquired from clinical scanners. Qualitative and quantitative validation of the process was accomplished by comparing phantom scans acquired under high and low dose conditions with simulated imagery. The importance of including alternative noise mechanisms (bowtie filter and system noise) was demonstrated. Henceforth, scans of clinical patients were acquired using conventional protocols; through simulations, image sets were presented at a variety of lower dose procedures. The methodology promises to be a useful tool for radiologists to explore dose reduction protocols in an effort to produce diagnostic images with radiation dose “as low as reasonably achievable”.

Aerobic Glycolysis as a Marker of Tumor Aggressiveness: Preliminary Data in High Grade Human Brain Tumors

Objectives. Glucose metabolism outside of oxidative phosphorylation, or aerobic glycolysis (AG), is a hallmark of active cancer cells that is not directly measured with standard 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET). In this study, we characterized tumor regions with elevated AG defined based on PET measurements of glucose and oxygen metabolism. Methods. Fourteen individuals with high-grade brain tumors underwent structural MR scans and PET measurements of cerebral blood flow (CBF), oxygen (CMRO2) and glucose (CMRGlu) metabolism, and AG, using 15O-labeled CO, O2 and H2O, and FDG, and were compared to a normative cohort of 20 age-matched individuals. Results. Elevated AG was observed in most high-grade brain tumors and it was associated with decreased CMRO2 and CBF, but not with significant changes in CMRGlu. Elevated AG was a dramatic and early sign of tumor growth associated with decreased survival. AG changes associated with tumor growth were differentiated from the effects of nonneoplastic processes such as epileptic seizures. Conclusions. Our findings demonstrate that high-grade brain tumors exhibit elevated AG as a marker of tumor growth and aggressiveness. AG may detect areas of active tumor growth that are not evident on conventional FDG PET.

MO‐D‐I‐611‐07: The Influence of Bowtie Filters On X‐Ray CT Signals

Purpose: Bowtie filters are commonly employed in CT scanners to minimize radiation dose by reducing intensity variations across detector elements in the presence of patient anatomy. This filtration modifies a number of x‐ray beam properties (effective energy, flux, first and second order statistics), making them non‐uniform across the fan beam field of view. Because these phenomena are not usually included in analysis of CT performance, this presentation will quantify the influence of bowtie filters on sinogram measurements and demonstrate their effect in reconstructed images.Method and Materials: A model developed for energy integrating x‐ray detectors, extended to bowtie filters, was used to compute signal means, variances, and beam quality for a realistic scanner configuration. Experimental measurements were acquired with cylinder phantoms located off‐isocenter, allowing objects to sample different portions of the fan beam. Using measured bowtie filter profiles and clinical CT patient scan data, simulated dose reduction images were generated to show visual effects. This approach was used to study a novel dose reduction method, region of interest (ROI) scanning, wherein full intensity is applied only to a local volume of interest while surrounding tissue receives a significantly lower dose. Results: The dominant effect of the bowtie filter is to increase noise in the periphery of the image field. A variation of effective energy results in a small amount of nonlinearity, which can be effectively corrected through calibration. Differences in second order statistics are at the threshold of observer detection. ROI scanning achieves very good local image quality while reducing dose in surrounding tissue.Conclusion: When bowtie filters are properly implemented, they provide reduction of patient exposure with minimal image degradation. The impact of bowtie filters on CT signals is significant and must be accounted for in accurate modeling of the scanning process.

Comparison of Perfusion- and Diffusion-weighted Imaging Parameters in Brain Tumor Studies Processed Using Different Software Platforms

RATIONALE AND OBJECTIVES To compare quantitative imaging parameter measures from diffusion- and perfusion-weighted imaging magnetic resonance imaging (MRI) sequences in subjects with brain tumors that have been processed with different software platforms. MATERIALS AND METHODS Scans from 20 subjects with primary brain tumors were selected from the Comprehensive Neuro-oncology Data Repository at Washington University School of Medicine (WUSM) and the Swedish Neuroscience Institute. MR images were coregistered, and each subjects data set was processed by three software packages: 1) vendor-specific scanner software, 2) research software developed at WUSM, and 3) a commercially available, Food and Drug Administration-approved, processing platform (Nordic Ice). Regions of interest (ROIs) were chosen within the brain tumor and normal nontumor tissue. The results obtained using these methods were compared. RESULTS For diffusion parameters, including mean diffusivity and fractional anisotropy, concordance was high when comparing different processing methods. For perfusion-imaging parameters, a significant variance in cerebral blood volume, cerebral blood flow, and mean transit time (MTT) values was seen when comparing the same raw data processed using different software platforms. Correlation was better with larger ROIs (radii ≥ 5 mm). Greatest variance was observed in MTT. CONCLUSIONS Diffusion parameter values were consistent across different software processing platforms. Perfusion parameter values were more variable and were influenced by the software used. Variation in the MTT was especially large suggesting that MTT estimation may be unreliable in tumor tissues using current MRI perfusion methods.

Diffusion Tensor Imaging as a Biomarker to Differentiate Acute Disseminated Encephalomyelitis From Multiple Sclerosis at First Demyelination

BACKGROUND There are no clinical features or biomarkers that can reliably differentiate acute disseminated encephalomyelitis from multiple sclerosis at the first demyelination attack. Consequently, a final diagnosis is sometimes delayed by months and years of follow-up. Early treatment for multiple sclerosis is recommended to reduce long-term disability. Therefore, we intend to explore neuroimaging biomarkers that can reliably distinguish between the two diagnoses. METHODS We reviewed prospectively collected clinical, standard MRI and diffusion tensor imaging data from 12 pediatric patients who presented with acute demyelination with and without encephalopathy. Patients were followed for an average of 6.5 years to determine the accuracy of final diagnosis. Final diagnosis was determined using 2013 International Pediatric MS Study Group criteria. Control subjects consisted of four age-matched healthy individuals for each patient. RESULTS The study population consisted of six patients with central nervous system demyelination with encephalopathy with a presumed diagnosis of acute disseminated encephalomyelitis and six without encephalopathy with a presumed diagnosis of multiple sclerosis or clinically isolated syndrome at high risk for multiple sclerosis. During follow-up, two patients with initial diagnosis of acute disseminated encephalomyelitis were later diagnosed with multiple sclerosis. Diffusion tensor imaging region of interest analysis of baseline scans showed differences between final diagnosis of multiple sclerosis and acute disseminated encephalomyelitis patients, whereby low fractional anisotropy and high radial diffusivity occurred in multiple sclerosis patients compared with acute disseminated encephalomyelitis patients and the age-matched controls. CONCLUSIONS Fractional anisotropy and radial diffusivity measures may have the potential to serve as biomarkers for distinguishing acute disseminated encephalomyelitis from multiple sclerosis at the onset.

TU-C-304A-09: Radiation Dose Reduction for Pediatric CT

Purpose: The purpose of this study was to evaluate potential radiation dose reduction for pediatric CT.Materials and Methods: A dose‐reduction simulation tool, which adds synthetic noise to raw projection measurements and reconstructsimages at a simulated lower dose (Massoumzadeh, et al. Med. Phys. Vol. 36, pp. 174–189, 2009), was used to simulate low‐dose CTimages. Simulated low‐dose CTimages are created from full‐dose CTimages of normal and pathological slices (lung nodules and abdominal visceral lesions). The amount of added noise was task dependent, with 10 sets of simulated low dose ranged from 1% to 85% of original dose. Sixteen pediatric cases were selected, including eight normal, three patients with pulmonary nodules, two patients with abdominal visceral organ lesions, and three patients with appendicitis without perforation. All studies performed on a 16‐row scanner, with the effects of tube current modulation and bow tie filters included. Following a short training session, 19 volunteer radiologists, from various clinical centers in the world who were attending the International Pediatric Radiology conference in Montreal in 2006, used a 5‐point scale to rate a sequence of simulated, low‐to‐high exposure images for the presence or absence of lesions. They were total of 176 images with viewing sessions limited to 45 minutes. Diagnostic agreement between full‐dose and reduced‐dose images was assessed with a weighted Kappa statistic. Result: For detection of pulmonary nodules, a decrease in the average intra‐observer agreement (kappa= 0.90) was found at 80% dose reduction, while for the detection of abdominal lesions or appendicitis a 50% reduction was observed. Conclusion: There is potential for dose reduction in CT studies, which is task dependent and greater for pulmonary nodule than for abdominal visceral lesions and appendicitis. The noise simulation methodology is a powerful tool to help understand the relationship among dose,noise, and observer agreement.

WE‐D‐332‐06: Validation of CT Dose‐Reduction Simulation

Purpose: To develop and validate a CTdose reduction simulation model. Method and Materials: A noise model was developed incorporating the mechanisms of stochastic noise in energy‐integrating x‐ray detectors, tube current modulation, bowtie beam filtering, and electronic system noise by adding synthetic noise to projection data (sinogram).Experiments were performed to determine the parameters required for the noise model, and the effects of various components were studied. Various scans were performed on an empty gantry, cylinder phantom, cadaver head, and a skull (an asymmetric object) at varyious flux levels. Seventeen clinical scans from three different centers were included. As a validation, the outputs of the simulations were compared to actual measurements in both the sinogram and image domain. Four‐alternative forced‐choice (4AFC) observer studies were performed to confirm the realistic appearance of simulated images. Tests were conducted to establish the “just noticeable difference (JND)” in noise levels, and the sensitivities of observers to changes in noise levels were determined. Results: The gaussian random noise generator was found to be appropriate for simulations. Measurements demonstrated a match of the noise variance to within 5% in the sinogram domain, which propagates into the image domain. The 4AFC observer studies indicated that the simulated imagery was realistic, with no detectable difference between simulated and original imagery (25%±7.9%). The JND studies indicated that observers reliably detected noise‐levels differences corresponding to 20–30% changes in tube current, implying that accuracies in simulation on the order of ∼9% would result in images that could not be reliably differentiated from original images.Conclusion: The dose‐reduction simulation tool demonstrated excellent image fidelity. The methodology promises to be a useful tool for radiologists to explore dose reduction protocols in efforts to produce diagnostic images with radiation dose “as low as reasonably achievable” (ALARA).

SU‐GG‐I‐36: Measuring CT Incident Flux Using CT Sinogram Data

Purpose: The value of incident flux in a multi detector CTscanner is necessary to model noise properties of a CT scan. A novel method is proposed to determine the incident flux of any clinical CT scan by utilizing direct exposure (air) regions of the sinogram data. Method and Materials: 17 clinical patient scans, with tube current modulation on (11 scans) and off (6 scans), were collected from three sixteen‐row scanners at different locations. In addition a non‐symmetric object (skull with contrast) was scanned with two mAs settings (low and high), and with tube modulation on and off. Air regions of sinogram data were segmented by setting a threshold and collecting samples of direct exposures. Meanwhile, the extracted data were normalized using predetermined information about Case‐a: the bowtie profile, or Case‐b: the bowtie profile and the tube current (which was extracted from the header of CTsinogram data file). The variance of the gradient of the selected points of normalized data in transmission space as a function of gantry angle was obtained, which is inversely proportional to the incident flux. With this innovative method the result of Case‐a is proportional to the tube current, and the result of Case‐b is a constant scaling factor(K) relating incident flux to tube current for a given CTscanner.Results: The rms relative error between predicted and measured K was found to be 2.2%. The standard deviation in the value of K was 4.5%, indicating the scan‐to‐scan variation in flux scaling for an individual scanner. Computing a mean inter‐scanner variation indicates calibration of individual scanners is required to achieve simulation noise accuracies less than 5%. Conclusion: Utilizing this method, the incident flux or scaling factor characterizing incident flux for particular CTscanner can be computed.

TU-D-L100J-02: Noise Stationarity in Spiral CT

Purpose: The properties of noise in CTimages are important in system design, algorithm development, and assessment of diagnostic observer performance, as well as for developing accurate dose reduction simulators. Conventional analysis assumes noise stationarity, even though it is well known that elements in a CT scanner, such as bowtie filters or tube‐current modulation, as well as filtered backprojection reconstruction, introduce variation in measurement statistics. In this presentation, two additional contributors to nonstationarity, hardware component variation and data interpolation, are analyzed, and the properties of noise in CTimages are characterized. Method and Materials: Raw sinogram data was collected from open gantry scans. Offline reconstruction software, with access to data at individual processing steps, such as linear interpolation, was used for image reconstruction. White noise data was injected at various stages of reconstruction and changes in statistical properties at subsequent processing steps measured. Multiple simulations were used to create data ensembles for analysis. Covariance measures were calculated in the sinogram and image domains. Results: Raw CT data is nonstationary during the acquisition cycle, with covariances as high as 20% in unattenuated signals. Image reconstruction steps, such as linear interpolation and filtered backprojection, introduce variations in noise power on the order of +/− 50% in sinogram data. The variation of noise power can be up to 600% across a uniform image. The off‐diagonal covariance coefficients are on the order of 10% with nearest neighbor pixels. Conclusion: The conventional assumption of noise stationarity in CTimages needs to be extended to include additional processes that introduce local variations in image statistics.