Noise reduction in nuclear medicine images.

Common methods of reducing random noise in nuclear medicine use lowpass filtering, which has the disadvantage that it affects high-frequency components of the image. We developed a noise-reduction approach that estimates signal and noise levels in each of several frequency bands and removes the appropriate amount of noise with little effect on the signal in each band.

A comparison of 180 degrees and 360 degrees acquisition for attenuation-compensated thallium-201 SPECT images.

UNLABELLED: This study compared attenuation compensated, myocardial SPECT images reconstructed from 180 degrees and 360 degrees data to determine if either data acquisition method might yield improved image quality. Specifically, this study analyzed how the use of either 180 degrees or 360 degrees data affects: (a) the relative count density distribution, (b) defect contrast and (c) level of statistical noise in the left ventricular (LV) wall in the reconstructed SPECT images. METHODS: Using the three-dimensional MCAT phantom simulating 201Tl uptake in the upper torso and the SIMSET Monte Carlo code, noise-free projection datasets for both 180 degrees (45 degrees LPO to 45 degrees RAO) and 360 degrees acquisition were generated with the effects of nonuniform attenuation, collimator-detector response and scatter. In addition, low-noise experimental phantom data were acquired over 180 degrees and 360 degrees. Assuming the same total acquisition time, four sets of noisy projection data were simulated from scaled noise-free, simulated data for the following acquisitions: (a) 180 degrees and (b) 360 degrees data acquired on a 90 degrees dual-detector system and (c) 180 degrees and (d) 360 degrees data acquired on a 120 degrees triple-detector system. For each of the four acquisition schemes, 400 realizations of noisy projection data were generated, and the normalized s.d. in the reconstructed images was calculated for five ROIs in the LV wall. Images were reconstructed with nonuniform attenuation compensation using ML-EM algorithm for 25, 50 and 75 iterations. RESULTS: Both the simulated noise-free and experimental low-noise images reconstructed from 180 degrees and 360 degrees data showed nearly identical count densities and defect contrasts in the LV wall. For the 90 degrees dual-detector system, 180 degrees images showed less noise, while for the 120 degrees triple-detector system, 360 degrees showed less noise; however, these differences in noise level were extremely small after a smoothing filter was applied. The 180 degrees images acquired with the 90 degrees dual-detector system showed the same noise level as the 360 degrees images acquired with the 120 degrees triple-detector system, so neither system geometry had an advantage with respect to reduced noise in the SPECT images. CONCLUSION: When nonuniform attenuation compensation is included in the reconstruction, the count density in the LV wall is nearly identical for 180 degrees and 360 degrees SPECT images, and the 90 degrees dual-detector and 120 degrees triple-detector SPECT systems produced similar SPECT images for the same total acquisition time.

Intrinsic dual-energy processing of myocardial perfusion images.

UNLABELLED: We have developed a software-based method for processing dual-energy 201TI SPECT emission projection data with the goal of calculating a spatially dependent index of the local impact of gamma-ray attenuation. We refer to this method as intrinsic dual-energy processing (IDEP). METHODS: IDEP exploits the differential attenuation of lower energy emissions (69-83 keV) and higher energy emissions (167 keV) resulting from the decay of 201TI to characterize the relative degree of low-energy gamma-ray attenuation throughout the myocardium. In particular, IDEP can be used to estimate the relative probability that a low-energy gamma-ray emitted from a particular region of the myocardium is detected during the acquisition of SPECT projection data. Studies on phantoms and healthy human volunteers were performed to determine whether the IDEP method yielded detection probability images with systematic structure visible above the noise of these images and whether the systematic structure in the detection probability images could be rationalized physically. In patient studies, the relative regional detection probabilities were applied qualitatively to determine the likely effects of attenuation on the distribution of mapped photon emissions. RESULTS: Measurements of the detection probability in uniform phantoms showed excellent agreement with those obtained from computer simulations for both 180 degrees and 360 degrees acquisitions. Additional simulations with digital phantoms showed good correlation between IDEP-estimated detection probabilities and calculated detection probabilities. In patient studies, the IDEP-derived detection probability maps showed qualitative agreement with known nonuniform attenuation characteristics of the human thorax. When IDEP data were integrated with the findings on the emission scan, the correlation with coronary anatomy (known in 6 patients and hypothesized on the basis of clinical and electrocardiographic parameters in 5 patients) was improved compared with evaluating the mapped emission image alone. CONCLUSION: The IDEP method has the potential to characterize the attenuation properties of an object without use of a separate transmission scan. Coupled with the emission data, it may aid coronary diagnosis.

Description of a prototype emission-transmission computed tomography imaging system.

We have developed a prototype imaging system that can perform simultaneous x-ray transmission CT and SPECT phantom studies. This system employs a 23-element high-purity-germanium detector array. The detector array is coupled to a collimator with septa angled toward the focal spot of an x-ray tube. During image acquisition, the x-ray fan beam and the detector array move synchronously along an arc pivoted at the x-ray source. Multiple projections are obtained by rotating the object, which is mounted at the center of rotation of the system. The detector array and electronics can count up to 10(6) cps/element with sufficient energy-resolution to discriminate between x-rays at 100-120 kVp and gamma rays from 99mTc. We have used this device to acquire x-ray CT and SPECT images of a three-dimensional Hoffman brain phantom. The emission and transmission images may be superimposed in order to localize the emission image on the transmission map.

Quality control of scintillation cameras using a minicomputer.

A minicomputer-based technique compiles objective indicators of scintigraphic system performance. The evaluation begins with the acquisition of a single image of an orthogonal hole pattern from which quantitative and regional measurements of point-source sensitivity, spatial resolution, and spatial linearity are derived. Two computer programs offer the user different but complementary features. The first program is the basis of an evaluation performed by a technologist for purposes of quality control. Operator intervention is minimal, and the entire protocol, including data acquisition and processing, can be completed in 20 min. The results are automatically compiled and displayed as graphs showing 100 consecutive sets of daily performance measurements. A second computer program is designed as an interactive diagnostic and research tool to display measurements as histograms and functional images. The operator can use the program to determine the quantitative and spatial characteristics of the system's intrinsic performance measurements made during the quality-control evaluations.

Absolute quantification of regional myocardial uptake of 99mTc-sestamibi with SPECT: experimental validation in a porcine model.

UNLABELLED: We have evaluated a method for absolute in vivo quantification of 99mTc-sestamibi uptake in a porcine model of myocardial perfusion. METHODS: Correlated CT and radionuclide images were obtained from eight adult pigs using a combined CT-SPECT imaging system. In each case, the CT image is used to generate an object-specific attenuation map that is incorporated into an iterative algorithm for reconstruction and attenuation correction of the radionuclide image. Anatomic information available from the correlated CT image is used to correct the radionuclide image for partial-volume errors by mathematically modeling the radionuclide imaging process. A volume of interest, or template, that approximates the geometric extent of the myocardium is defined from the CT image. Once defined, the template is assigned unit activity and is mathematically projected using a realistic physical model of the radionuclide imaging process including nonideal collimation and object-specific attenuation. The template is then reconstructed from these projections to obtain a pixel-by-pixel partial-volume correction for the myocardium in the radionuclide image. The CT image is also used to delimit the anatomic boundaries of the myocardium for quantification of the radionuclide images. The pixel intensities in the corrected radionuclide image are calibrated in units of activity concentration (MBq/g) and compared with the ex vivo activity concentration measured directly from the excised myocardium. RESULTS: Without corrections, the measured in vivo activity concentration in the porcine myocardium was only 10% of the true value. Correcting for object-specific attenuation improved the accuracy of this measurement but resulted in values that were still only 42% of the true value. By correcting for both attenuation and partial-volume errors, we were able to achieve absolute quantification with an accuracy error near 10%. CONCLUSION: We have shown that, by applying object-specific attenuation corrections and suitable partial-volume corrections, absolute regional activity concentration can be measured accurately in the porcine myocardium.

Neuroblastoma imaging using a combined CT scanner-scintillation camera and 131I-MIBG.

UNLABELLED: High-dose administration of 131I-metaiodobenzylguanidine (131I-MIBG) continues to be a promising treatment for neuroblastoma. However, currently used methods of estimating 131I-MIBG uptake in vivo may be too inaccurate to properly monitor patient radiation exposure doses. To improve localization and uptake measurements over currently practiced techniques, we evaluated different methodologies that take advantage of the correlated patient data available from a combined CT-scintillation camera imaging system. METHODS: Serial CT and radionuclide scans of three patients were obtained on a combined imaging system. SPECT images were reconstructed using both filtered backprojection and maximum-likelihood expectation maximization (MLEM). Volumes of interest (VOIs) were defined on anatomic images and automatically correlated to spatial volumes in reconstructed SPECT images. Several radionuclide quantification methods were then compared. First, the mean reconstructed values within coregistered SPECT VOIs were estimated from MLEM reconstructed images. Next, we assumed that reconstructed activity in SPECT voxels were linear combinations of activities present in individual objects, weighted by geometric factors derived from CT images. After calculating the weight factors by modeling the SPECT imaging process with anatomically defined VOIs, least-squares fitting was used to estimate the activities within lesion volumes. We also estimated the lesion activities directly from planar radionuclide images of the patients using similar linearity assumptions. Finally, for comparison, lesion activities were estimated using a standard conjugate view method. RESULTS: Activities were quantified from three patients having a total of six lesions with volumes ranging from 0.67 to 117 mL. Methods that used CT data to quantify lesion activities gave similar results for planar and tomographic radionuclide data. Estimating activity directly from mean VOI values in MLEM-reconstructed images alone consistently provided estimates lower than CT-aided methods because of the limited spatial resolution of SPECT. Values obtained with conjugate views produced differences up to fivefold in comparison with CT-aided methods. CONCLUSION: These results show that anatomic information available from coregistered CT images may improve in vivo localization and measurement of 131I-MIBG uptake in tumors.

Myocardial perfusion imaging with a combined x-ray CT and SPECT system.

UNLABELLED: We evaluated a novel combined x-ray CT and SPECT medical imaging system for quantitative in vivo measurements of 99mTc-sestamibi uptake in an animal model of myocardial perfusion. METHODS: Correlated emission-transmission myocardial images were obtained from 7- to 10-kg pigs. The x-ray CT image was used to generate an object-specific attenuation map that was incorporated into an iterative ML-EM algorithm for reconstruction and attenuation correction of the coregistered SPECT images. The pixel intensities in the SPECT images were calibrated in units of radionuclide concentrations (MBq/g), then compared against in vitro 99mTc activity concentration measured from the excised myocardium. In addition, the coregistered x-ray CT image was used to determine anatomical boundaries for quantitation of myocardial regions with low perfusion. RESULTS: The accuracy of the quantitative measurement of in vivo activity concentration in the porcine myocardium was improved by object-specific attenuation correction. However, an additional correction for partial volume errors was required to retrieve the true activity concentration from the reconstructed SPECT images. CONCLUSION: Accurate absolute SPECT quantitation required object-specific correction for attenuation and partial volume effects. Additional anatomical information from the x-ray CT image was helpful in defining regions of interest for quantitation of the SPECT images.

Noise, resolution, and sensitivity considerations in the design of a single-slice emission-transmission computed tomographic system.

A prototype Emission-Transmission Computed Tomography (ETCT) system is being developed that will acquire single-slice x-ray transmission CT images simultaneously with single photon emission computed tomography (SPECT) images. This system will permit the correlation of anatomical information from x-ray CT with functional information from SPECT images. The patient-specific attenuation map derived from the x-ray CT images can be used to perform attenuation correction of the SPECT images, so that accurate quantitative information can be obtained. The fan-beam scanning geometry and the use of a segmented HPGe detector array impose special constraints on the design of the collimator for the system. Based on a signal detection model, an efficiency-resolution figure of merit (ERFM) as a function of the collimator geometric efficiency, system resolution width, and object diameter is defined. The ERFM is proportional to the square of the detection signal-to-noise ratio. The collimator design parameters can then be optimized by optimizing the ERFM for an anticipated object diameter. The collimator point-spread function, geometric efficiency, and resolution are calculated. The collimator optimized for the detection of a 1-cm object will have a single-slice point source efficiency of 1.2 X 10(-4), and a FWHM of 6.5 mm at the center of the reconstruction circle, at 12 cm from the collimator face. The minimum object contrast which will give a detection SNR of 5 is 74%, for a total accumulated count per slice of 2 X 10(6).

Systematic bias in basis material decomposition applied to quantitative dual-energy x-ray imaging.

Basis material decomposition represents dual-energy x-ray attenuation measurements in terms of the attenuation coefficients or thickness of two standard materials which, when combined, produce attenuation equivalent to the object being measured. In tomographic imaging, the reconstructed attenuation coefficient is calculated in terms of the attenuation coefficients of the basis materials, while in projection imaging, the thicknesses of two materials can be specified in terms of the basis materials. This analysis shows that basis material decomposition is exact in a dual-monoenergetic system, but for broad spectra, x-ray beam hardening introduces a bias into quantitative measurements. The error is small enough that it can be ignored when dual-energy imaging is used primarily to enhance the contrast of one material over another. The magnitude of the error in quantitative measurements depends on the details of the specific application including the energy of the x-ray beam, and the composition and thickness of the materials included in the object. The magnitude of the error for dual-energy bone densitometry has been analyzed using a first-order propagation of error analysis and the calculations verified by computer simulation. This analysis shows that the magnitude of the systematic error can be as high as 3% for 1 g/cm2 of bone mineral when aluminum and acrylic basis materials are used for the calibration. This systematic error is eliminated when the basis materials are the same as the materials that are being quantified (i.e., bone mineral and water).

Characterization and correction of pulse pile-up in simultaneous emission-transmission computed tomography.

We have developed an emission-transmission CT (ETCT) system capable of both single-photon emission computed tomography (SPECT) imaging and x-ray transmission CT imaging using a common photon counting detector. In principle, SPECT and x-ray CT projection data can be acquired simultaneously with the ETCT system; however, doing so results in contamination of the SPECT projection data due to pulse pile-up caused by the relatively high x-ray fluence rate. In this study, we characterize the effects of pulse pile-up for simultaneous ETCT imaging through computer simulation and experimental studies. We demonstrate that pulse pile-up in the SPECT energy window can be well approximated by a simple quadratic relationship between the pile-up rate and the x-ray fluence rate for sufficiently small x-ray fluence rates. Using this quadratic relationship, we developed a simple pile-up correction scheme that subtracts the pile-up counts from the emission data and also truncates the exterior regions of the emission projection data. Analysis of difference images and profiles indicate that this method permits us to reconstruct SPECT images with no apparent noise or resolution degradation in comparison to those obtained via sequential emission and transmission scans.

Geometrical properties of a digital beam attenuator system.

A digital beam attenuator system has been developed to automatically generate patient-specific compensating filters for chest radiography. An initial low-dose test image is used to generate the attenuator, which is fabricated by overprinting multiple layers of a heavy-metal material onto a nonattenuating substrate. The attenuator is subsequently inserted into the x-ray beam for a final compensated radiograph. The effects of focal spot blurring and limited attenuator resolution result in the final compensated image containing only high-spatial frequency information. The frequency response of the process is not strictly describable by a modulation transfer function, but an approximation of the frequencies remaining in the compensated image is obtained for low-contrast conditions. It is found that a 4 X 4 blurring function on the original 64 X 64 test image is required for the attenuator to give appropriate compensated image appearance. A proposed attenuator printing scheme prints the attenuator in a 16 X 16 matrix, staggering successively printed layers to achieve the required 64 X 64 sampling with appropriate blurring. The resulting compensated image has good anatomical definition and contains a frequency response similar to that obtained by compensation techniques being investigated by Plewes and Sorenson.

A bound on the energy resolution required for quantitative SPECT.

Scattered radiation is one of several physical perturbations that limit the accuracy of quantitative measurements in single-photon emission computed tomography (SPECT). Improvement in detector energy resolution leads to a reduction of scatter counts and a corresponding improvement in the quantitative accuracy of the SPECT measurement. In this study, simulated SPECT projections of a simple myocardial perfusion phantom were used to investigate the effect of detector energy resolution on the data. The phantom consists of a spherical shell of radionuclide within a 15 cm radius water-filled cylinder. Each projection contains on the order of 3 x 10(5) counts. The results demonstrate that a full-width, half-maximum energy resolution of 3-4 keV is sufficient to render the error due to scatter insignificant compared to the uncertainty due to photon statistics in this case. Further simulations verify that because smaller objects produce less scatter, they can be imaged accurately with degraded energy resolution. These results are useful when designing prototype systems that utilize solid-state detectors and low-noise electronics to achieve improved energy resolution.

A prototype high-purity germanium detector system with fast photon-counting circuitry for medical imaging.

A data-acquisition system designed for x-ray medical imaging utilizes a segmented high-purity germanium (HPGe) detector array with 2-mm wide and 6-mm thick elements. The detectors are contained within a liquid-nitrogen cryostat designed to minimize heat losses. The 50-ns pulse-shaping time of the preamplifier electronics is selected as the shortest time constant compatible with the 50-ns charge collection time of the detector. This provides the detection system with the fastest count-rate capabilities and immunity from microphonics, with moderate energy resolution performance. A theoretical analysis of the preamplifier electronics shows that its noise performance is limited primarily by its input capacitance, and is independent of detector leakage current up to approximately 100 nA. The system experimentally demonstrates count rates exceeding 1 million counts per second per element with an energy resolution of 7 keV for the 60-keV gamma ray photon from 241Am. The results demonstrate the performance of a data acquisition system utilizing HPGe detector systems which would be suitable for dual-energy imaging as well as systems offering simultaneous x-ray transmission and radionuclide emission imaging.

The propagation of stochastic pixel noise into magnitude and phase values in the Fourier analysis of digital images.

The use of Fourier analysis in nuclear medicine gated blood pool ventriculography provides a useful example of the application of Fourier methods to digital medical imaging. In particular, the nuclear medicine experience demonstrates that there is diagnostic significance not only in the pixel averages of temporal Fourier magnitude and phase computed in various image regions, but also in the distributions of the individual pixel values about those averages. However, a region containing pixels that are perfectly synchronous on average would still yield a finite distribution of calculated Fourier coefficients due to the propagation of stochastic pixel noise into the calculated values. We have studied this noise component of both the magnitude and phase distributions using phantom studies and computer simulation. In both approaches, several thousand one-pixel 'ventriculograms' were generated, all identical to each other except for stochastic noise. Fourier magnitudes and phases at several frequencies were calculated and histograms generated. A theoretical prediction of the distributions was developed and shown to fit the experimental results well. Our formalism can be used to estimate study count requirements or, for fixed study counts, to assess the stochastic noise contribution in the interpretation of measured phase and magnitude distributions.

Iterative concurrent reconstruction algorithms for emission computed tomography.

Direct reconstruction techniques, such as those based on filtered backprojection, are typically used for emission computed tomography (ECT), even though it has been argued that iterative reconstruction methods may produce better clinical images. The major disadvantage of iterative reconstruction algorithms, and a significant reason for their lack of clinical acceptance, is their computational burden. We outline a new class of 'concurrent' iterative reconstruction techniques for ECT in which the reconstruction process is reorganized such that a significant fraction of the computational processing occurs concurrently with the acquisition of ECT projection data. These new algorithms use the 10-30 min required for acquisition of a typical SPECT scan to iteratively process the available projection data, significantly reducing the requirements for post-acquisition processing. These algorithms are tested on SPECT projection data from a Hoffman brain phantom acquired with a 2 x 10(5) counts in 64 views each having 64 projections. The SPECT images are reconstructed as 64 x 64 tomograms, starting with six angular views. Other angular views are added to the reconstruction process sequentially, in a manner that reflects their availability for a typical acquisition protocol. The results suggest that if T s of concurrent processing are used, the reconstruction processing time required after completion of the data acquisition can be reduced by at least 1/3T s.

Noise propagation in SPECT images reconstructed using an iterative maximum-likelihood algorithm.

The effects of photon noise in the emission projection data and uncertainty in the attenuation map on the image noise in attenuation-corrected SPECT images reconstructed using a maximum-likelihood expectation-maximization algorithm were investigated. Emission projection data of a physical Hoffman brain phantom and a thorax-like phantom were acquired from a prototype emission-transmission computed tomography (ETCT) scanner being developed at UCSF. Computer-simulated emission projection data from a head-like phantom and a thorax-like phantom were also obtained using a fan-beam geometry consistent with the ETCT system. The simulation assumed a 99Tcm source, included collimator blurring but ignored photon scatter. For each phantom, a region of interest (ROI) at the centre of the reconstructed image was chosen for the purpose of noise analysis. In all cases, the mean value (m) in the ROI approached a constant value after approximately 20 iterations. The standard deviation (sigma) generally increased with the number of iterations. The ratio (sigma/m) was found to be inversely proportional to the square root of the total detected counts and proportional to the relative uncertainty in the attenuation maps. These two noise components contributed independently towards the noise in the reconstructed image. In the ETCT system employing an x-ray tube for attenuation map acquisition, the uncertainty in the reconstructed radionuclide distribution is limited mainly by photon noise in the emission projection data. Our results are expected to be generally applicable to other emission-transmission systems, including those using external radionuclide sources for the acquisition of attenuation maps.

Effect of bone distribution on vertebral strength: assessment with patient-specific nonlinear finite element analysis.

Three-dimensional quantitative computed tomographic (QCT) studies of the lumbar spine were extended with finite element analysis (FEA) to include bone distribution in assessment of vertebral body strength. Fifty-nine FEA models were created from data from 43 patients, 28 with no evidence of osteoporosis and 15 with previous vertebral fractures. Simulated loads were applied to the vertebral models to estimate vertebral strength. Yield strength in the models from patients with osteoporosis was 0.22-1.05 MPa (average, 0.57 MPa +/- 0.26 [mean +/- standard deviation]), compared with 0.80-2.79 MPa (1.46 +/- 0.52, P less than .001) in patients with normal bone. Yield strength of vertebrae in patients with osteoporosis uniformly fell below approximately 1.0 MPa, with minimal overlap between patients with osteoporosis and those with normal bone compared with the overlap in bone mineral content and trabecular mineral density. Reproducibility of the FEA technique was 12.1% in a subgroup of patients with normal bone. A constant relationship between cortical and trabecular contributions was observed in patients with osteoporosis but not in control patients.

Problems with contrast-detail curves for CT performance evaluation.

Contrast-detail curves have been used frequently to describe the low contrast performance characteristics of computed tomography (CT) scanners. However, such curves can produce misleading conclusions if the effects of all variables influencing CT images are not considered. As shown in this experimental study, improperly designed contrast-detail curves disguise differences in CT performance when the same object is imaged with different x-ray spectra. These problems arise because contrast is defined as the difference in system-dependent CT numbers rather than the actual difference in the object. An alternate approach to CT performance evaluation using "difference-detail" curves is offered.