Motion correction of PET brain images through deconvolution: II. Practical implementation and algorithm optimization.

Image quality is significantly degraded even by small amounts of patient motion in very high-resolution PET scanners. When patient motion is known, deconvolution methods can be used to correct the reconstructed image and reduce motion blur. This paper describes the implementation and optimization of an iterative deconvolution method that uses an ordered subset approach to make it practical and clinically viable. We performed ten separate FDG PET scans using the Hoffman brain phantom and simultaneously measured its motion using the Polaris Vicra tracking system (Northern Digital Inc., Ontario, Canada). The feasibility and effectiveness of the technique was studied by performing scans with different motion and deconvolution parameters. Deconvolution resulted in visually better images and significant improvement as quantified by the Universal Quality Index (UQI) and contrast measures. Finally, the technique was applied to human studies to demonstrate marked improvement. Thus, the deconvolution technique presented here appears promising as a valid alternative to existing motion correction methods for PET. It has the potential for deblurring an image from any modality if the causative motion is known and its effect can be represented in a system matrix.

Motion correction of PET brain images through deconvolution: I. Theoretical development and analysis in software simulations.

Image quality is significantly degraded even by small amounts of patient motion in very high-resolution PET scanners. Existing correction methods that use known patient motion obtained from tracking devices either require multi-frame acquisitions, detailed knowledge of the scanner, or specialized reconstruction algorithms. A deconvolution algorithm has been developed that alleviates these drawbacks by using the reconstructed image to estimate the original non-blurred image using maximum likelihood estimation maximization (MLEM) techniques. A high-resolution digital phantom was created by shape-based interpolation of the digital Hoffman brain phantom. Three different sets of 20 movements were applied to the phantom. For each frame of the motion, sinograms with attenuation and three levels of noise were simulated and then reconstructed using filtered backprojection. The average of the 20 frames was considered the motion blurred image, which was restored with the deconvolution algorithm. After correction, contrast increased from a mean of 2.0, 1.8 and 1.4 in the motion blurred images, for the three increasing amounts of movement, to a mean of 2.5, 2.4 and 2.2. Mean error was reduced by an average of 55% with motion correction. In conclusion, deconvolution can be used for correction of motion blur when subject motion is known.

Gated stress-only 99mTc myocardial perfusion SPECT imaging accurately assesses coronary artery disease.

In today's cost containment environment it is important to consider changes to standard protocols which would reduce cost, particularly if there is no significant loss of diagnostic accuracy. The aim of the present study was to assess the usefulness of a gated stress-only Tc sestamibi protocol in comparison to conventional gated dual isotope rest-stress myocardial perfusion single photon emission computed tomography (SPECT) in the detection and localization of coronary artery disease (CAD). Sixty-five consecutive patients (65+/-10 years, 22 women) who had undergone conventional gated perfusion SPECT were chosen retrospectively. Fifty-three of these 65 patients had previous coronary arteriography, 45 with at least one stenosis, eight without stenosis, and 12 of these 65 patients had <5% likelihood of CAD. Three readers interpreted the gated stress-only and dual isotope studies in separate sessions blinded to (1). their previous readings, (2). the interpretation by others, and (3). the angiographic results. Readers used a five-point scale to score their visual and quantitative assessment of perfusion, function and compromised vascular territory. Their average score was used for determination of the accuracy by using receiver operating characteristic (ROC) analysis of the techniques. The areas under ROC curves were determined for the detection of CAD and localization of vascular territories. Fifty-four of these patients had 97 significant stress induced perfusion defects as determined by the CEqual quantitative program. The agreement between protocols for the assessment of reversibility in these 97 defects was analysed. There were no statistically significant differences between dual isotope rest/stress and gated stress-only studies for the detection and localization of CAD. The area under the dual isotope rest/stress ROC curve was 0.78+/-0.07 compared to the area under the gated stress-only ROC curve of 0.80+/-0.06, resulting in P=0.30. For the combined vessels comparison of the area under the dual isotope rest/stress ROC curve was 0.73+/-0.04 versus the area under the gated stress-only ROC curve of 0.74+/-0.04, resulting in P=0.27. Similar non-significant differences were obtained when comparing the area under the dual isotope versus gated stress-only ROC curves for the left anterior descending vascular territory (LAD, 0.61+/-0.08 vs 0.660.08, P=0.14), the left circumflex vascular territory (LCX, 0.82+/-0.07 vs 0.81+/-0.06, P =0.47) or the right coronary vascular territory (RCA, 0.80+/-0.06 vs 0.78+/-0.06, P=0.28). The analysis of the reversibility of stress induced perfusion defects yielded a global agreement between protocols of 93% (kappa=0.42). The differences were due to the expert readers, using the gated stress-only protocol, misinterpreting some patients with attenuation artefacts, subendocardial infarction and functional stunning. These results show that the lower cost gated stress-only myocardial Tc myocardial perfusion SPECT studies are comparable to the conventional dual isotope studies when the clinical question is the detection and localization of coronary artery disease. Nevertheless, we also showed that this approach is limited when attempting to interpret the reversibility of stress induced perfusion defects in patients who exhibit attenuation artefacts, subendocardial infarction and functional stunning.

Evaluation of left ventricular endocardial volumes and ejection fractions computed from gated perfusion SPECT with magnetic resonance imaging: comparison of two methods.

BACKGROUND: Two methods of computing left ventricular volumes and ejection fraction (EF) from 8-frame gated perfusion single photon emission computed tomography (SPECT) were compared with each other and with magnetic resonance (MR) imaging. METHODS AND RESULTS: Thirty-five subjects underwent 8-frame gated dual-isotope SPECT imaging and 12- to 16-frame gated MR imaging. Endocardial boundaries on short-axis MR images were hand traced by experts blinded to any SPECT results. Volumes and EF were computed with the use of Simpson's rule. SPECT images were analyzed for the same functional variables with the use of 2 automatic programs, Quantitative Gated SPECT (QGS) and the Emory Cardiac Toolbox (ECTb). The mean difference between MR and SPECT EF was 0.008 for ECTb and 0.08 for QGS. QGS showed a slight trend toward higher correlation for EF (r = 0.72, SE of the estimate = 0.08) than ECTb (r = 0.70, SE of the estimate = 0.09). For both SPECT methods, left ventricular volumes were similarly correlated with MR, although SPECT volumes were higher than MR values by approximately 30%. CONCLUSIONS: QGS and ECTb values of cardiac function computed from 8-frame gated perfusion SPECT correlate very well with each other and correlate well with MR. Averaged over all subjects, ECTb measurements of EF are not significantly different from MR values but QGS significantly underestimates the MR values.

Neural activity related to drug craving in cocaine addiction.

BACKGROUND: Crack cocaine dependence and addiction is typically associated with frequent and intense drug wanting or craving triggered by internal or environmental cues associated with past drug use. METHODS: Water O 15 positron emission tomography (PET) studies were used to localize alterations in synaptic activity related to cue-induced drug craving in 8 crack cocaine-dependent African American men. In a novel approach, script-guided imagery of autobiographical memories were used as individualized cues to internally generate a cocaine craving state and 2 control (ie, anger and neutral episodic memory recall) states during PET image acquisition. RESULTS: The mental imagery of personalized drug use and anger-related scripts was associated with self-ratings of robust drug craving or anger, and comparable alterations in heart rate. Compared with the neutral imagery control condition, imagery-induced drug craving was associated with bilateral (right hemisphere amygdala activation greater than left) activation of the amygdala, the left insula and anterior cingulate gyrus, and the right subcallosal gyrus and nucleus accumbens area. Compared with the anger control condition, internally generated drug craving was associated with bilateral activation of the insula and subcallosal cortex, left hippocampus, and anterior cingulate cortex and brainstem. A brain-wide pixel-by-pixel search indicated significant positive and negative correlations between imagery-induced cocaine craving and regional cerebral blood flow (rCBF) in distributed sites. CONCLUSIONS: The collected findings suggest the craving-related activation of a network of limbic, paralimbic, and striatal brain regions, including structures involved in stimulus-reward association (amygdala), incentive motivation (subcallosal gyrus/nucleus accumbens), and anticipation (anterior cingulate cortex).

Linking sight and sound: fMRI evidence of primary auditory cortex activation during visual word recognition.

We describe two studies that used repetition priming paradigms to investigate brain activity during the reading of single words. Functional magnetic resonance images were collected during a visual lexical decision task in which nonword stimuli were manipulated with regard to phonological properties and compared to genuine English words. We observed a region in left-hemisphere primary auditory cortex linked to a repetition priming effect. The priming effect activity was observed only for stimuli that sound like known words; moreover, this region was sensitive to strategic task differences. Thus, a brain region involved in the most basic aspects of auditory processing appears to be engaged in reading even when there is no environmental oral or auditory component.

Clinical impact of arrhythmias on gated SPECT cardiac myocardial perfusion and function assessment.

BACKGROUND: We reported previously that mean quantified cardiac functional parameters computed by one gated single photon emission computed tomography (SPECT) technique were not significantly altered by common gating errors. However, it is not known to what extent other gated SPECT approaches that are based on different ventricular modeling assumptions are influenced by arrhythmias, nor are the effects of gating errors on visual analyses and their subsequent clinical implications known. METHODS: Projection data for 50 patients (aged 64 +/- 12 years; 68% men; 76% with myocardial perfusion defects) undergoing technetium-99m sestamibi gated SPECT who were in sinus rhythm during data acquisition were altered to simulate common arrhythmias. To determine quantitative effects, we performed calculations for original control and altered images by Gaussian myocardial detection (Quantitative Gated SPECT [QGS] program) and by wall thickening derived from gated perfusion polar maps (Emory Cardiac Toolbox program). To evaluate visual assessment in control and simulated-arrhythmia tomograms, 2 experienced blinded observers independently interpreted perfusion from polar maps and wall motion and thickening from tomographic cines, using a 4-point scale. RESULTS: Although mean functional parameters were scarcely altered, paired t tests showed ejection fraction fluctuations to be significantly different from control values, causing patients to change between abnormal and normal ejection fraction categories (2% of patients by QGS and 14% by Emory Cardiac Toolbox). Visual examination of QGS polar perfusion and function maps showed changes for 72% of cases, although in only 4% were these considered to have potential clinical consequences. The kappa statistic for visual analysis of concordance between control and arrhythmia readings showed that agreement was "excellent" for perfusion, "good" for motion, and "marginal" for thickening. CONCLUSIONS: As with quantitative measurements, thickening is the parameter most prone to error in the presence of arrhythmias. It is important to test data for gating errors to avoid potentially erroneous measurements and visual readings.

Alterations in the functional anatomy of working memory in adult attention deficit hyperactivity disorder.

OBJECTIVE: The authors used a functional neuroimaging study with a working memory probe to investigate the pathophysiology of attention deficit hyperactivity disorder (ADHD). Their goal was to compare regional cerebral blood flow (rCBF) changes related to working memory in adults with and without ADHD. METHOD: Using [(15)O]H(2)O positron emission tomography (PET) studies, the authors compared the sites of neural activation related to working memory in six adult men diagnosed with ADHD and six healthy men without ADHD who were matched in age and general intelligence. RESULTS: Task-related changes in rCBF in the men without ADHD were more prominent in the frontal and temporal regions, but rCBF changes in men with ADHD were more widespread and primarily located in the occipital regions. CONCLUSIONS: These data suggest the use of compensatory mental and neural strategies by subjects with ADHD in response to a disrupted ability to inhibit attention to nonrelevant stimuli and the use of internalized speech to guide behavior.

Three-dimensional color-modulated display of myocardial SPECT perfusion distributions accurately assesses coronary artery disease.

UNLABELLED: The objective of this study was to compare visual assessment of 3-dimensional color-modulated (to counts) surface displays with visual assessment of oblique tomographic slices of myocardial SPECT perfusion distributions in the detection and localization of coronary artery disease (CAD). METHODS: Sixty-two consecutive patients (41 men, 21 women; mean age, 61 +/- 11 y) who had undergone conventional dual-isotope perfusion SPECT were retrospectively chosen; 50 had undergone coronary arteriography previously, and 12 had less than a 5% likelihood of CAD. Four readers interpreted the 3-dimensional displays and slices in separate sessions while unaware of the findings of previous readings, the interpretations of others, and the angiographic results. The readers used a 5-point scoring system. Their average score was used for receiver operating characteristic (ROC) analysis. The area under the ROC curves was determined so that the ability of the 2 methods to detect and localize CAD could be compared. RESULTS: No significant differences were found between visual interpretation of 3-dimensional displays and visual interpretation of slices. CONCLUSION: These preliminary results indicate that visual interpretation of 3-dimensional displays of myocardial perfusion SPECT distributions may someday replace visual assessment of conventional slices without loss of diagnostic accuracy.

A confrontational naming task produces congruent increases and decreases in PET and fMRI.

This work uses the well-established (by PET) confrontation naming task to compare PET and fMRI in a cognitive activation experiment. The signal changes from this task are much less than the changes caused by visual or motor activation tasks used in previous comparisons. ANOVA methods adjusted for multiple comparisons were used to determine significant changes in signal between confrontation naming and figure size discrimination tasks. All 17 significantly increased regions (confrontation naming signal greater) seen on one modality were increased on both modalities. Ten of 13 regions that were significantly decreased on one modality were decreased on the other. Three mismatched regions showed a significant decrease on one modality and a nonsignificant increase on the other. This study could not detect a consistent difference in activation site location between PET and fMRI.

Left ventricular function and perfusion from gated SPECT perfusion images: an integrated method.

UNLABELLED: A new technique for computing left ventricular function, including left ventricular volumes, mass and ejection fraction, has been developed. This method is a logical extension of the results of a standard perfusion quantification technique; thus, it allows integration of perfusion and functional information. METHODS: Anatomically based models of the endocardial and epicardial surfaces are generated using the myocardial samples for which perfusion values are quantified, for all frames in the cardiac cycle. With these surface points, left ventricular chamber volume and myocardial volume can be computed. A computer simulation was used to determine the sensitivity of the approach to the assumptions of the model. Validation of volume, mass and ejection fraction was performed with correlative MR studies, and ejection fraction and left ventricular volumes were further investigated using correlative first-pass studies. RESULTS: Automated processing was successful in 96% of the cases analyzed. End diastolic volume, end systolic volume, left ventricular mass and left ventricular ejection fraction correlated with MRI with r = 0.97, 0.99, 0.87, and 0.85, respectively. Ejection fraction from tomography correlated with first-pass values with r = 0.82, and end diastolic and end systolic volumes from tomography correlated with first-pass values with r = 0.85 and r = 0.91, respectively. CONCLUSION: The new integrated approach is accurate and robust for computing both perfusion and function from perfusion tomograms.

Acute blood flow changes and efficacy of vagus nerve stimulation in partial epilepsy.

OBJECTIVE: To determine possible sites of therapeutic action of vagus nerve stimulation (VNS), by correlating acute VNS-induced regional cerebral blood flow (rCBF) alterations and chronic therapeutic responses. BACKGROUND: We previously found that VNS acutely induces rCBF alterations at sites that receive vagal afferents and higher-order projections, including dorsal medulla, somatosensory cortex (contralateral to stimulation), thalamus and cerebellum bilaterally, and several limbic structures (including hippocampus and amygdala bilaterally). METHODS: VNS-induced rCBF changes were measured by subtracting resting rCBF from rCBF during VNS, using [O-15]water and PET, immediately before ongoing VNS began, in 11 partial epilepsy patients. T-statistical mapping established relative rCBF increases and decreases for each patient. Percent changes in frequency of complex partial seizures (with or without secondary generalization) during three months of VNS compared with pre-VNS baseline, and T-thresholded rCBF changes (for each of the 25 regions of previously observed significant CBF change), were rank ordered across patients. Spearman rank correlation coefficients assessed associations of seizure-frequency change and t-thresholded rCBF change. RESULTS: Seizure-frequency changes ranged from 71% decrease to 12% increase during VNS. Only the right and left thalami showed significant associations of rCBF change with seizure-frequency change. Increased right and left thalamic CBF correlated with decreased seizures (p < 0.001). CONCLUSIONS: Increased thalamic synaptic activities probably mediate some antiseizure effects of VNS. Future studies should examine neurotransmitter-receptor alterations in reticular and specific thalamic nuclei during VNS.

Brain blood flow alterations induced by therapeutic vagus nerve stimulation in partial epilepsy: I. Acute effects at high and low levels of stimulation.

PURPOSE: Left cervical vagus nerve stimulation (VNS) decreases complex partial seizures (CPS) by unknown mechanisms of action. We hypothesized that therapeutic VNS alters synaptic activities at vagal afferent terminations and in sites that receive polysynaptic projections from these medullary nuclei. METHODS: Ten patients with partial epilepsy underwent positron emission tomographic (PET) measurements of cerebral blood flow (BF) three times before and three times during VNS. Parameters for VNS were at high levels for 5 patients and at low levels for 5. Resting BF measurements were subtracted from measurements during VNS in each subject. Subtraction data were averaged in each of 2 groups of 5 patients. t Tests were applied to BF changes in brain regions that receive vagal afferents and projections (significant at p < 0.05, corrected for repeated measures). RESULTS: In both the low- and high-stimulation groups during VNS, brain BF was (a) increased in the rostral, dorsal-central medulla; (b) increased in the right postcentral gyrus, (c) increased bilaterally in the hypothalami, thalami, and insular cortices, and in cerebellar hemispheres inferiorly; and (d) decreased bilaterally in hippocampus, amygdala, and posterior cingulate gyri. The high-stimulation group had greater volumes of activation and deactivation sites. CONCLUSIONS: Our findings suggest that left cervical VNS acutely increases synaptic activity in structures directly innervated by central vagal structures and areas that process left-sided somatosensory information, but VNS also acutely alters synaptic activity in multiple limbic system structures bilaterally. These findings may reflect sites of therapeutic actions of VNS.

Clinical validation of three-dimensional color-modulated displays of myocardial perfusion.

BACKGROUND: Two-dimensional polar maps have been validated previously with coronary arteriography for determining vascular involvement of defects from a patient's myocardial perfusion distributions with and without quantification. The purpose of this study was to validate previously developed three-dimensional color-modulated surface displays representing myocardial perfusion. METHODS AND RESULTS: The validation consisted of comparing the agreement between the three-dimensional displays and two-dimensional polar maps in localizing perfusion defects to vascular territories in 30 patients (16 men/14 women) who underwent both a 1-day rest/stress exercise 99mTc-labeled sestamibi study and coronary arteriography. Reading by two experts was used to identify the size and location of quantified defects and corresponding areas of reversibility seen in the polar maps and, on a separate day, in the three-dimensional displays. Agreement between the two-dimensional polar maps and the three-dimensional displays resulted in identical percentages for the localization of both defects and reversibilities: left anterior descending coronary artery, 87% (26/30); left circumflex coronary artery, 97% (29/30); right coronary artery, 97% (29/30); and coronary artery disease, 97% (29/30). CONCLUSIONS: These results show that the color-modulated three-dimensional displays are at least as good as the CEqual polar maps in localizing a perfusion defect and its reversibility to angiographically defined vascular territories and thus could be used in the routine clinical evaluation of myocardial perfusion.

Role of posterior parietal cortex in the recalibration of visually guided reaching.

Visually guided reaching requires complex neural transformations to link visual and proprioceptive inputs with appropriate motor outputs. Despite the complexity of these transformations, hand-eye coordination in humans is remarkably flexible, as demonstrated by the ease with which reaching can be adapted to distortions in visual feedback. If subjects attempt to reach to visual targets while wearing displacing prisms, they initially misreach in the direction of visual displacement. Given feedback about their reaching errors, however, they quickly adapt to the visual distortion. This is shown by the gradual resumption of accurate reaching while the prisms remain in place, and by the immediate onset of reaching errors in the opposite direction after the prisms have been removed. Despite an abundance of psychophysical data on adaptation to prisms, the functional localization of this form of sensorimotor adaptation is uncertain. Here we use positron emission tomography (PET) to localize changes in regional cerebral blood flow (rCBF) in subjects who performed a prism-adaptation task as well as a task that controlled for the sensory, motor and cognitive conditions of the adaptation experiment. Difference images that reflected the net effects of the adaptation process showed selective activation of posterior parietal cortex contralateral to the reaching limb.

Three-dimensional displays of left ventricular epicardial surface from standard cardiac SPECT perfusion quantification techniques.

UNLABELLED: Two methods for generating left ventricular epicardial surface from SPECT perfusion tomograms are described and validated. Both methods use the locations of the maximal reconstructed count values determined from a perfusion quantification procedure as a basis for generating surfaces. METHODS: The first method fits circular contours, which are perpendicular to the long-axis, to the points obtained from perfusion quantification. The second method applies median and linear filters to the points to remove noise but maintain the basic shape of the surface. Both models are validated against an automatic technique and against the user-traced surfaces of both the perfusion image and an MR image of the same patient. RESULTS: The median-filtered model was found to be closer to the standard surfaces than the circular model in all cases, and 85% of the points on the median-filtered surfaces were within one SPECT pixel length of the hand-traced MR surfaces. CONCLUSION: Accurate, three-dimensional left ventricular epicardial surfaces can be generated quickly and easily from already existing perfusion quantification software. The resulting images may be useful for realistic displays of ventricular size, shape and the three-dimensional distribution of perfusion.

Determining the accuracy of calculating systolic wall thickening using a fast Fourier transform approximation: a simulation study based on canine and patient data.

UNLABELLED: The high court yields of 99mTc-sestamibi make possible the acquisition of multiple gated SPECT studies with relatively high count densities. By reorienting these studies into gated short-axis slices, and extracting the three-dimensional myocardial perfusion distribution, we can study wall thickening using an amplitude and phase analysis methodology that examines the change in counts throughout the cardiac cycle. There have been two main concerns raised about this count-based technique: (1) What effect does the sampling rate have on the calculation of systolic wall thickening? and (2) What effect does count density have on the calculation of systolic wall thickening? METHODS: We designed a simulation study using myocardial wall thickening data obtained from ultrasonic crystals implanted in the myocardium of a normal canine. This data was modified to produce wall thickening curves with various percent systolic wall thickening measurements, sampling rates and count densities. RESULTS: The results show that using at least eight frames per cardiac cycle, systolic wall thickening can be calculated with enough accuracy to separate normal patients from those with cardiac dysfunction, even in areas of hypoperfused myocardium. Also, the results show the importance of calculating and interpreting phase (onset of contraction) information. CONCLUSIONS: This count-based technique continues to show promise as a tool for calculating systolic wall-thickening from multiple gated myocardial perfusion SPECT studies, but needs to be validated in a prospective multi-center trial before being applied in a clinical setting.

Multiheaded rotating gamma cameras in cardiac single-photon emission computed tomographic imaging.

Multiheaded rotating gamma cameras can do more than simply decrease the time required for cardiac single-photon emission computed tomographic (SPECT) acquisitions. They give their users a flexibility to improve image quality that cannot be achieved so easily with single-headed systems. Multiheaded cameras can be used to acquire quickly those radiopharmaceuticals whose distributions washout very rapidly, increase count levels in noisy images without lengthening imaging time, permit high-resolution collimation or electrocardiographic gating with little or no decrease in counts, or acquire transmission images for attenuation correction concurrently with an emission study. This new generation of SPECT scanners gives the nuclear cardiology community a unique opportunity to create a new generation of cardiac SPECT images.