Lingxiong Shao

Cross-plane scattering correction-point source deconvolution in PET

Most previous scattering correction techniques for PET are based on assumptions made for a single transaxial plane and are independent of axial variations. These techniques will incorrectly estimate the scattering fraction for volumetric PET imaging systems since they do not take the cross-plane scattering into account. We propose a new point source scattering deconvolution method (2-D). The cross-plane scattering is incorporated into the algorithm by modelling a scattering point source function. In the model, the scattering dependence both on axial and transaxial directions is reflected in the exponential fitting pa- rameters and these parameters are directly estimated from a limited number of measured point response functions. Our re- sults comparing the standard in-plane line source deconvolu- tion to our cross-plane point source deconvolution show that for a small source, the former technique overestimates the scat- ter fraction in the plane of the source and underestimates the scatter fraction in adjacent planes. In addition, we also propose a simple approximation technique for deconvolution.

Triple energy window scatter correction technique in PET

A practical triple energy window technique (TEW) is proposed, which is based on using the information in two lower energy windows and one single calibration, to estimate the scatter within the photopeak window. The technique is basically a conventional dual-window technique plus a modification factor, which can partially compensate object-distribution dependent scatters. The modification factor is a function of two lower scatter windows of both the calibration phantom and the actual object. In order to evaluate the technique, a Monte Carlo simulation program, which simulates the PENN-PET scanner geometry, was used. Different phantom activity distributions and phantom sizes were tested to simulate brain studies, including uniform and nonuniform distributions. The results indicate that the TEW technique works well for a wide range of activity distributions and object sizes. The comparisons between the TEW and dual window techniques show better quantitative accuracy for the TEW, especially for different phantom sizes. The technique is also applied to experimental data from a PENN-PET scanner to test its practicality.

Evaluation of volume imaging with the HEAD PENN-PET scanner

A volume-imaging PET scanner for brain, animal, and pediatric imaging has been designed and built to achieve high performance, specifically in sensitivity and spatial resolution. The scanner is unique in its use of a single annular crystal of NaI(Tl), which allows a 30 cm diameter patient port and an axial field-of-view of 25.6 cm with a maximum axial acceptance angle a of /spl plusmn/26/spl deg/. By using a narrow photopeak energy window, this leads to high sensitivity with relatively low scatter and randoms. Image resolution is 3.5 mm in both the transverse and axial directions. Images are reconstructed with 128 slices, and are presented for phantom and patient studies (FDG) with both the multi-slice rebinning and the 3D-reprojection algorithm.<<ETX>>

Composite Dual Window Scattering Correction Technique In PET

Most current energy spectrum based scattering correction techniques in PET suffer from source distribution dependencies. As an alternative method to compensate for these problems, we propose a composite dual energy window (CDW) scattering correction technique, which is based on using the scatter kernels from both the scatter and the photopeak windows. The CDW technique can be broken down to two steps: convolution and energy window correction. In the convolution part, we can obtain the first order scatter projections for both the photopeak and scatter windows by using the calibrated scatter kernels. Then we find the object dependent scatter ratio by dividing the convolution-estimated scatter projections. By this simple division, the overestimation of scatters during convolution is mostly canceled out. In the second step, we simply use the measured scatter profiles in the scatter window and the estimated scatter ratio to obtain the scatter projection in the photopeak window. The CDW technique is evaluated with a Monte Carlo simulation program which simulates the UGM PENN-PET scanner. The results indicate that this technique is better than both the deconvolution-subtraction with one iteration and dual window scattering correction techniques.

Modified convolution-subtraction scattering correction technique for 3D PET

A modified convolution-subtraction (CS) scatter correction technique in 3D PET imaging has been proposed and evaluated. It can compensate for the shortcomings with the conventional CS techniques, which are commonly used in 2D PET data. The shortcomings include requiring the position-dependent scatter-kernel calibration, the non-standard convolution and long processing time. The modified CS technique assumes that the position-dependent scatter response function (SRF) can be modeled by a product of an average SRF, an adjustable parameter and a relative scatter fraction function. The relative scatter fraction function is a function of source position. The technique was applied to both simulated and measured data. The preliminary results indicate that the modified CS scattering correction technique is practical and robust to the sizes of objects. It provides more accurate scatter estimates than the conventional CS techniques, especially for highly nonuniform distributed sources.

Cross Plane Scattering Correction - Point Source Deconvolution In PET

The authors propose a new 2-D point source scattering deconvolution method. The cross-plane scattering is incorporated into the algorithm by modeling a scattering point source function. In the model, the scattering dependence on axial and transaxial directions is reflected in the exponential fitting parameters, and these parameters are directly estimated from a limited number of measured point response functions. The results comparing the standard in-plane line source deconvolution to the cross-plane point source deconvolution show that for a small source the former technique overestimates the scatter fraction in the plane of the source and underestimates the scatter fraction in adjacent planes. In addition, the authors also propose a simple approximation technique for deconvolution.

Practical approaches for oxygen flow studies in PET

Two simple regional cerebral blood flow (rCBF) models have been proposed and investigated with the ramped infusion injection method: step and scaling. The step model is similar to the Alpert et al.s (1984) model except that the authors used step-function weights with which they only need to acquire two time frames of images compared to the 18 time frames of images that the authors currently use in their dynamic fitting model. The scaling model is based on the fact that in most patient studies, the rCBF varies around the whole brain CBF. Hence, within a small range of the whole brain CBF, the rCBF can be obtained by linearly scaling the density image corresponding to the whole brain CBF. Both models were evaluated and compared to other conventional models with simulated and patient data. In this study, 10 patients were analyzed. The average percentage errors on rCBF over a large group of regions for the step and the scaling models with respect to the dynamic model are 4.0% and 4.8%, respectively.<<ETX>>

Information gain from count corrections in SPECT image reconstruction and classification

The authors present a method for quantifying the impact of count correction side information on the capability of a SPECT (single-photon-emission computed tomography) projective tomography system to perform image (emitter) reconstruction and feature classification. The method involves computing the image or feature related information gain which results from the presence of count correction side information at the detector. For image reconstruction this information gain is computed using Shannons mutual information, while for feature classification the cutoff rate of the SPECT information channel is used. For reconstruction the gain is proportional to the information divergence between the spatially dependent probability of deletion of a gamma -ray originating at a particular emitter location and the spatially independent average deletion probability. For classification the gain is proportional to the difference between the arithmetic mean and the geometric mean of the average number of gamma -ray deletions for each of the image classes. Results of analysis and numerical study are presented which indicate that the information gain associated with using count correction data is much more significant for reconstruction of the emitters than for classification of the emitter density when the total detected fluence is low. >

Quantitative measurements of cerebral blood flow in volume imaging PET scanners

Quantitative measurements of cerebral blood flow (CBF) are performed in a volume imaging PET scanner by means of moderate activity infusions. In equilibrium infusions, activations are measured by scanning over 10 minutes with 16 minute activations. Typical measured whole brain CBF values are 37/spl plusmn/8 ml/min/dl, close to the value of 42 ml/min/dl reported by other groups using this method. For ramped infusions, scanning over 4 minutes with 5 minute activations results in whole brain CBFs of 49/spl plusmn/9 ml/min/dl, close to the Kety and Schmidt value of 50 ml/min/dl. Both equilibrium and ramped infusion methods have been used to study face and word memory in human subjects. Both methods were able to detect significant activations in regions implicated in human memory (amygdala and hippocampus). It is concluded that both accurate and precise quantitation of regional CBF (rCBF) is achieved using the methods described. In addition a simplified protocol for ramped infusion studies has been investigated. In this method the whole brain tissue time activity curve generated from dynamic scanning is replaced by an appropriately scaled camera coincidence countrate curve. The resulting whole brain CBF values are only 7% different from the dynamic scan results. Methods for obtaining the rCBF from the summed image and the resulting whole brain CBF are under investigation.<<ETX>>

Parameter estimation using the impulse response function for high affinity neuroreceptor ligands in PET

A tool for the complex task of multiparameter estimates from positron emission tomography (PET) and single photon emission computed tomography (SPECT) data, based on models of high affinity neuroreceptors ligands is presented. A simplified kinetic model of high affinity ligands and the mathematical background for the estimation of an impulse response function (IRF) from measured time activity curves are discussed. The advantages of this approach are that there is a straightforward relationship between the rate constants and the shape of the theoretical IRF, allowing a simplified parameter estimation process that in some cases will only yield one solution. Additionally, the estimated IRF allows a qualitative assessment of reliability of the estimates for each parameter and also an intuitive method of selectively weighting the data. Results based on simulated and measured N-methy-spiperone (NMSP) studies show that the IRF method is robust and also is a useful tool in conjunction with a standard three-parameter estimation algorithm.<<ETX>>