Time-of-flight (TOF) positron emission tomography (PET) has gained remarkable development recently due to the advances in scintillator, silicon photomultipliers (SiPM), and fast electronics. However, current clinical reconstruction algorithms in TOF-PET are still based on ordered-subset-expectation-maximization (OSEM) and its variants, which may face challenges in non-conventional imaging applications, such as fast imaging within short scan time. In this work, we propose an image-TV constrained optimization problem, and tailor a primal- dual algorithm for solving the problem and reconstructing images. We collect list-mode data of a Jaszczak phantom with a prototype digital TOF-PET scanner. We focus on investigating image reconstruction from data collected within reduced scan time, and thus of lower count levels. Results of the study indicate that our proposed algorithm can 1) yield image reconstruction with suppressed noise, extended axial volume coverage, and improved spatial resolution over that obtained in conventional reconstructions, and 2) yield reconstructions with potential clinical utility from data collected within shorter scan time.
We report the development of a modularized compact positron emission tomography (PET) detector that outputs serial streams of digital samples of PET event pulses via an Ethernet interface using the UDP/IP protocol to enable rapid configuration of a PET system by connecting multiple such detectors via a network switch to a computer. Presently, the detector is 76 mm×50 mm×55 mm in extent (excluding I/O connectors) and contains an 18×12 array of 4.2×4.2×20 mm3 one-to-one coupled lutetium-yttrium oxyorthosilicate/silicon photomultiplier pixels. It employs cross-wire and stripline readouts to merge the outputs of the 216 detector pixels to 24 channels. Signals at these channels are sampled using a built-in 24-ch, 4-level field programmable gate arrays-only multivoltage threshold digitizer. In the computer, software programs are implemented to analyze the digital samples to extract event information and to perform energy qualification and coincidence filtering. We have developed two such detectors. We show that all their pixels can be accurately discriminated and measure a crystal-level energy resolution of 14.4% to 19.4% and a detector-level coincidence time resolution of 1.67 ns FWHM. Preliminary imaging results suggests that a PET system based on the detectors can achieve an image resolution of ∼1.6 mm.
There exists interest in designing a PET system with reduced detectors due to cost concerns, while not significantly compromising the PET utility. Recently developed optimization-based algorithms, which have demonstrated the potential clinical utility in image reconstruction from sparse CT data, may be used for enabling such design of innovative PET systems. In this work, we investigate a PET configuration with reduced number of detectors, and carry out preliminary studies from patient data collected by use of such sparse-PET configuration. We consider an optimization problem combining Kullback-Leibler (KL) data fidelity with an image TV constraint, and solve it by using a primal-dual optimization algorithm developed by Chambolle and Pock. Results show that advanced algorithms may enable the design of innovative PET configurations with reduced number of detectors, while yielding potential practical PET utilities.
Positron emission tomography (PET) is an important imaging modality in both clinical usage and research
studies. For small-animal PET imaging, it is of major interest to improve the sensitivity and resolution. We
have developed a compact high-sensitivity PET system that consisted of two large-area panel PET detector
heads. The highly accurate system response matrix can be computed by use of Monte Carlo simulations, and
stored for iterative reconstruction methods. The detector head employs 2.1x2.1x20 mm<sup>3</sup> LSO/LYSO crystals of
pitch size equal to 2.4 mm, and thus will produce more than 224 millions lines of response (LORs). By exploiting
the symmetry property in the dual-head system, the computational demands can be dramatically reduced.
Nevertheless, the tremendously large system size and repetitive reading of system response matrix from the hard
drive will result in extremely long reconstruction times. The implementation of an ordered subset expectation
maximization (OSEM) algorithm on a CPU system (four Athlon x64 2.0 GHz PCs) took about 2 days for 1
iteration. Consequently, it is imperative to significantly accelerate the reconstruction process to make it more
useful for practical applications. Specifically, the graphic processing unit (GPU), which possesses highly parallel
computational architecture of computing units can be exploited to achieve a substantial speedup. In this work, we
employed the state-of-art GPU, NVIDIA Tesla C2050 based on the Fermi-generation of the compute united device
architecture (CUDA) architecture, to yield a reconstruction process within a few minutes. We demonstrated
that reconstruction times can be drastically reduced by using the GPU. The OSEM reconstruction algorithms
were implemented employing both GPU-based and CPU-based codes, and their computational performance was
quantitatively analyzed and compared.
We present our work toward implementing all-digital signal processing for Positron Emission Tomography (PET)
event detection. In the conventional PET system, proper calibration and extending event processing are challenging
tasks due to the huge number of channels and multiplexing of input signals in the mixed-signal front-end.
To alleviate such limitations, we have proposed a simple all-digital PET system utilizing digital signal processing
(DSP) technologies for analyzing event pulses generated in PET. In this work, we implement a Gaussian shaper
circuit for scintillation pulses, which followed by a moderate sampling rate Analog-to-Digital Converter (ADC).
We also evaluate two DSP algorithms for extracting time information from the digitized pulse samples, and the
two algorithms examined could generate a coincidence timing resolution of ~ 2.4ns FWHM, by using a 125MSps
sampling rate ADC.
In recent years, the clinical status of positron emission tomography(PET)/computed tomography(CT) in achieving
more accurate staging of lung cancer has been established and the technology has been enthusiastically
accepted by the medical community. However, its capability in chest imaging is still limited by several physical
factors. As a result of typical PET/CT imaging protocol, respiration-averaged PET data and free of respiration-averaged
CT data are collected in a PET/CT scanning. In this work, we investigate the effects of respiration
motion. We employ mathematical and Monte-Carlo simulations for generating PET/CT data. We scale a
Zubal phantom to generate 30 phantoms having various sizes in order to represent different torso anatomic
states during respiration. Images reconstructed from selected scaling PET data using the respective scaling
PET attenuation maps serve as baseline results. PET/CT imaging protocol is simulated by reconstruction
from respiration-averaged PET data with the selected PET attenuation maps. We also reconstruct PET images
from respiratory-averaged PET data with respiration-averaged PET attenuation maps, which simulates conventional
PET imaging protocol. We will compare the resulting images reconstructed from the above-mentioned
approaches to evaluate the effects of respiration motion in PET/CT.
Positron emission tomography~(PET) systems employ mixed-signal front-end to carry out relatively simple, and <i>ad hoc</i>, processing of the charge pulses generated upon event detection. To obtain, and maintain over time, proper calibrations of the mixed-signal circuitry for generating accurate event information is a challenging task due to the simplicity of the event processing, and the huge number of channels and multiplexing of the input signals found in modern PET systems. It is also difficult to modify or extend the event-processing technologies when needs arise because it would involve making changes to the circuitry. These limitations can be circumvented by applying digital signal-processing technologies for analyzing event pulses generated in PET. With digital technologies, optimized event-processing algorithms can be implemented and they can be modified or extended with ease when needed. The resulting PET data-acquisition (DAQ) system is easier to calibrate and maintain, can generate more accurate event information, and has better extendibility. In this paper, we present our work toward developing a scalable all-digital DAQ system for PET, built upon a personal-computer platform for reducing cost. We will present the overall architecture of this digital DAQ system, and describe our implementations of several components of the system.
We believe that small-animal positron emission tomography (<i>μ</i>PET) can play an important role in phenotyping and drug screening. For such applications, imaging throughput becomes an important issue because one needs to image a
considerable number of subjects in a study. Toward enabling high-throughput <i>μ</i>PET imaging, we are developing
a prototype that consists of two large-area, high-performance flat detectors. These detectors are placed opposed to each other with a small spacing for
providing large detection solid angle and detection sensitive volume. The resulting scanner geometry produces data having missing views and projection truncations, therefore posing a particular challenge in
reconstruction. In this paper, we developed a new iterative reconstruction method that addresses this challenge. By using 2D simulated data, we find that this new method can accurately reconstruct an extended
detection volume of the prototype. Because our prototype shares the same configuration with positron emission mammography (PEM), the new reconstruction method is also applicable for PEM reconstruction.
Image reconstruction from few-view CT is of interest because of the potential to reduce scanning time and radiation dose. The challenge of few-view CT for image reconstruction is essentially a problem of interpolation from under-sampled data. Recently, a new algorithm for inverting the Fourier transform from under-sampled data has been developed by Candes et al. <i>IEEE Trans. Inf. Theory</i> , <b>52</b> 489 (2006). This algorithm can be directly applied to image reconstruction in 2D parallel-beam CT because of the central slice theorem. This article presents a discussion of the new algorithm, showing examples for different degrees of under-sampling.
Because the effects of physical factors such as photon attenuation and spatial resolution are distance-dependent in single-photon emission computed tomography (SPECT), it has been widely assumed that accurate image reconstruction requires knowledge of the data function over 2(π) . In SPECT with uniform attenuation, Noo and Wagner recently showed that an accurate image can be reconstructed from knowledge of the data function over a contiguous (π) -segment. More generally, we proposed (π) -scheme SPECT that entails data acquisition over disjoint angular intervals without conjugate views, totaling to (π) radians, thereby allowing flexibility in choosing projection views at which the emitted gamma-rays may undergo the least attenuation and blurring. In this work, we study the general properties of the (π) -scheme inverse exponential Radon Transform, and discuss how to take advantage of the (π) -scheme flexibility to improve noise properties of short-scan SPECT.
We propose a new noise-reduction method for tomographic reconstruction. The method incorporates a priori information on the source image for allowing the derivation of the energy spectrum of its ideal sinogram. In combination with the energy spectrum of the Poisson noise in the measured sinogram, we are able to derive a Wiener-like filter for effective suppression of the sinogram noise. The filtered backprojection (FBP) algorithm, with a ramp filter, is then applied to the filtered sinogram to produce tomographic images. The resulting filter has a closed-form expression in the frequency space and contains a single user-adjustable regularization parameter. The proposed method is hence simple to implement and easy to use. In contrast to the ad hoc apodizing windows, such as Hanning and Butterworth filters, that are commonly used in the conventional FBP reconstruction, the proposed filter is theoretically more rigorous as it is derived by basing upon an optimization criterion, subject to a known class of source image intensity distributions.
Proc. SPIE. 3978, Medical Imaging 2000: Physiology and Function from Multidimensional Images
KEYWORDS: Signal to noise ratio, Principal component analysis, Statistical analysis, Data modeling, Interference (communication), Medical imaging, Functional imaging, Analytical research, Positron emission tomography, Factor analysis
Factor analysis of medical image sequences (FAMIS), in which one concerns the problem of simultaneous identification of homogeneous regions (factor images) and the characteristic temporal variations (factors) inside these regions from a temporal sequence of images by statistical analysis, is one of the major challenges in medical imaging. In this research, we contribute to this important area of research by proposing a two-step approach. First, we study the use of the noise- adjusted principal component (NAPC) analysis developed by Lee et. al. for identifying the characteristic temporal variations in dynamic scans acquired by PET and MRI. NAPC allows us to effectively reject data noise and substantially reduce data dimension based on signal-to-noise ratio consideration. Subsequently, a simple spatial analysis based on the criteria of minimum spatial overlapping and non-negativity of the factor images is applied for extraction of the factors and factor images. In our simulation study, our preliminary results indicate that the proposed approach can accurately identify the factor images. However, the factors are not completely separated.
KEYWORDS: 3D acquisition, 3D image reconstruction, Signal attenuation, Fourier transforms, Computer simulations, 3D modeling, Collimators, Spatial resolution, Single photon emission computed tomography, 3D image processing
SPECT can potentially be used for quantitative imaging of in vivo 3D radiopharmaceutical distributions. Attempts for accurate quantitation in 3D SPECT images have been compromise not only by the physical effects of photon attenuation, distance-dependent spatial resolution, and scattering, but also by the lack of effective and efficient methods that will correct for these effects. In this work, we introduce a one-step method that can effectively compensate for the effects of photon attenuation and distance-dependent spatial resolution in 3D SPECT. The correction for these effects requires only a very limited amount of computation in addition to that for 3D reconstruction and hence has the potential for routine clinical application. We use both computer-generated simulations and real data to validate the approach. The results demonstrate that the proposed one-step compensation method results in reconstructed 3D SPECT images with good quantitative information.
Positron emission tomography (PET) is a medical imaging modality which produces valuable functional information, but is limited by the poor image quality it provides. Considerable attention has been payed to the problem of reconstructing images in a way that produces better image resolution and noise properties. In dynamic imaging applications PET data are particularly noisy, thus preventing successful recovery of spatial resolution by signal processing applications. In this paper we show that smoothing of image data using a low-order approximation along the time axis can greatly enhance restoration performance.
The Bayesian approach that employs the concepts of cliques and line sites in its Gibbs prior provides the potential of realistic and objective characterization of boundaries between different regions exhibiting intensity variations in the reconstructed images. In this work, we develop an improved Bayesian approach for accurate detection of boundaries by introducing symmetric cliques and new types of line sites as well as a new calculation scheme. This improved Bayesian approach has been applied to positron emission tomography data from both computer simulations and patient studies. The results demonstrate that the new cliques, line sites, and calculation scheme can enhance the boundary detectability of the Bayesian approach and yield realistic boundaries and hence improved reconstructed images.
We describe an adaptive regularization scheme and show how to incorporate it into either the Algebraic Reconstruction Technique (ART) or Maximum Likelihood-Expectation Maximization (ML-EM) based algorithms for reconstruction of Positron Emission Tomography (PET) images. We demonstrate through qualitative and quantitative experiments that the adaptive regularization technique effectively reduces the noise level in the image, while preserving the fine details of the edge structures in the image. The technique does not introduce any visible artifacts during reconstruction.