Maximum-Likelihood Estimation With a Contracting-Grid Search Algorithm

Neural Architecture Search Survey: A Computer Vision Perspective

In recent years, deep learning (DL) has been widely studied using various methods across the globe, especially with respect to training methods and network structures, proving highly effective in a wide range of tasks and applications, including image, speech, and text recognition. One important aspect of this advancement is involved in the effort of designing and upgrading neural architectures, which has been consistently attempted thus far. However, designing such architectures requires the combined knowledge and know-how of experts from each relevant discipline and a series of trial-and-error steps. In this light, automated neural architecture search (NAS) methods are increasingly at the center of attention; this paper aimed at summarizing the basic concepts of NAS while providing an overview of recent studies on the applications of NAS. It is worth noting that most previous survey studies on NAS have been focused on perspectives of hardware or search strategies. To the best knowledge of the present authors, this study is the first to look at NAS from a computer vision perspective. In the present study, computer vision areas were categorized by task, and recent trends found in each study on NAS were analyzed in detail.

Stroke Risk Prediction with Hybrid Deep Transfer Learning Framework

Stroke has become a leading cause of death and long-term disability in the world, and there is no effective treatment.Deep learning-based approaches have the potential to outperform existing stroke risk prediction models, they rely on large well-labeled data. Due to the strict privacy protection policy in health-care systems, stroke data is usually distributed among different hospitals in small pieces. In addition, the positive and negative instances of such data are extremely imbalanced. Transfer learning solves small data issue by exploiting the knowledge of a correlated domain, especially when multiple source are available.In this work, we propose a novel Hybrid Deep Transfer Learning-based Stroke Risk Prediction (HDTL-SRP) scheme to exploit the knowledge structure from multiple correlated sources (i.e.,external stroke data, chronic diseases data, such as hypertension and diabetes). The proposed framework has been extensively tested in synthetic and real-world scenarios, and it outperforms the state-of-the-art stroke risk prediction models. It also shows the potential of real-world deployment among multiple hospitals aided with 5G/B5G infrastructures.

Deep learning based methods for gamma ray interaction location estimation in monolithic scintillation crystal detectors

In this work, we explore deep learning based techniques using the information from mean detector response functions (MDRFs) as a new method to estimate gamma ray interaction location in monolithic scintillation crystal detectors. Compared with searching based methods, deep learning techniques do not require recording all the MDRF information once the prediction networks are trained, which means the memory cost could be significantly reduced. In addition, the event positioning process using deep learning techniques only requires running through the network once, without the need to do searching in the reference dataset. This could greatly speed up the positioning process for each event. We have designed and trained four different neural networks to estimate the gamma ray interaction location given the MDRF data. We have studied network structures consisting only of fully connected (FC) layers, as well as Conv neural networks (CNNs). In addition, we tried to use both regression and classification to generate the final prediction of the gamma ray interaction position. We evaluated the estimation accuracy, testing speed and memory cost (numbers of parameters) of different network architectures, and also compared them with the exhaustive search method. Our results indicate that deep learning based estimation methods with a well designed network structure can achieve a relative positioning error with respect to the ground truth determined by the exhaustive search method of below 1 mm in both x and y directions (depth information is not considered in this work), which would imply a very high performance positioning algorithm for practical monolithic scintillation crystal detectors. The deep learning network also achieves a testing speed that is more than 400 times faster than the exhaustive search method. With proper design of the network structure, the deep learning based positioning methods have the potential to save memory cost by a factor of up to 100.

Quantifying the loss of information from binning list-mode data

List-mode data are increasingly being used in single photon emission computed tomography (SPECT) and positron emission tomography (PET) imaging, among other imaging modalities. However, there are still many imaging designs that effectively bin list-mode data before image reconstruction or other estimation tasks are performed. Intuitively, the binning operation should result in a loss of information. In this work, we show that this is true for Fisher information and provide a computational method for quantifying the information loss. In the end, we find that the information loss depends on three factors. The first factor is related to the smoothness of the mean data function for the list-mode data. The second factor is the actual object being imaged. Finally, the third factor is the binning scheme in relation to the other two factors.

More than BOLD: Dual-spin populations create functional contrast

PURPOSE

Functional MRI contrast has generally been associated with changes in transverse relaxivity caused by blood oxygen concentration, the so-called blood oxygen level dependent contrast. However, this interpretation of fMRI contrast has been called into question by several recent experiments at high spatial resolution. Experiments were conducted to examine contrast dependencies that cannot be explained only by differences in relaxivity in a single-spin population.

METHODS

Measurements of functional signal and contrast were obtained in human early visual cortex during a high-contrast visual stimulation over a large range of TEs and for several flip angles. Small voxels (1.5 mm) were used to restrict the measurements to cortical gray matter in early visual areas identified using retinotopic mapping procedures.

RESULTS

Measurements were consistent with models that include 2 spin populations. The dominant population has a relatively short transverse lifetime that is strongly modulated by activation. However, functional contrast is also affected by volume changes between this short-lived population and the longer-lived population.

CONCLUSION

Some of the previously observed "nonclassical" behaviors of functional contrast can be explained by these interacting dual-spin populations.

Fast gamma-ray interaction-position estimation using k-d tree search

We have developed a fast gamma-ray interaction-position estimation method using k-d tree search, which can be combined with various kinds of closeness metrics such as Euclidean distance, maximum-likelihood estimation, etc. Compared with traditional search strategies, this method can achieve both speed and accuracy at the same time using the k-d tree data structure. The k-d tree search method has a time complexity of [Formula: see text], where N is the number of entries in the reference data set, which means large reference datasets can be used to efficiently estimate each event's interaction position. This method's accuracy was found to be equal to that of the exhaustive search method, yielding the highest achievable accuracy. Most importantly, this method has no restriction on the data structure of the reference dataset and can still work with complicated mean-detector-response functions (MDRFs), meaning that it is more robust than other popular methods such as contracting-grid-search (CG) or vector-search (VS) methods that could yield locally optimal instead of globally optimal results.

Design and Analysis of a High-Gain and Robust Multi-DOF Electro-thermally Actuated MEMS Gyroscope

This paper presents the design and analysis of a multi degree of freedom (DOF) electro-thermally actuated non-resonant MEMS gyroscope with a 3-DOF drive mode and 1-DOF sense mode system. The 3-DOF drive mode system consists of three masses coupled together using suspension beams. The 1-DOF system consists of a single mass whose motion is decoupled from the drive mode using a decoupling frame. The gyroscope is designed to be operated in the flat region between the first two resonant peaks in drive mode, thus minimizing the effect of environmental and fabrication process variations on device performance. The high gain in the flat operational region is achieved by tuning the suspension beams stiffness. A detailed analytical model, considering the dynamics of both the electro-thermal actuator and multi-mass system, is developed. A parametric optimization is carried out, considering the microfabrication process constraints of the Metal Multi-User MEMS Processes (MetalMUMPs), to achieve high gain. The stiffness of suspension beams is optimized such that the sense mode resonant frequency lies in the flat region between the first two resonant peaks in the drive mode. The results acquired through the developed analytical model are verified with the help of 3D finite element method (FEM)-based simulations. The first three resonant frequencies in the drive mode are designed to be 2.51 kHz, 3.68 kHz, and 5.77 kHz, respectively. The sense mode resonant frequency is designed to be 3.13 kHz. At an actuation voltage of 0.2 V, the dynamically amplified drive mode gain in the sense mass is obtained to be 18.6 µm. With this gain, a capacitive change of 28.11   f F and 862.13   f F is achieved corresponding to the sense mode amplitude of 0.15   μ m and 4.5   μ m at atmospheric air pressure and in a vacuum, respectively.

An edge-readout, multilayer detector for positron emission tomography

PURPOSE

We present a novel gamma-ray-detector design based on total internal reflection (TIR) of scintillation photons within a crystal that addresses many limitations of traditional PET detectors. Our approach has appealing features, including submillimeter lateral resolution, DOI positioning from layer thickness, and excellent energy resolution. The design places light sensors on the edges of a stack of scintillator slabs separated by small air gaps and exploits the phenomenon that more than 80% of scintillation light emitted during a gamma-ray event reaches the edges of a thin crystal with polished faces due to TIR. Gamma-ray stopping power is achieved by stacking multiple layers, and DOI is determined by which layer the gamma ray interacts in.

METHOD

The concept of edge readouts of a thin slab was verified by Monte Carlo simulation of scintillation light transport. An LYSO crystal of dimensions 50.8 mm × 50.8 mm × 3.0 mm was modeled with five rectangular SiPMs placed along each edge face. The mean-detector-response functions (MDRFs) were calculated by simulating signals from 511 keV gamma-ray interactions in a grid of locations. Simulations were carried out to study the influence of choice of scintillator material and dimensions, gamma-ray photon energies, introduction of laser or mechanically induced optical barriers (LIOBs, MIOBs), and refractive indices of optical-coupling media and SiPM windows. We also analyzed timing performance including influence of gamma-ray interaction position and presence of optical barriers. We also modeled and built a prototype detector, a 27.4 mm × 27.4 mm × 3.0 mm CsI(Tl) crystal with 4 SiPMs per edge to experimentally validate the results predicted by the simulations. The prototype detector used CsI(Tl) crystals from Proteus outfitted with 16 Hamamatsu model S13360-6050PE MPPCs read out by an AiT-16-channel readout. The MDRFs were measured by scanning the detector with a collimated beam of 662-keV photons from a ¹³⁷ Cs source. The spatial resolution was experimentally determined by imaging a tungsten slit that created a beam of 0.44 mm (FWHM) width normal to the detector surface. The energy resolution was evaluated by analyzing list-mode data from flood illumination by the ¹³⁷ Cs source.

RESULT

We find that in a block-detector-sized LYSO layer read out by five SiPMs per edge, illuminated by 511-keV photons, the average resolution is 1.49 mm (FWHM). With the introduction of optical barriers, average spatial resolution improves to 0.56 mm (FWHM). The DOI resolution is the layer thickness of 3.0 mm. We also find that optical-coupling media and SiPM-window materials have an impact on spatial resolution. The timing simulation with LYSO crystal yields a coincidence resolving time (CRT) of 200-400 ps, which is slightly position dependent. And the introduction of optical barriers has minimum influence. The prototype CsI(Tl) detector, with a smaller area and fewer SiPMs, was measured to have central-area spatial resolutions of 0.70 and 0.39 mm without and with optical barriers, respectively. These results match well with our simulations. An energy resolution of 6.4% was achieved at 662 keV.

CONCLUSION

A detector design based on a stack of monolithic scintillator layers that uses edge readouts offers several advantages over current block detectors for PET. For example, there is no tradeoff between spatial resolution and detection sensitivity since no reflector material displaces scintillator crystal, and submillimeter resolution can be achieved. DOI information is readily available, and excellent timing and energy resolutions are possible.

Null functions in three-dimensional imaging of alpha and beta particles

Null functions of an imaging system are functions in the object space that give exactly zero data. Hence, they represent the intrinsic limitations of the imaging system. Null functions exist in all digital imaging systems, because these systems map continuous objects to discrete data. However, the emergence of detectors that measure continuous data, e.g. particle-processing (PP) detectors, has the potential to eliminate null functions. PP detectors process signals produced by each particle and estimate particle attributes, which include two position coordinates and three components of momentum, as continuous variables. We consider Charged-Particle Emission Tomography (CPET), which relies on data collected by a PP detector to reconstruct the 3D distribution of a radioisotope that emits alpha or beta particles, and show empirically that the null functions are significantly reduced for alpha particles if ≥3 attributes are measured or for beta particles with five attributes measured.

Maximum-Likelihood Estimation of Scintillation Pulse Timing

Including time-of-flight information in positron emission tomography (PET) reconstruction increases the signal-to-noise ratio if the timing information is sufficiently accurate. We estimate timing information by analyzing sampled waveforms, where the sampling frequency and number of samples acquired affect the accuracy of timing estimation. An efficient data-acquisition system acquires the minimum number of samples that contains the most timing information for a desired resolution. We describe a maximum-likelihood (ML) estimation algorithm to assign a time stamp to digital pulses. The method is based on a contracting-grid search algorithm that can be implemented in a field-programmable gate array and in graphics processing units. The Fisher-information (FI) matrix quantifies the amount of timing information that can be extracted from the waveforms. FI analyses on different segments of the waveform allow us to determine the smallest amount of data that we need to acquire in order to obtain a desired timing resolution. We describe the model and the procedure used to simulate waveforms for ML estimation and FI analysis, the ML-estimation algorithm and the timing resolution obtained from experimental data using a LaBr₃:Ce crystal and two photomultiplier tubes. The results show that for lengthening segments of the pulse, timing resolution approaches a limit. We explored the method as a function of sampling frequency and compared the results to other digital time pickoff methods. This information will be used to build an efficient data-acquisition system with reduced complexity and cost that nonetheless preserves full timing performance.

Charged-particle emission tomography

Purpose

Conventional charged‐particle imaging techniques — such as autoradiography — provide only two‐dimensional (2D) black ex vivo images of thin tissue slices. In order to get volumetric information, images of multiple thin slices are stacked. This process is time consuming and prone to distortions, as registration of 2D images is required. We propose a direct three‐dimensional (3D) autoradiography technique, which we call charged‐particle emission tomography (CPET). This 3D imaging technique enables imaging of thick tissue sections, thus increasing laboratory throughput and eliminating distortions due to registration. CPET also has the potential to enable in vivo charged‐particle imaging with a window chamber or an endoscope.

Methods

Our approach to charged‐particle emission tomography uses particle‐processing detectors (PPDs) to estimate attributes of each detected particle. The attributes we estimate include location, direction of propagation, and/or the energy deposited in the detector. Estimated attributes are then fed into a reconstruction algorithm to reconstruct the 3D distribution of charged‐particle‐emitting radionuclides. Several setups to realize PPDs are designed. Reconstruction algorithms for CPET are developed.

Results

Reconstruction results from simulated data showed that a PPD enables CPET if the PPD measures more attributes than just the position from each detected particle. Experiments showed that a two‐foil charged‐particle detector is able to measure the position and direction of incident alpha particles.

Conclusions

We proposed a new volumetric imaging technique for charged‐particle‐emitting radionuclides, which we have called charged‐particle emission tomography (CPET). We also proposed a new class of charged‐particle detectors, which we have called particle‐processing detectors (PPDs). When a PPD is used to measure the direction and/or energy attributes along with the position attributes, CPET is feasible.

Radiance and photon noise: imaging in geometrical optics, physical optics, quantum optics and radiology

The statistics of detector outputs produced by an imaging system are derived from basic radiometric concepts and definitions. We show that a fundamental way of describing a photon-limited imaging system is in terms of a Poisson random process in spatial, angular, and wavelength variables. We begin the paper by recalling the concept of radiance in geometrical optics, radiology, physical optics, and quantum optics. The propagation and conservation laws for radiance in each of these domains are reviewed. Building upon these concepts, we distinguish four categories of imaging detectors that all respond in some way to the incident radiance, including the new category of photon-processing detectors (capable of measuring radiance on a photon-by-photon basis). This allows us to rigorously show how the concept of radiance is related to the statistical properties of detector outputs and to the information content of a single detected photon. A Monte-Carlo technique, which is derived from the Boltzmann transport equation, is presented as a way to estimate probability density functions to be used in reconstruction from photon-processing data.

Singular value decomposition for photon-processing nuclear imaging systems and applications for reconstruction and computing null functions

Recent advances in technology are enabling a new class of nuclear imaging systems consisting of detectors that use real-time maximum-likelihood (ML) methods to estimate the interaction position, deposited energy, and other attributes of each photon-interaction event and store these attributes in a list format. This class of systems, which we refer to as photon-processing (PP) nuclear imaging systems, can be described by a fundamentally different mathematical imaging operator that allows processing of the continuous-valued photon attributes on a per-photon basis. Unlike conventional photon-counting (PC) systems that bin the data into images, PP systems do not have any binning-related information loss. Mathematically, while PC systems have an infinite-dimensional null space due to dimensionality considerations, PP systems do not necessarily suffer from this issue. Therefore, PP systems have the potential to provide improved performance in comparison to PC systems. To study these advantages, we propose a framework to perform the singular-value decomposition (SVD) of the PP imaging operator. We use this framework to perform the SVD of operators that describe a general two-dimensional (2D) planar linear shift-invariant (LSIV) PP system and a hypothetical continuously rotating 2D single-photon emission computed tomography (SPECT) PP system. We then discuss two applications of the SVD framework. The first application is to decompose the object being imaged by the PP imaging system into measurement and null components. We compare these components to the measurement and null components obtained with PC systems. In the process, we also present a procedure to compute the null functions for a PC system. The second application is designing analytical reconstruction algorithms for PP systems. The proposed analytical approach exploits the fact that PP systems acquire data in a continuous domain to estimate a continuous object function. The approach is parallelizable and implemented for graphics processing units (GPUs). Further, this approach leverages another important advantage of PP systems, namely the possibility to perform photon-by-photon real-time reconstruction. We demonstrate the application of the approach to perform reconstruction in a simulated 2D SPECT system. The results help to validate and demonstrate the utility of the proposed method and show that PP systems can help overcome the aliasing artifacts that are otherwise intrinsically present in PC systems.

Evaluation of position-estimation methods applied to CZT-based photon-counting detectors for dedicated breast CT

Semiconductor photon-counting detectors based on high atomic number, high density materials [cadmium zinc telluride (CZT)/cadmium telluride (CdTe)] for x-ray computed tomography (CT) provide advantages over conventional energy-integrating detectors, including reduced electronic and Swank noise, wider dynamic range, capability of spectral CT, and improved signal-to-noise ratio. Certain CT applications require high spatial resolution. In breast CT, for example, visualization of microcalcifications and assessment of tumor microvasculature after contrast enhancement require resolution on the order of [Formula: see text]. A straightforward approach to increasing spatial resolution of pixellated CZT-based radiation detectors by merely decreasing the pixel size leads to two problems: (1) fabricating circuitry with small pixels becomes costly and (2) inter-pixel charge spreading can obviate any improvement in spatial resolution. We have used computer simulations to investigate position estimation algorithms that utilize charge sharing to achieve subpixel position resolution. To study these algorithms, we model a simple detector geometry with a [Formula: see text] array of [Formula: see text] pixels, and use a conditional probability function to model charge transport in CZT. We used COMSOL finite element method software to map the distribution of charge pulses and the Monte Carlo package PENELOPE for simulating fluorescent radiation. Performance of two x-ray interaction position estimation algorithms was evaluated: the method of maximum-likelihood estimation and a fast, practical algorithm that can be implemented in a readout application-specific integrated circuit and allows for identification of a quadrant of the pixel in which the interaction occurred. Both methods demonstrate good subpixel resolution; however, their actual efficiency is limited by the presence of fluorescent [Formula: see text]-escape photons. Current experimental breast CT systems typically use detectors with a pixel size of [Formula: see text], with [Formula: see text] binning during the acquisition giving an effective pixel size of [Formula: see text]. Thus, it would be expected that the position estimate accuracy reported in this study would improve detection and visualization of microcalcifications as compared to that with conventional detectors.

Estimating ROI activity concentration with photon-processing and photon-counting SPECT imaging systems

Recently a new class of imaging systems, referred to as photon-processing (PP) systems, are being developed that uses real-time maximum-likelihood (ML) methods to estimate multiple attributes per detected photon and store these attributes in a list format. PP systems could have a number of potential advantages compared to systems that bin photons based on attributes such as energy, projection angle, and position, referred to as photon-counting (PC) systems. For example, PP systems do not suffer from binning-related information loss and provide the potential to extract information from attributes such as energy deposited by the detected photon. To quantify the effects of this advantage on task performance, objective evaluation studies are required. We performed this study in the context of quantitative 2-dimensional single-photon emission computed tomography (SPECT) imaging with the end task of estimating the mean activity concentration within a region of interest (ROI). We first theoretically outline the effect of null space on estimating the mean activity concentration, and argue that due to this effect, PP systems could have better estimation performance compared to PC systems with noise-free data. To evaluate the performance of PP and PC systems with noisy data, we developed a singular value decomposition (SVD)-based analytic method to estimate the activity concentration from PP systems. Using simulations, we studied the accuracy and precision of this technique in estimating the activity concentration. We used this framework to objectively compare PP and PC systems on the activity concentration estimation task. We investigated the effects of varying the size of the ROI and varying the number of bins for the attribute corresponding to the angular orientation of the detector in a continuously rotating SPECT system. The results indicate that in several cases, PP systems offer improved estimation performance compared to PC systems.

Impact of the Fano Factor on Position and Energy Estimation in Scintillation Detectors

The Fano factor for an integer-valued random variable is defined as the ratio of its variance to its mean. Light from various scintillation crystals have been reported to have Fano factors from sub-Poisson (Fano factor < 1) to super-Poisson (Fano factor > 1). For a given mean, a smaller Fano factor implies a smaller variance and thus less noise. We investigated if lower noise in the scintillation light will result in better spatial and energy resolutions. The impact of Fano factor on the estimation of position of interaction and energy deposited in simple gamma-camera geometries is estimated by two methods - calculating the Cramér-Rao bound and estimating the variance of a maximum likelihood estimator. The methods are consistent with each other and indicate that when estimating the position of interaction and energy deposited by a gamma-ray photon, the Fano factor of a scintillator does not affect the spatial resolution. A smaller Fano factor results in a better energy resolution.

Task-based measures of image quality and their relation to radiation dose and patient risk

The theory of task-based assessment of image quality is reviewed in the context of imaging with ionizing radiation, and objective figures of merit (FOMs) for image quality are summarized. The variation of the FOMs with the task, the observer and especially with the mean number of photons recorded in the image is discussed. Then various standard methods for specifying radiation dose are reviewed and related to the mean number of photons in the image and hence to image quality. Current knowledge of the relation between local radiation dose and the risk of various adverse effects is summarized, and some graphical depictions of the tradeoffs between image quality and risk are introduced. Then various dose-reduction strategies are discussed in terms of their effect on task-based measures of image quality.

Volume estimation of low-contrast lesions with CT: a comparison of performances from a phantom study, simulations and theoretical analysis

Measurements of lung nodule volume with multi-detector computed tomography (MDCT) have been shown to be more accurate and precise compared to conventional lower dimensional measurements. Quantifying the size of lesions is potentially more difficult when the object-to-background contrast is low as with lesions in the liver. Physical phantom and simulation studies are often utilized to analyze the bias and variance of lesion size estimates because a ground truth or reference standard can be established. In addition, it may also be useful to derive theoretical bounds as another way of characterizing lesion sizing methods. The goal of this work was to study the performance of a MDCT system for a lesion volume estimation task with object-to-background contrast less than 50 HU, and to understand the relation among performances obtained from phantom study, simulation and theoretical analysis. We performed both phantom and simulation studies, and analyzed the bias and variance of volume measurements estimated by a matched-filter-based estimator. We further corroborated results with a theoretical analysis to estimate the achievable performance bound, which was the Cramer-Rao's lower bound (CRLB) of minimum variance for the size estimates. Results showed that estimates of non-attached solid small lesion volumes with object-to-background contrast of 31-46 HU can be accurate and precise, with less than 10.8% in percent bias and 4.8% in standard deviation of percent error (SPE), in standard dose scans. These results are consistent with theoretical (CRLB), computational (simulation) and empirical phantom bounds. The difference between the bounds is rather small (for SPE less than 1.9%) indicating that the theoretical- and simulation-based performance bounds can be good surrogates for physical phantom studies.

Evaluation of a compact, high-resolution SPECT detector based on digital silicon photomultipliers

MicroSPECT is one of the main functional imaging techniques used in the preclinical setting. Even though high-resolution images can be obtained with currently available systems, their sensitivity is often quite low due to the use of multi-pinhole collimation. This results in long acquisition times and hampers dynamic imaging. However, it has already been shown that this limited sensitivity can be overcome using high-resolution detectors. In this article, we therefore investigated the use of a digital photon counter (DPC) in combination with a 2 mm thick monolithic LYSO crystal for SPECT imaging. These light sensors contain arrays of avalanche photodiodes whose signals are directly digitised. The DPCs have the advantage that they are very compact, have a high intrinsic resolution, are MR compatible and allow disabling cells with a high dark count rate. In order to investigate the influence of the temperature dependent dark count rate on the detector performance, we compared it at 3 °C and 18 °C. At 3 °C, we observed an energy resolution of 28.8% and an intrinsic spatial resolution of 0.48 mm. Furthermore, the count rate at 10% loss is 60 kcps. Next, we looked at the event loss at 18 °C caused by the higher dark count rate and found a 5% loss compared to the 3 °C measurements. At this higher temperature the energy resolution becomes 29.2% and the intrinsic spatial resolution decreases to 0.52 mm. Due to the 5% count loss, the count rate at 10% loss increases to 63 kcps. A small degradation of the detector performance is thus observed at 18 °C.These results show the usefulness of this detector for SPECT imaging together with its excellent intrinsic spatial resolution. A drawback of the detector is its low, spatially varying energy resolution. Even though the detection efficiency and intrinsic spatial resolution are better at 3 °C, results are still acceptable at 18 °C.

Molecular Imaging in the College of Optical Sciences - An Overview of Two Decades of Instrumentation Development

During the past two decades, researchers at the University of Arizona's Center for Gamma-Ray Imaging (CGRI) have explored a variety of approaches to gamma-ray detection, including scintillation cameras, solid-state detectors, and hybrids such as the intensified Quantum Imaging Device (iQID) configuration where a scintillator is followed by optical gain and a fast CCD or CMOS camera. We have combined these detectors with a variety of collimation schemes, including single and multiple pinholes, parallel-hole collimators, synthetic apertures, and anamorphic crossed slits, to build a large number of preclinical molecular-imaging systems that perform Single-Photon Emission Computed Tomography (SPECT), Positron Emission Tomography (PET), and X-Ray Computed Tomography (CT). In this paper, we discuss the themes and methods we have developed over the years to record and fully use the information content carried by every detected gamma-ray photon.

A SPECT system simulator built on the SolidWorks TM 3D-Design package

We have developed a GPU-accelerated SPECT system simulator that integrates into instrument-design workflow [1]. This simulator includes a gamma-ray tracing module that can rapidly propagate gamma-ray photons through arbitrary apertures modeled by SolidWorks ^TM -created stereolithography (.STL) representations with a full complement of physics cross sections [2, 3]. This software also contains a scintillation detector simulation module that can model a scintillation detector with arbitrary scintillation crystal shape and light-sensor arrangement. The gamma-ray tracing module enables us to efficiently model aperture and detector crystals in SolidWorks ^TM and save them as STL file format, then load the STL-format model into this module to generate list-mode results of interacted gamma-ray photon information (interaction positions and energies) inside the detector crystals. The Monte-Carlo scintillation detector simulation module enables us to simulate how scintillation photons get reflected, refracted and absorbed inside a scintillation detector, which contributes to more accurate simulation of a SPECT system.

RADIANCE AND PHOTON NOISE: Imaging in geometrical optics, physical optics, quantum optics and radiology

A fundamental way of describing a photon-limited imaging system is in terms of a Poisson random process in spatial, angular and wavelength variables. The mean of this random process is the spectral radiance. The principle of conservation of radiance then allows a full characterization of the noise in the image (conditional on viewing a specified object). To elucidate these connections, we first review the definitions and basic properties of radiance as defined in terms of geometrical optics, radiology, physical optics and quantum optics. The propagation and conservation laws for radiance in each of these domains are reviewed. Then we distinguish four categories of imaging detectors that all respond in some way to the incident radiance, including the new category of photon-processing detectors. The relation between the radiance and the statistical properties of the detector output is discussed and related to task-based measures of image quality and the information content of a single detected photon.

Design and performance of a compact and stationary microSPECT system

PURPOSE

Over the last ten years, there has been an extensive growth in the development of microSPECT imagers. Most of the systems are based on the combination of conventional, relatively large gamma cameras with poor intrinsic spatial resolution and multipinhole collimators working in large magnification mode. Spatial resolutions range from 0.58 to 0.76 mm while peak sensitivities vary from 0.06% to 0.4%. While pushing the limits of performance is of major importance, the authors believe that there is a need for smaller and less complex systems that bring along a reduced cost. While low footprint and low-cost systems can make microSPECT available to more researchers, the ease of operation and calibration and low maintenance cost are additional factors that can facilitate the use of microSPECT in molecular imaging. In this paper, the authors simulate the performance of a microSPECT imager that combines high space-bandwidth detectors and pinholes with truncated projection, resulting in a small and stationary system.

METHODS

A system optimization algorithm is used to determine the optimal SPECT systems, given our high resolutions detectors and a fixed field-of-view. These optimal system geometries are then used to simulate a Defrise disk phantom and a hot rod phantom. Finally, a MOBY mouse phantom, with realistic concentrations of Tc99m-tetrofosmin is simulated.

RESULTS

Results show that the authors can successfully reconstruct a Defrise disk phantom of 24 mm in diameter without any rotating system components or translation of the object. Reconstructed spatial resolution is approximately 800 μm while the peak sensitivity is 0.23%. Finally, the simulation of the MOBY mouse phantom shows that the authors can accurately reconstruct mouse images.

CONCLUSIONS

These results show that pinholes with truncated projections can be used in small magnification or minification mode to obtain a compact and stationary microSPECT system. The authors showed that they can reach state-of-the-art system performance and can successfully reconstruct images with realistic noise levels in a preclinical context. Such a system can be useful for dynamic SPECT imaging.

High-Resolution Anamorphic SPECT Imaging

We have developed a gamma-ray imaging system that combines a high-resolution silicon detector with two sets of movable, half-keel-edged copper-tungsten blades configured as crossed slits. These apertures can be positioned independently between the object and detector, producing an anamorphic image in which the axial and transaxial magnifications are not constrained to be equal. The detector is a 60 mm × 60 mm, one-millimeter-thick, one-megapixel silicon double-sided strip detector with a strip pitch of 59 μm. The flexible nature of this system allows the application of adaptive imaging techniques. We present system details; calibration, acquisition, and reconstruction methods; and imaging results.

A high resolution SPECT detector based on thin continuous LYSO

Single-photon emission computed tomography (SPECT) detectors with improved spatial resolution can be used to build multi-pinhole SPECT systems that have a higher sensitivity or a higher spatial resolution. In order to improve the spatial resolution we investigate the performance of a 2 mm thick continuous Lutetium Yttrium Orthosilicate (LYSO) scintillator and compare it to the performance of a 5 mm thick continuous NaI(Tl) scintillator. The advantages of LYSO are its high stopping power and its non-hygroscopicity. Drawbacks are the lower light output and the intrinsic radioactivity. The hypothesis of this study is that such a thin LYSO scintillator will have a small light spread and, as a consequence, will also have an improved spatial resolution when coupled to a Hamamatsu H8500 position sensitive photomultiplier tube. To optimize the spatial resolution and the useful detector area we used a mean nearest neighbor event-positioning method. Beam source measurements ((99m)Tc, 140 keV) were done to investigate the energy resolution and the spatial resolution of both detectors. The effect of the intrinsic radioactivity of the LYSO scintillator in the energy window was quantified. The mean energy resolution is 9.3% for the NaI(Tl) scintillator and 21.3% for the LYSO scintillator. The LYSO spectrum shows an X-ray escape peak which decreases the detection efficiency with 9.1%. The spatial resolution of the LYSO detector (0.93 mm full width at half maximum (FWHM)) is superior to the spatial resolution of the NaI(Tl) detector (1.37 mm FWHM). The intrinsic radioactivity in the energy window (42% window centered at 140 keV) is low (125.6 cps, 0.024 cps mm(-3)). LYSO is a promising scintillator for small-animal SPECT imaging, where spatial resolution is more important than energy resolution.

Scanning linear estimation: improvements over region of interest (ROI) methods

In tomographic medical imaging, a signal activity is typically estimated by summing voxels from a reconstructed image. We introduce an alternative estimation scheme that operates on the raw projection data and offers a substantial improvement, as measured by the ensemble mean-square error (EMSE), when compared to using voxel values from a maximum-likelihood expectation-maximization (MLEM) reconstruction. The scanning-linear (SL) estimator operates on the raw projection data and is derived as a special case of maximum-likelihood estimation with a series of approximations to make the calculation tractable. The approximated likelihood accounts for background randomness, measurement noise and variability in the parameters to be estimated. When signal size and location are known, the SL estimate of signal activity is unbiased, i.e. the average estimate equals the true value. By contrast, unpredictable bias arising from the null functions of the imaging system affect standard algorithms that operate on reconstructed data. The SL method is demonstrated for two different tasks: (1) simultaneously estimating a signal's size, location and activity; (2) for a fixed signal size and location, estimating activity. Noisy projection data are realistically simulated using measured calibration data from the multi-module multi-resolution small-animal SPECT imaging system. For both tasks, the same set of images is reconstructed using the MLEM algorithm (80 iterations), and the average and maximum values within the region of interest (ROI) are calculated for comparison. This comparison shows dramatic improvements in EMSE for the SL estimates. To show that the bias in ROI estimates affects not only absolute values but also relative differences, such as those used to monitor the response to therapy, the activity estimation task is repeated for three different signal sizes.

Objective assessment of image quality. V. Photon-counting detectors and list-mode data

A theoretical framework for detection or discrimination tasks with list-mode data is developed. The object and imaging system are rigorously modeled via three random mechanisms: randomness of the object being imaged, randomness in the attribute vectors, and, finally, randomness in the attribute vector estimates due to noise in the detector outputs. By considering the list-mode data themselves, the theory developed in this paper yields a manageable expression for the likelihood of the list-mode data given the object being imaged. This, in turn, leads to an expression for the optimal Bayesian discriminant. Figures of merit for detection tasks via the ideal and optimal linear observers are derived. A concrete example discusses detection performance of the optimal linear observer for the case of a known signal buried in a random lumpy background.

High-resolution, anamorphic, adaptive small-animal SPECT imaging with silicon double-sided strip detectors

We are developing a prototype gamma-ray imaging system that consists of two sets of movable, keel-edged copper-tungsten blades configured as crossed slits. These apertures can be positioned independently between the object and detector, producing an anamorphic image in which the axial and transaxial magnifications are not constrained to be equal. The detector is a 60 mm × 60 mm, millimeter thick, one-megapixel silicon double-sided strip detector. The flexible nature of this system allows the application of adaptive imaging techniques. We will discuss system details, calibration and acquisition methods, and our progress towards biological imaging applications.

SPECT detectors: the Anger Camera and beyond

The development of radiation detectors capable of delivering spatial information about gamma-ray interactions was one of the key enabling technologies for nuclear medicine imaging and, eventually, single-photon emission computed tomography (SPECT). The continuous sodium iodide scintillator crystal coupled to an array of photomultiplier tubes, almost universally referred to as the Anger Camera after its inventor, has long been the dominant SPECT detector system. Nevertheless, many alternative materials and configurations have been investigated over the years. Technological advances as well as the emerging importance of specialized applications, such as cardiac and preclinical imaging, have spurred innovation such that alternatives to the Anger Camera are now part of commercial imaging systems. Increased computing power has made it practical to apply advanced signal processing and estimation schemes to make better use of the information contained in the detector signals. In this review we discuss the key performance properties of SPECT detectors and survey developments in both scintillator and semiconductor detectors and their readouts with an eye toward some of the practical issues at least in part responsible for the continuing prevalence of the Anger Camera in the clinic.

Maximum-likelihood scintillation detection for EM-CCD based gamma cameras

Gamma cameras based on charge-coupled devices (CCDs) coupled to continuous scintillation crystals can combine a good detection efficiency with high spatial resolutions with the aid of advanced scintillation detection algorithms. A previously developed analytical multi-scale algorithm (MSA) models the depth-dependent light distribution but does not take statistics into account. Here we present and validate a novel statistical maximum-likelihood algorithm (MLA) that combines a realistic light distribution model with an experimentally validated statistical model. The MLA was tested for an electron multiplying CCD optically coupled to CsI(Tl) scintillators of different thicknesses. For (99m)Tc imaging, the spatial resolution (for perpendicular and oblique incidence), energy resolution and signal-to-background counts ratio (SBR) obtained with the MLA were compared with those of the MSA. Compared to the MSA, the MLA improves the energy resolution by more than a factor of 1.6 and the SBR is enhanced by more than a factor of 1.3. For oblique incidence (approximately 45°), the depth-of-interaction corrected spatial resolution is improved by a factor of at least 1.1, while for perpendicular incidence the MLA resolution does not consistently differ significantly from the MSA result for all tested scintillator thicknesses. For the thickest scintillator (3 mm, interaction probability 66% at 141 keV) a spatial resolution (perpendicular incidence) of 147 µm full width at half maximum (FWHM) was obtained with an energy resolution of 35.2% FWHM. These results of the MLA were achieved without prior calibration of scintillations as is needed for many statistical scintillation detection algorithms. We conclude that the MLA significantly improves the gamma camera performance compared to the MSA.

Synthesis and preliminary evaluation of radiolabeled bis(zinc(II)-dipicolylamine) coordination complexes as cell death imaging agents

The aim of this study was the development of (⁹⁹m)Tc labeled bis(zinc(II)-dipicolylamine) (Zn²⁺-DPA) coordination complexes, and the in vivo evaluation of their usefulness as radiotracers for the detection of cell death. DPA ligand 1 was labeled with (⁹⁹m)Tc via the (⁹⁹m)Tc-tricarbonyl core ([(⁹⁹m)Tc(CO)₃-1]³⁺) or via HYNIC ((⁹⁹m)Tc-HYNIC-1) in good radiochemical yields. Highest in vitro stabilities were demonstrated for [(⁹⁹m)Tc(CO)₃-1]³⁺. A mouse model of hepatic apoptosis (anti-Fas mAb) was used to demonstrate binding to apoptotic cells. (⁹⁹m)Tc-HYNIC-1 showed the best targeting of apoptotic hepatic tissue with a 2.2 times higher liver uptake in anti-Fas treated mice as compared to healthy animals. A rat model of ischemia-reperfusion injury was used to further explore the ability of the (⁹⁹m)Tc-labeled Zn²⁺-DPA coordination complexes to target cell death. Selective accumulation could be detected for both tracers in the area at risk, correlating with histological proof of cell death. Area at risk to normal tissue uptake ratios were 3.82 for [(⁹⁹m)Tc(CO)₃-1]³⁺ and 5.45 for (⁹⁹m)Tc-HYNIC-1.

Maximum-Likelihood Calibration of an X-ray Computed Tomography System

We present a maximum-likelihood (ML) method for calibrating the geometrical parameters of an x-ray computed tomography (CT) system. This method makes use of the full image data and not a reduced set of data. This algorithm is particularly useful for CT systems that change their geometry during the CT acquisition, such as an adaptive CT scan. Our ML search method uses a contracting-grid algorithm that does not require initial starting values to perform its estimate, thus avoiding problems associated with choosing initialization values.

Progress in BazookaSPECT

Recent progress on a high-resolution, photon-counting gamma-ray and x-ray imager called BazookaSPECT is presented. BazookaSPECT is an example of a new class of scintillation detectors based on integrating detectors such as CCD(charge-coupled device) or CMOS(complementary metal-oxide semiconductor) sensors. BazookaSPECT is unique in that it makes use of a scintillator in close proximity to a microchannel plate-based image intensifier for up-front optical amplification of scintillation light. We discuss progress made in bringing about compact BazookaSPECT modules and in real-time processing of event data using graphics processing units (GPUs). These advances are being implemented in the design of a high-resolution rodent brain imager called FastSPECT III. A key benefit of up-front optical gain is that any CCD/CMOS sensor can now be utilized for photon counting. We discuss the benefits and feasibility of using CMOS sensors as photon-counting detectors for digital radiography, with application in mammography and computed tomography (CT). We present as an appendix a formal method for comparing various photon-counting integrating detectors using objective statistical criteria.