Honors & Awards
Young Investigator Award, Society of Nuclear Medicine and Molecular Imaging (2014)
Radiation Physics Impact Award, Stanford University (2013)
Young Investigator Award - Semifinalist, World Molecular Imaging Congress (2013)
Dean's Fellowship, Stanford (2010)
Postdoctoral Fellowship, DoD Breast Cancer Research Program (2010)
Research Seed Grant, American Association of Physicists in Medicine (2010)
Top Student Paper, IEEE Medical Imaging Conference (2008)
Bio-X Graduate Fellowship, Stanford University (2006)
Bradley-Alavi Fellow, Society of Nuclear Medicine (2006)
NVIDIA Fellowship, NVIDIA Corp (2006)
Ph.D., Stanford University, Electrical Engineering (2010)
M.S., Stanford University, Electrical Engineering (2006)
B.S., Ecole Centrale Paris, Engineering (2004)
Current Research and Scholarly Interests
(1) Imaging of radionuclides with single-cell resolution: Currently, radionuclide tracer imaging is the most sensitive assay for probing subtle biochemical processes in living subjects. Molecular imaging with PET and SPECT has become crucial both for basic science research and for patient management. However, little is known about how those radionuclide tracers interact with cell at the single cell level. I am currently developing a new imaging tool call the radioluminescence microscope that can image these tracers in a standard microscopy environment. This new tool allows researchers at Stanford to visualize how radionuclide tracers distribute in a living cell population.
(2) X-ray molecular imaging: Molecular imaging offers the ability to probe subtle biological signals that are characteristic of disease onset and progression. It can also monitor the response of a disease to treatment before any anatomical changes occur. My research explores two emerging imaging techniques that can probe multiple disease biomarkers in a non-invasive fashion. In both imaging techniques, a contrast agent is introduced that can produce a distinguishable signal when irradiated with X-ray. This feature makes it possible to obtain molecular information during a CT examination. The two imaging techniques differ in the following: In X-ray luminescence imaging, the contrast agent is a radioluminescent nanoparticle that produces near-infrared light under X-ray irradiation. In X-ray fluorescence imaging, the contrast agent is a high-atomic-number element that emits a characteristic X-ray signal under irradiation.
(3) High-performance medical computing: Efficient computing now requires using multi- and many-core processors--which embed multiple computing elements in a single chip. New medical imaging algorithms must be designed that are aware of the parallel computing capabilities of new computer hardware. In my work, I develop medical imaging algorithms adapted to these new parallel architectures. Clinically, those algorithms can shorten the time required to process data by as much as tenfold, removing a critical bottleneck in the clinical workflow. One of the most promising platform for medical computing is the graphics processing unit: originally a gadget sought by serious computer gamers, it is now used as an inexpensive supercomputer on-a-chip by researchers in all fields.
X-ray-Induced Shortwave Infrared Biomedical Imaging Using Rare-Earth Nanoprobes.
2015; 15 (1): 96-102
Shortwave infrared (SWIR or NIR-II) light provides significant advantages for imaging biological structures due to reduced autofluorescence and photon scattering. Here, we report on the development of rare-earth nanoprobes that exhibit SWIR luminescence following X-ray irradiation. We demonstrate the ability of X-ray-induced SWIR luminescence (X-IR) to monitor biodistribution and map lymphatic drainage. Our results indicate X-IR imaging is a promising new modality for preclinical applications and has potential for dual-modality molecular disease imaging.
View details for DOI 10.1021/nl504123r
View details for PubMedID 25485705
- Cerenkov Luminescence Endoscopy: Improved Molecular Sensitivity with beta(-)-Emitting Radiotracers JOURNAL OF NUCLEAR MEDICINE 2014; 55 (11): 1905-1909
- Fiber-Optic System for Dual-Modality Imaging of Glucose Probes F-18-FDG and 6-NBDG in Atherosclerotic Plaques PLOS ONE 2014; 9 (9)
Clinical evaluation of a novel intraoperative handheld gamma camera for sentinel lymph node biopsy
PHYSICA MEDICA-EUROPEAN JOURNAL OF MEDICAL PHYSICS
2014; 30 (3): 340-345
Preoperative lymphoscintigraphy (PLS) combined with intraoperative gamma probe (GP) localization is standard procedure for localizing the sentinel lymph nodes (SLN) in melanoma and breast cancer. In this study, we evaluated the ability of a novel intraoperative handheld gamma camera (IHGC) to image SLNs during surgery.The IHGC is a small-field-of-view camera optimized for real-time imaging of lymphatic drainage patterns. Unlike conventional cameras, the IHGC can acquire useful images in a few seconds in a free-running fashion and be moved manually around the patient to find a suitable view of the node. Thirty-nine melanoma and eleven breast cancer patients underwent a modified SLN biopsy protocol in which nodes localized with the GP were imaged with the IHGC. The IHGC was also used to localize additional nodes that could not be found with the GP.The removal of 104 radioactive SLNs was confirmed ex vivo by GP counting. In vivo, the relative node detection sensitivity was 88.5 (82.3, 94.6)% for the IHGC (used in conjunction with the GP) and 94.2 (89.7, 98.7)% for the GP alone, a difference not found to be statistically significant (McNemar test, p = 0.24).Small radioactive SLNs can be visualized intraoperatively using the IHGC with exposure time of 20 s or less, with no significant difference in node detection sensitivity compared to a GP. The IHGC is a useful complement to the GP, especially for SLNs that are difficult to locate with the GP alone.
View details for DOI 10.1016/j.ejmp.2013.10.005
View details for Web of Science ID 000334094200012
L-shell x-ray fluorescence computed tomography (XFCT) imaging of Cisplatin
PHYSICS IN MEDICINE AND BIOLOGY
2014; 59 (1): 219-232
X-ray fluorescence computed tomography (XFCT) imaging has been focused on the detection of K-shell x-rays. The potential utility of L-shell x-ray XFCT is, however, not well studied. Here we report the first Monte Carlo (MC) simulation of preclinical L-shell XFCT imaging of Cisplatin. We built MC models for both L- and K-shell XFCT with different excitation energies (15 and 30 keV for L-shell and 80 keV for K-shell XFCT). Two small-animal sized imaging phantoms of 2 and 4 cm diameter containing a series of objects of 0.6 to 2.7 mm in diameter at 0.7 to 16 mm depths with 10 to 250 µg mL(-1) concentrations of Pt are used in the study. Transmitted and scattered x-rays were collected with photon-integrating transmission detector and photon-counting detector arc, respectively. Collected data were rearranged into XFCT and transmission CT sinograms for image reconstruction. XFCT images were reconstructed with filtered back-projection and with iterative maximum-likelihood expectation maximization without and with attenuation correction. While K-shell XFCT was capable of providing an accurate measurement of Cisplatin concentration, its sensitivity was 4.4 and 3.0 times lower than that of L-shell XFCT with 15 keV excitation beam for the 2 cm and 4 cm diameter phantom, respectively. With the inclusion of excitation and fluorescence beam attenuation correction, we found that L-shell XFCT was capable of providing fairly accurate information of Cisplatin concentration distribution. With a dose of 29 and 58 mGy, clinically relevant Cisplatin Pt concentrations of 10 µg mg(-1) could be imaged with L-shell XFCT inside a 2 cm and 4 cm diameter object, respectively.
View details for DOI 10.1088/0031-9155/59/1/219
View details for Web of Science ID 000328549200011
Hard X-ray-induced optical luminescence via biomolecule-directed metal clusters
2014; 50 (27): 3549-3551
Here, we demonstrate that biomolecule-directed metal clusters are applicable in the study of hard X-ray excited optical luminescence, promising a new direction in the development of novel X-ray-activated imaging probes.
View details for DOI 10.1039/c3cc48661c
View details for Web of Science ID 000332483200003
View details for PubMedID 24463467
Seeing the invisible: Direct visualization of therapeutic radiation beams using air scintillation
2014; 41 (1)
Purpose: To assess whether air scintillation produced during standard radiation treatments can be visualized and used to monitor a beam in a nonperturbing manner.Methods: Air scintillation is caused by the excitation of nitrogen gas by ionizing radiation. This weak emission occurs predominantly in the 300-430 nm range. An electron-multiplication charge-coupled device camera, outfitted with an f∕0.95 lens, was used to capture air scintillation produced by kilovoltage photon beams and megavoltage electron beams used in radiation therapy. The treatment rooms were prepared to block background light and a short-pass filter was utilized to block light above 440 nm.Results: Air scintillation from an orthovoltage unit (50 kVp, 30 mA) was visualized with a relatively short exposure time (10 s) and showed an inverse falloff (r(2) = 0.89). Electron beams were also imaged. For a fixed exposure time (100 s), air scintillation was proportional to dose rate (r(2) = 0.9998). As energy increased, the divergence of the electron beam decreased and the penumbra improved. By irradiating a transparent phantom, the authors also showed that Cherenkov luminescence did not interfere with the detection of air scintillation. In a final illustration of the capabilities of this new technique, the authors visualized air scintillation produced during a total skin irradiation treatment.Conclusions: Air scintillation can be measured to monitor a radiation beam in an inexpensive and nonperturbing manner. This physical phenomenon could be useful for dosimetry of therapeutic radiation beams or for online detection of gross errors during fractionated treatments.
View details for DOI 10.1118/1.4851595
View details for Web of Science ID 000329182200004
View details for PubMedID 24387491
Detection and quantitation of circulating tumor cell dynamics by bioluminescence imaging in an orthotopic mammary carcinoma model.
2014; 9 (9)
Circulating tumor cells (CTCs) have been detected in the bloodstream of both early-stage and advanced cancer patients. However, very little is know about the dynamics of CTCs during cancer progression and the clinical relevance of longitudinal CTC enumeration. To address this, we developed a simple bioluminescence imaging assay to detect CTCs in mouse models of metastasis. In a 4T1 orthotopic metastatic mammary carcinoma mouse model, we demonstrated that this quantitative method offers sensitivity down to 2 CTCs in 0.1-1mL blood samples and high specificity for CTCs originating from the primary tumor, independently of their epithelial status. In this model, we simultaneously monitored blood CTC dynamics, primary tumor growth, and lung metastasis progression over the course of 24 days. Early in tumor development, we observed low numbers of CTCs in blood samples (10-15 cells/100 µL) and demonstrated that CTC dynamics correlate with viable primary tumor growth. To our knowledge, these data represent the first reported use of bioluminescence imaging to detect CTCs and quantify their dynamics in any cancer mouse model. This new assay is opening the door to the study of CTC dynamics in a variety of animal models. These studies may inform clinical decision on the appropriate timing of blood sampling and value of longitudinal CTC enumeration in cancer patients.
View details for DOI 10.1371/journal.pone.0105079
View details for PubMedID 25188396
High-Resolution Radioluminescence Microscopy of F-18-FDG Uptake by Reconstructing the beta-Ionization Track
JOURNAL OF NUCLEAR MEDICINE
2013; 54 (10): 1841-1846
Radioluminescence microscopy is a new method for imaging radionuclide uptake by single live cells with a fluorescence microscope. Here, we report a particle-counting scheme that improves spatial resolution by overcoming the β-range limit.Short frames (10 μs-1 s) were acquired using a high-gain camera coupled to a microscope to capture individual ionization tracks. Optical reconstruction of the β-ionization track (ORBIT) was performed to localize individual β decays, which were aggregated into a composite image. The new approach was evaluated by imaging the uptake of (18)F-FDG in nonconfluent breast cancer cells.After image reconstruction, ORBIT resulted in better definition of individual cells. This effect was particularly noticeable in small clusters (2-4 cells), which occur naturally even for nonconfluent cell cultures. The annihilation and Bremsstrahlung photon background signal was markedly lower. Single-cell measurements of (18)F-FDG uptake that were computed from ORBIT images more closely matched the uptake of the fluorescent glucose analog (Pearson correlation coefficient, 0.54 vs. 0.44, respectively).ORBIT can image the uptake of a radiotracer in living cells with spatial resolution better than the β range. In principle, ORBIT may also allow for greater quantitative accuracy because the decay rate is measured more directly, with no dependency on the β-particle energy.
View details for DOI 10.2967/jnumed.112.113365
View details for Web of Science ID 000325341300027
Distributed MLEM: An Iterative Tomographic Image Reconstruction Algorithm for Distributed Memory Architectures
IEEE TRANSACTIONS ON MEDICAL IMAGING
2013; 32 (5): 957-967
The processing speed for positron emission tomography (PET) image reconstruction has been greatly improved in recent years by simply dividing the workload to multiple processors of a graphics processing unit (GPU). However, if this strategy is generalized to a multi-GPU cluster, the processing speed does not improve linearly with the number of GPUs. This is because large data transfer is required between the GPUs after each iteration, effectively reducing the parallelism. This paper proposes a novel approach to reformulate the maximum likelihood expectation maximization (MLEM) algorithm so that it can scale up to many GPU nodes with less frequent inter-node communication. While being mathematically different, the new algorithm maximizes the same convex likelihood function as MLEM, thus converges to the same solution. Experiments on a multi-GPU cluster demonstrate the effectiveness of the proposed approach.
View details for DOI 10.1109/TMI.2013.2252913
View details for Web of Science ID 000318643500011
View details for PubMedID 23529079
Development of XFCT imaging strategy for monitoring the spatial distribution of platinum-based chemodrugs: Instrumentation and phantom validation
2013; 40 (3)
Developing an imaging method to directly monitor the spatial distribution of platinum-based (Pt) drugs at the tumor region is of critical importance for early assessment of treatment efficacy and personalized treatment. In this study, the authors investigated the feasibility of imaging platinum (Pt)-based drug distribution using x-ray fluorescence (XRF, a.k.a. characteristic x ray) CT (XFCT).A 5-mm-diameter pencil beam produced by a polychromatic x-ray source equipped with a tungsten anode was used to stimulate emission of XRF photons from Pt drug embedded within a water phantom. The phantom was translated and rotated relative to the stationary pencil beam in a first-generation CT geometry. The x-ray energy spectrum was collected for 18 s at each position using a cadmium telluride detector. The spectra were then used for the K-shell XRF peak isolation and sinogram generation for Pt. The distribution and concentration of Pt were reconstructed with an iterative maximum likelihood expectation maximization algorithm. The capability of XFCT to multiplexed imaging of Pt, gadolinium (Gd), and iodine (I) within a water phantom was also investigated.Measured XRF spectrum showed a sharp peak characteristic of Pt with a narrow full-width at half-maximum (FWHM) (FWHMK?1 = 1.138 keV, FWHMK?2 = 1.052 keV). The distribution of Pt drug in the water phantom was clearly identifiable on the reconstructed XRF images. Our results showed a linear relationship between the XRF intensity of Pt and its concentrations (R(2) = 0.995), suggesting that XFCT is capable of quantitative imaging. A transmission CT image was also obtained to show the potential of the approach for providing attenuation correction and morphological information. Finally, the distribution of Pt, Gd, and I in the water phantom was clearly identifiable in the reconstructed images from XFCT multiplexed imaging.XFCT is a promising modality for monitoring the spatial distribution of Pt drugs. The technique may be useful in tailoring tumor treatment regimen in the future.
View details for DOI 10.1118/1.4789917
View details for Web of Science ID 000316369400003
View details for PubMedID 23464279
First Demonstration of Multiplexed X-Ray Fluorescence Computed Tomography (XFCT) Imaging
IEEE TRANSACTIONS ON MEDICAL IMAGING
2013; 32 (2): 262-267
Simultaneous imaging of multiple probes or biomarkers represents a critical step toward high specificity molecular imaging. In this work, we propose to utilize the element-specific nature of the X-ray fluorescence (XRF) signal for imaging multiple elements simultaneously (multiplexing) using XRF computed tomography (XFCT). A 5-mm-diameter pencil beam produced by a polychromatic X-ray source (150 kV, 20 mA) was used to stimulate emission of XRF photons from 2% (weight/volume) gold (Au), gadolinium (Gd), and barium (Ba) embedded within a water phantom. The phantom was translated and rotated relative to the stationary pencil beam in a first-generation CT geometry. The X-ray energy spectrum was collected for 18 s at each position using a cadmium telluride detector. The spectra were then used to isolate the K shell XRF peak and to generate sinograms for the three elements of interest. The distribution and concentration of the three elements were reconstructed with the iterative maximum likelihood expectation maximization algorithm. The linearity between the XFCT intensity and the concentrations of elements of interest was investigated. We found that measured XRF spectra showed sharp peaks characteristic of Au, Gd, and Ba. The narrow full-width at half-maximum (FWHM) of the peaks strongly supports the potential of XFCT for multiplexed imaging of Au, Gd, and Ba ( FWHM(Au,Kα1) = 0.619 keV, FWHM(Au,Kα2)=1.371 keV , FWHM(Gd,Kα)=1.297 keV, FWHM(Gd,Kβ)=0.974 keV , FWHM(Ba,Kα)=0.852 keV, and FWHM(Ba,Kβ)=0.594 keV ). The distribution of Au, Gd, and Ba in the water phantom was clearly identifiable in the reconstructed XRF images. Our results showed linear relationships between the XRF intensity of each tested element and their concentrations ( R(2)(Au)=0.944 , R(Gd)(2)=0.986, and R(Ba)(2)=0.999), suggesting that XFCT is capable of quantitative imaging. Finally, a transmission CT image was obtained to show the potential of the approach for providing attenuation correction and morphological information. In conclusion, XFCT is a promising modality for multiplexed imaging of high atomic number probes.
View details for DOI 10.1109/TMI.2012.2223709
View details for Web of Science ID 000314367100011
View details for PubMedID 23076031
X-ray excitable luminescent polymer dots doped with an iridium(III) complex
2013; 49 (39): 4319-4321
In this study, cyclometalated iridium(III) complex-doped polymer dots were synthesized and shown to emit luminescence upon X-ray irradiation, potentially serving as a new probe for molecular imaging during X-ray computed tomography.
View details for DOI 10.1039/c2cc37169c
View details for Web of Science ID 000317931500063
View details for PubMedID 23320256
X-ray acoustic computed tomography with pulsed x-ray beam from a medical linear accelerator
2013; 40 (1)
The feasibility of medical imaging using a medical linear accelerator to generate acoustic waves is investigated. This modality, x-ray acoustic computed tomography (XACT), has the potential to enable deeper tissue penetration in tissue than photoacoustic tomography via laser excitation.Short pulsed (?s-range) 10 MV x-ray beams with dose-rate of approximately 30 Gy?min were generated from a medical linear accelerator. The acoustic signals were collected with an ultrasound transducer (500 KHz central frequency) positioned around an object. The transducer, driven by a computer-controlled step motor to scan around the object, detected the resulting acoustic signals in the imaging plane at each scanning position. A pulse preamplifier, with a bandwidth of 20 KHz-2 MHz at -3 dB, and switchable gains of 40 and 60 dB, received the signals from the transducer and delivered the amplified signals to a secondary amplifier. The secondary amplifier had bandwidth of 20 KHz-30 MHz at -3 dB, and a gain range of 10-60 dB. Signals were recorded and averaged 128 times by an oscilloscope. A sampling rate of 100 MHz was used to record 2500 data points at each view angle. One set of data incorporated 200 positions as the receiver moved 360°. The x-ray generated acoustic image was then reconstructed with the filtered back projection algorithm.The x-ray generated acoustic signals were detected from a lead rod embedded in a chicken breast tissue. The authors found that the acoustic signal was proportional to the x-ray dose deposition, with a correlation of 0.998. The two-dimensional XACT images of the lead rod embedded in chicken breast tissue were found to be in good agreement with the shape of the object.The first x-ray acoustic computed tomography image is presented. The new modality may be useful for a number of applications, such as providing the location of a fiducial, or monitoring x-ray dose distribution during radiation therapy. Although much work is needed to improve the image quality of XACT and to explore its performance in other irradiation energies, the benefits of this modality, as highlighted in this work, encourage further study.
View details for DOI 10.1118/1.4771935
View details for Web of Science ID 000313033200003
View details for PubMedID 23298069
Radioluminescence Microscopy: Measuring the Heterogeneous Uptake of Radiotracers in Single Living Cells
2012; 7 (10)
Radiotracers play an important role in interrogating molecular processes both in vitro and in vivo. However, current methods are limited to measuring average radiotracer uptake in large cell populations and, as a result, lack the ability to quantify cell-to-cell variations. Here we apply a new technique, termed radioluminescence microscopy, to visualize radiotracer uptake in single living cells, in a standard fluorescence microscopy environment. In this technique, live cells are cultured sparsely on a thin scintillator plate and incubated with a radiotracer. Light produced following beta decay is measured using a highly sensitive microscope. Radioluminescence microscopy revealed strong heterogeneity in the uptake of [(18)F]fluoro-deoxyglucose (FDG) in single cells, which was found consistent with fluorescence imaging of a glucose analog. We also verified that dynamic uptake of FDG in single cells followed the standard two-tissue compartmental model. Last, we transfected cells with a fusion PET/fluorescence reporter gene and found that uptake of FHBG (a PET radiotracer for transgene expression) coincided with expression of the fluorescent protein. Together, these results indicate that radioluminescence microscopy can visualize radiotracer uptake with single-cell resolution, which may find a use in the precise characterization of radiotracers.
View details for DOI 10.1371/journal.pone.0046285
View details for Web of Science ID 000309454000029
View details for PubMedID 23056276
Intraoperative Imaging of Tumors Using Cerenkov Luminescence Endoscopy: A Feasibility Experimental Study
JOURNAL OF NUCLEAR MEDICINE
2012; 53 (10): 1579-1584
Cerenkov luminescence imaging (CLI) is an emerging new molecular imaging modality that is relatively inexpensive, easy to use, and has high throughput. CLI can image clinically available PET and SPECT probes using optical instrumentation. Cerenkov luminescence endoscopy (CLE) is one of the most intriguing applications that promise potential clinical translation. We developed a prototype customized fiberscopic Cerenkov imaging system to investigate the potential in guiding minimally invasive surgical resection.All experiments were performed in a dark chamber. Cerenkov luminescence from (18)F-FDG samples containing decaying radioactivity was transmitted through an optical fiber bundle and imaged by an intensified charge-coupled device camera. Phantoms filled with (18)F-FDG were used to assess the imaging spatial resolution. Finally, mice bearing subcutaneous C6 glioma cells were injected intravenously with (18)F-FDG to determine the feasibility of in vivo imaging. The tumor tissues were exposed, and CLI was performed on the mouse before and after surgical removal of the tumor using the fiber-based imaging system and compared with a commercial optical imaging system.The sensitivity of this particular setup was approximately 45 kBq (1.21 ?Ci)/300 ?L. The 3 smallest sets of cylindric holes in a commercial SPECT phantom were identifiable via this system, demonstrating that the system has a resolution better than 1.2 mm. Finally, the in vivo tumor imaging study demonstrated the feasibility of using CLI to guide the resection of tumor tissues.This proof-of-concept study explored the feasibility of using fiber-based CLE for the detection of tumor tissue in vivo for guided surgery. With further improvements of the imaging sensitivity and spatial resolution of the current system, CLE may have a significant application in the clinical setting in the near future.
View details for DOI 10.2967/jnumed.111.098541
View details for Web of Science ID 000309432400017
View details for PubMedID 22904353
Investigation of X-ray Fluorescence Computed Tomography (XFCT) and K-Edge Imaging
IEEE TRANSACTIONS ON MEDICAL IMAGING
2012; 31 (8): 1620-1627
This work provides a comprehensive Monte Carlo study of X-ray fluorescence computed tomography (XFCT) and K-edge imaging system, including the system design, the influence of various imaging components, the sensitivity and resolution under various conditions. We modified the widely used EGSnrc/DOSXYZnrc code to simulate XFCT images of two acrylic phantoms loaded with various concentrations of gold nanoparticles and Cisplatin for a number of XFCT geometries. In particular, reconstructed signal as a function of the width of the detector ring, its angular coverage and energy resolution were studied. We found that XFCT imaging sensitivity of the modeled systems consisting of a conventional X-ray tube and a full 2-cm-wide energy-resolving detector ring was 0.061% and 0.042% for gold nanoparticles and Cisplatin, respectively, for a dose of ? 10 cGy. Contrast-to-noise ratio (CNR) of XFCT images of the simulated acrylic phantoms was higher than that of transmission K-edge images for contrast concentrations below 0.4%.
View details for DOI 10.1109/TMI.2012.2201165
View details for Web of Science ID 000307120600010
View details for PubMedID 22692896
Radioluminescent nanophosphors enable multiplexed small-animal imaging
2012; 20 (11): 11598-11604
We demonstrate the ability to image multiple nanoparticle-based contrast agents simultaneously using a nanophosphor platform excited by either radiopharmaceutical or X-ray irradiation. These radioluminescent nanoparticles emit optical light at unique wavelengths depending on their lanthanide dopant, enabling multiplexed imaging. This study demonstrates the separation of two distinct nanophosphor contrast agents in gelatin phantoms with a recovered phosphor separation correlation of -0.98. The ability to distinguish the two nanophosphors and a Cerenkov component is then demonstrated in a small animal phantom. Combined with the high-resolution potential of low-scattering X-ray excitation, this imaging technique may be a promising method to probe molecular processes in living organisms.
View details for Web of Science ID 000304403100002
View details for PubMedID 22714145
Ultrafast and scalable cone-beam CT reconstruction using MapReduce in a cloud computing environment
2011; 38 (12): 6603-6609
Four-dimensional CT (4DCT) and cone beam CT (CBCT) are widely used in radiation therapy for accurate tumor target definition and localization. However, high-resolution and dynamic image reconstruction is computationally demanding because of the large amount of data processed. Efficient use of these imaging techniques in the clinic requires high-performance computing. The purpose of this work is to develop a novel ultrafast, scalable and reliable image reconstruction technique for 4D CBCT?CT using a parallel computing framework called MapReduce. We show the utility of MapReduce for solving large-scale medical physics problems in a cloud computing environment.In this work, we accelerated the Feldcamp-Davis-Kress (FDK) algorithm by porting it to Hadoop, an open-source MapReduce implementation. Gated phases from a 4DCT scans were reconstructed independently. Following the MapReduce formalism, Map functions were used to filter and backproject subsets of projections, and Reduce function to aggregate those partial backprojection into the whole volume. MapReduce automatically parallelized the reconstruction process on a large cluster of computer nodes. As a validation, reconstruction of a digital phantom and an acquired CatPhan 600 phantom was performed on a commercial cloud computing environment using the proposed 4D CBCT?CT reconstruction algorithm.Speedup of reconstruction time is found to be roughly linear with the number of nodes employed. For instance, greater than 10 times speedup was achieved using 200 nodes for all cases, compared to the same code executed on a single machine. Without modifying the code, faster reconstruction is readily achievable by allocating more nodes in the cloud computing environment. Root mean square error between the images obtained using MapReduce and a single-threaded reference implementation was on the order of 10(-7). Our study also proved that cloud computing with MapReduce is fault tolerant: the reconstruction completed successfully with identical results even when half of the nodes were manually terminated in the middle of the process.An ultrafast, reliable and scalable 4D CBCT?CT reconstruction method was developed using the MapReduce framework. Unlike other parallel computing approaches, the parallelization and speedup required little modification of the original reconstruction code. MapReduce provides an efficient and fault tolerant means of solving large-scale computing problems in a cloud computing environment.
View details for DOI 10.1118/1.3660200
View details for Web of Science ID 000298250100028
View details for PubMedID 22149842
Monte Carlo simulation of photon migration in a cloud computing environment with MapReduce
JOURNAL OF BIOMEDICAL OPTICS
2011; 16 (12)
Monte Carlo simulation is considered the most reliable method for modeling photon migration in heterogeneous media. However, its widespread use is hindered by the high computational cost. The purpose of this work is to report on our implementation of a simple MapReduce method for performing fault-tolerant Monte Carlo computations in a massively-parallel cloud computing environment. We ported the MC321 Monte Carlo package to Hadoop, an open-source MapReduce framework. In this implementation, Map tasks compute photon histories in parallel while a Reduce task scores photon absorption. The distributed implementation was evaluated on a commercial compute cloud. The simulation time was found to be linearly dependent on the number of photons and inversely proportional to the number of nodes. For a cluster size of 240 nodes, the simulation of 100 billion photon histories took 22 min, a 1258 × speed-up compared to the single-threaded Monte Carlo program. The overall computational throughput was 85,178 photon histories per node per second, with a latency of 100 s. The distributed simulation produced the same output as the original implementation and was resilient to hardware failure: the correctness of the simulation was unaffected by the shutdown of 50% of the nodes.
View details for DOI 10.1117/1.3656964
View details for Web of Science ID 000299490300011
View details for PubMedID 22191916
Fully 3D list-mode time-of-flight PET image reconstruction on GPUs using CUDA
2011; 38 (12): 6775-6786
List-mode processing is an efficient way of dealing with the sparse nature of positron emission tomography (PET) data sets and is the processing method of choice for time-of-flight (ToF) PET image reconstruction. However, the massive amount of computation involved in forward projection and backprojection limits the application of list-mode reconstruction in practice, and makes it challenging to incorporate accurate system modeling.The authors present a novel formulation for computing line projection operations on graphics processing units (GPUs) using the compute unified device architecture (CUDA) framework, and apply the formulation to list-mode ordered-subsets expectation maximization (OSEM) image reconstruction. Our method overcomes well-known GPU challenges such as divergence of compute threads, limited bandwidth of global memory, and limited size of shared memory, while exploiting GPU capabilities such as fast access to shared memory and efficient linear interpolation of texture memory. Execution time comparison and image quality analysis of the GPU-CUDA method and the central processing unit (CPU) method are performed on several data sets acquired on a preclinical scanner and a clinical ToF scanner.When applied to line projection operations for non-ToF list-mode PET, this new GPU-CUDA method is >200 times faster than a single-threaded reference CPU implementation. For ToF reconstruction, we exploit a ToF-specific optimization to improve the efficiency of our parallel processing method, resulting in GPU reconstruction >300 times faster than the CPU counterpart. For a typical whole-body scan with 75?×?75?×?26 image matrix, 40.7 million LORs, 33 subsets, and 3 iterations, the overall processing time is 7.7 s for GPU and 42 min for a single-threaded CPU. Image quality and accuracy are preserved for multiple imaging configurations and reconstruction parameters, with normalized root mean squared (RMS) deviation less than 1% between CPU and GPU-generated images for all cases.A list-mode ToF OSEM library was developed on the GPU-CUDA platform. Our studies show that the GPU reformulation is considerably faster than a single-threaded reference CPU method especially for ToF processing, while producing virtually identical images. This new method can be easily adapted to enable more advanced algorithms for high resolution PET reconstruction based on additional information such as depth of interaction (DoI), photon energy, and point spread functions (PSFs).
View details for DOI 10.1118/1.3661998
View details for Web of Science ID 000298250100045
View details for PubMedID 22149859
Toward IMRT 2D dose modeling using artificial neural networks: A feasibility study
2011; 38 (10): 5807-5817
To investigate the feasibility of artificial neural networks (ANN) to reconstruct dose maps for intensity modulated radiation treatment (IMRT) fields compared with those of the treatment planning system (TPS).An artificial feed forward neural network and the back-propagation learning algorithm have been used to replicate dose calculations of IMRT fields obtained from PINNACLE(3) v9.0. The ANN was trained with fluence and dose maps of IMRT fields for 6 MV x-rays, which were obtained from the amorphous silicon (a-Si) electronic portal imaging device of Novalis TX. Those fluence distributions were imported to the TPS and the dose maps were calculated on the horizontal midpoint plane of a water equivalent homogeneous cylindrical virtual phantom. Each exported 2D dose distribution from the TPS was classified into two clusters of high and low dose regions, respectively, based on the K-means algorithm and the Euclidian metric in the fluence-dose domain. The data of each cluster were divided into two sets for the training and validation phase of the ANN, respectively. After the completion of the ANN training phase, 2D dose maps were reconstructed by the ANN and isodose distributions were created. The dose maps reconstructed by ANN were evaluated and compared with the TPS, where the mean absolute deviation of the dose and the ?-index were used.A good agreement between the doses calculated from the TPS and the trained ANN was achieved. In particular, an average relative dosimetric difference of 4.6% and an average ?-index passing rate of 93% were obtained for low dose regions, and a dosimetric difference of 2.3% and an average ?-index passing rate of 97% for high dose region.An artificial neural network has been developed to convert fluence maps to corresponding dose maps. The feasibility and potential of an artificial neural network to replicate complex convolution kernels in the TPS for IMRT dose calculations have been demonstrated.
View details for DOI 10.1118/1.3639998
View details for Web of Science ID 000295617400052
View details for PubMedID 21992395
Toward real-time Monte Carlo simulation using a commercial cloud computing infrastructure
PHYSICS IN MEDICINE AND BIOLOGY
2011; 56 (17): N175-N181
Monte Carlo (MC) methods are the gold standard for modeling photon and electron transport in a heterogeneous medium; however, their computational cost prohibits their routine use in the clinic. Cloud computing, wherein computing resources are allocated on-demand from a third party, is a new approach for high performance computing and is implemented to perform ultra-fast MC calculation in radiation therapy. We deployed the EGS5 MC package in a commercial cloud environment. Launched from a single local computer with Internet access, a Python script allocates a remote virtual cluster. A handshaking protocol designates master and worker nodes. The EGS5 binaries and the simulation data are initially loaded onto the master node. The simulation is then distributed among independent worker nodes via the message passing interface, and the results aggregated on the local computer for display and data analysis. The described approach is evaluated for pencil beams and broad beams of high-energy electrons and photons. The output of cloud-based MC simulation is identical to that produced by single-threaded implementation. For 1 million electrons, a simulation that takes 2.58 h on a local computer can be executed in 3.3 min on the cloud with 100 nodes, a 47× speed-up. Simulation time scales inversely with the number of parallel nodes. The parallelization overhead is also negligible for large simulations. Cloud computing represents one of the most important recent advances in supercomputing technology and provides a promising platform for substantially improved MC simulation. In addition to the significant speed up, cloud computing builds a layer of abstraction for high performance parallel computing, which may change the way dose calculations are performed and radiation treatment plans are completed.
View details for DOI 10.1088/0031-9155/56/17/N02
View details for Web of Science ID 000294786400003
View details for PubMedID 21841211
Online detector response calculations for high-resolution PET image reconstruction
PHYSICS IN MEDICINE AND BIOLOGY
2011; 56 (13): 4023-4040
Positron emission tomography systems are best described by a linear shift-varying model. However, image reconstruction often assumes simplified shift-invariant models to the detriment of image quality and quantitative accuracy. We investigated a shift-varying model of the geometrical system response based on an analytical formulation. The model was incorporated within a list-mode, fully 3D iterative reconstruction process in which the system response coefficients are calculated online on a graphics processing unit (GPU). The implementation requires less than 512 Mb of GPU memory and can process two million events per minute (forward and backprojection). For small detector volume elements, the analytical model compared well to reference calculations. Images reconstructed with the shift-varying model achieved higher quality and quantitative accuracy than those that used a simpler shift-invariant model. For an 8 mm sphere in a warm background, the contrast recovery was 95.8% for the shift-varying model versus 85.9% for the shift-invariant model. In addition, the spatial resolution was more uniform across the field-of-view: for an array of 1.75 mm hot spheres in air, the variation in reconstructed sphere size was 0.5 mm RMS for the shift-invariant model, compared to 0.07 mm RMS for the shift-varying model.
View details for DOI 10.1088/0031-9155/56/13/018
View details for Web of Science ID 000291866800020
View details for PubMedID 21677367
- Synthesis and Radioluminescence of PEGylated Eu3+-doped Nanophosphors as Bioimaging Probes ADVANCED MATERIALS 2011; 23 (24): H195-H199
Limited-angle x-ray luminescence tomography: methodology and feasibility study
PHYSICS IN MEDICINE AND BIOLOGY
2011; 56 (12): 3487-3502
X-ray luminescence tomography (XLT) has recently been proposed as a new imaging modality for biological imaging applications. This modality utilizes phosphor nanoparticles which luminesce near-infrared light when excited by x-ray photons. The advantages of this modality are that it uniquely combines the high sensitivity of radioluminescent nanoparticles and the high spatial localization of collimated x-ray beams. Currently, XLT has been demonstrated using x-ray spatial encoding to resolve the imaging volume. However, there are applications where the x-ray excitation may be limited by geometry, where increased temporal resolution is desired, or where a lower dose is mandatory. This paper extends the utility of XLT to meet these requirements by incorporating a photon propagation model into the reconstruction algorithm in an x-ray limited-angle (LA) geometry. This enables such applications as image-guided surgery, where the ability to resolve lesions at depths of several centimeters can be the key to successful resection. The hybrid x-ray/diffuse optical model is first formulated and then demonstrated in a breast-sized phantom, simulating a breast lumpectomy geometry. Both numerical and experimental phantoms are tested, with lesion-simulating objects of various sizes and depths. Results show localization accuracy with median error of 2.2 mm, or 4% of object depth, for small 2-14 mm diameter lesions positioned from 1 to 4.5 cm in depth. This compares favorably with fluorescence optical imaging, which is not able to resolve such small objects at this depth. The recovered lesion size has lower size bias in the x-ray excitation direction than the optical direction, which is expected due to the increased optical scatter. However, the technique is shown to be quite invariant in recovered size with respect to depth, as the standard deviation is less than 2.5 mm. Sensitivity is a function of dose; radiological doses are found to provide sufficient recovery for µg ml(-1) concentrations, while therapy dosages provide recovery for ng ml(-1) concentrations. Experimental phantom results agree closely with the numerical results, with positional errors recovered within 8.6% of the effective depth for a 5 mm object, and within 5.2% of the depth for a 10 mm object. Object-size median error is within 2.3% and 2% for the 5 and 10 mm objects, respectively. For shallow-to-medium depth applications where optical and radio-emission imaging modalities are not ideal, such as in intra-operative procedures, LAXLT may be a useful tool to detect molecular signatures of disease.
View details for DOI 10.1088/0031-9155/56/12/003
View details for Web of Science ID 000291095700004
View details for PubMedID 21606553
GPU computing in medical physics: A review
2011; 38 (5): 2685-2697
The graphics processing unit (GPU) has emerged as a competitive platform for computing massively parallel problems. Many computing applications in medical physics can be formulated as data-parallel tasks that exploit the capabilities of the GPU for reducing processing times. The authors review the basic principles of GPU computing as well as the main performance optimization techniques, and survey existing applications in three areas of medical physics, namely image reconstruction, dose calculation and treatment plan optimization, and image processing.
View details for DOI 10.1118/1.3578605
View details for Web of Science ID 000290625700044
View details for PubMedID 21776805
Convex Optimization of Coincidence Time Resolution for a High-Resolution PET System
IEEE TRANSACTIONS ON MEDICAL IMAGING
2011; 30 (2): 391-400
We are developing a dual panel breast-dedicated positron emission tomography (PET) system using LSO scintillators coupled to position sensitive avalanche photodiodes (PSAPD). The charge output is amplified and read using NOVA RENA-3 ASICs. This paper shows that the coincidence timing resolution of the RENA-3 ASIC can be improved using certain list-mode calibrations. We treat the calibration problem as a convex optimization problem and use the RENA-3's analog-based timing system to correct the measured data for time dispersion effects from correlated noise, PSAPD signal delays and varying signal amplitudes. The direct solution to the optimization problem involves a matrix inversion that grows order (n(3)) with the number of parameters. An iterative method using single-coordinate descent to approximate the inversion grows order (n). The inversion does not need to run to convergence, since any gains at high iteration number will be low compared to noise amplification. The system calibration method is demonstrated with measured pulser data as well as with two LSO-PSAPD detectors in electronic coincidence. After applying the algorithm, the 511 keV photopeak paired coincidence time resolution from the LSO-PSAPD detectors under study improved by 57%, from the raw value of 16.3 ±0.07 ns full-width at half-maximum (FWHM) to 6.92 ±0.02 ns FWHM ( 11.52 ±0.05 ns to 4.89 ±0.02 ns for unpaired photons).
View details for Web of Science ID 000286931000019
View details for PubMedID 20876008
- Fast List-Mode Reconstruction for Time-of-Flight PET Using Graphics Hardware IEEE TRANSACTIONS ON NUCLEAR SCIENCE 2011; 58 (1): 105-109
- Facile Synthesis of Amine-Functionalized Eu3+-Doped La(OH)(3) Nanophosphors for Bioimaging NANOSCALE RESEARCH LETTERS 2011; 6
X-Ray Luminescence Computed Tomography via Selective Excitation: A Feasibility Study
IEEE TRANSACTIONS ON MEDICAL IMAGING
2010; 29 (12): 1992-1999
X-ray luminescence computed tomography (XLCT) is proposed as a new molecular imaging modality based on the selective excitation and optical detection of X-ray-excitable phosphor nanoparticles. These nano-sized particles can be fabricated to emit near-infrared (NIR) light when excited with X-rays, and, because because both X-rays and NIR photons propagate long distances in tissue, they are particularly well suited for in vivo biomedical imaging. In XLCT, tomographic images are generated by irradiating the subject using a sequence of programmed X-ray beams, while sensitive photo-detectors measure the light diffusing out of the subject. By restricting the X-ray excitation to a single, narrow beam of radiation, the origin of the optical photons can be inferred regardless of where these photons were detected, and how many times they scattered in tissue. This study presents computer simulations exploring the feasibility of imaging small objects with XLCT, such as research animals. The accumulation of 50 nm phosphor nanoparticles in a 2-mm-diameter target can be detected and quantified with subpicomolar sensitivity using less than 1 cGy of radiation dose. Provided sufficient signal-to-noise ratio, the spatial resolution of the system can be made as high as needed by narrowing the beam aperture. In particular, 1 mm spatial resolution was achieved for a 1-mm-wide X-ray beam. By including an X-ray detector in the system, anatomical imaging is performed simultaneously with molecular imaging via standard X-ray computed tomography (CT). The molecular and anatomical images are spatially and temporally co-registered, and, if a single-pixel X-ray detector is used, they have matching spatial resolution.
View details for DOI 10.1109/TMI.2010.2055883
View details for Web of Science ID 000284848700004
View details for PubMedID 20615807
Tomographic molecular imaging of x-ray-excitable nanoparticles
2010; 35 (20): 3345-3347
X-ray luminescence computed tomography (XLCT) is proposed as a new dual molecular/anatomical imaging modality. XLCT is based on the selective excitation and optical detection of x-ray-excitable nanoparticles. As a proof of concept, we built a prototype XLCT system and imaged near-IR-emitting Gd(2)O(2)S:Eu phosphors in various phantoms. Imaging in an optically diffusive medium shows that imaging performance is not affected by optical scatter; furthermore, the linear response of the reconstructed images suggests that XLCT is capable of quantitative imaging.
View details for Web of Science ID 000283048100013
View details for PubMedID 20967061
Effects of multiple-interaction photon events in a high-resolution PET system that uses 3-D positioning detectors
2010; 37 (10): 5494-5508
The authors' laboratory is developing a dual-panel, breast-dedicated PET system. The detector panels are built from dual-LSO-position-sensitive avalanche photodiode (PSAPD) modules-units holding two 8 x 8 arrays of 1 mm3 LSO crystals, where each array is coupled to a PSAPD. When stacked to form an imaging volume, these modules are capable of recording the 3-D coordinates of individual interactions of a multiple-interaction photon event (MIPE). The small size of the scintillation crystal elements used increases the likelihood of photon scattering between crystal arrays. In this article, the authors investigate how MIPEs impact the system photon sensitivity, the data acquisition scheme, and the quality and quantitative accuracy of reconstructed PET images.A Monte Carlo simulated PET scan using the dual-panel system was performed on a uniformly radioactive phantom for the photon sensitivity study. To establish the impact of MIPEs on a proposed PSAPD multiplexing scheme, experimental data were collected from a dual-LSO-PSAPD module edge-irradiated with a 22Na point source, the data were compared against simulation data based on an identical setup. To assess the impact of MIPEs on the dual-panel PET images, a simulated PET of a phantom comprising a matrix of hot spherical radiation sources of varying diameters immersed in a warm background was performed. The list-mode output data were used for image reconstruction, where various methods were used for estimating the location of the first photon interaction in MIPEs for more accurate line of response positioning. The contrast recovery coefficient (CRC), contrast to noise ratio (CNR), and the full width at half maximum spatial resolution of the spheres in the reconstructed images were used as figures of merit to facilitate comparison.Compared to image reconstruction employing only events with interactions confined to one LSO array, a potential single photon sensitivity gain of > 46.9% (> 115.7% for coincidence) was noted for a uniform phantom when MIPEs with summed-energy falling within a +/- 12% window around the photopeak were also included. Both experimental and simulation data demonstrate that < 0.4% of the events whose summed-energy deposition falling within that energy window interacted with both crystal arrays within the same dual-LSO-PSAPD module. This result establishes the feasibility of a proposed multiplexed readout of analog output signals of the two PSAPDs within each module. Using MIPEs with summed-energy deposition within the 511 keV +/- 12% photopeak window and a new method for estimating the location of the first photon interaction in MIPEs, the corresponding reconstructed image exhibited a peak CNR of 7.23 for the 8 mm diameter phantom spheres versus a CNR of 6.69 from images based solely on single LSO array interaction events. The improved system photon sensitivity could be exploited to reduce the scan time by up to approximately 10%, while still maintaining image quality comparable to that achieved if MIPEs were excluded.MIPE distribution in the detectors allows the proposed photodetector multiplexing arrangement without significant information loss. Furthermore, acquiring MIPEs can enhance system photon sensitivity and improve PET image CNR and CRC. The system under development can therefore competently acquire and analyze MIPEs and produce high-resolution PET images.
View details for DOI 10.1118/1.3483262
View details for Web of Science ID 000283483700038
View details for PubMedID 21089785
Hybrid x-ray/optical luminescence imaging: Characterization of experimental conditions
2010; 37 (8): 4011-4018
The feasibility of x-ray luminescence imaging is investigated using a dual-modality imaging system that merges x-ray and optical imaging. This modality utilizes x-ray activated nanophosphors that luminesce when excited by ionizing photons. By doping phosphors with lanthanides, which emit light in the visible and near infrared range, the luminescence is suitable for biological applications. This study examines practical aspects of this new modality including phosphor concentration, light emission linearity, detector damage, and spectral emission characteristics. Finally, the contrast produced by these phosphors is compared to that of x-ray fluoroscopy.Gadolinium and lanthanum oxysulfide phosphors doped with terbium (green emission) or europium (red emission) were studied. The light emission was imaged in a clinical x-ray scanner with a cooled CCD camera and a spectrophotometer; dose measurements were determined with a calibrated dosimeter. Using these properties, in addition to luminescence efficiency values found in the literature for a similar phosphor, minimum concentration calculations are performed. Finally, a 2.5 cm agar phantom with a 1 cm diameter cylindrical phosphor-filled inclusion (diluted at 10 mg/ml) is imaged to compare x-ray luminescence contrast with x-ray fluoroscopic contrast at a superficial location.Dose to the CCD camera in the chosen imaging geometry was measured at less than 0.02 cGy/s. Emitted light was found to be linear with dose (R(2)= 1) and concentration (R(2)= 1). Emission peaks for clinical x-ray energies are less than 3 nm full width at half maximum, as expected from lanthanide dopants. The minimum practical concentration necessary to detect luminescent phosphors is dependent on dose; it is estimated that subpicomolar concentrations are detectable at the surface of the tissue with typical mammographic doses, with the minimum detectable concentration increasing with depth and decreasing with dose. In a reflection geometry, x-ray luminescence had nearly a 430-fold greater contrast to background than x-ray fluoroscopy.X-ray luminescence has the potential to be a promising new modality for enabling molecular imaging within x-ray scanners. Although much work needs to be done to ensure biocompatibility of x-ray exciting phosphors, the benefits of this modality, highlighted in this work, encourage further study.
View details for DOI 10.1118/1.3457332
View details for Web of Science ID 000281112900011
View details for PubMedID 20879562
Bayesian reconstruction of photon interaction sequences for high-resolution PET detectors
PHYSICS IN MEDICINE AND BIOLOGY
2009; 54 (17): 5073-5094
Realizing the full potential of high-resolution positron emission tomography (PET) systems involves accurately positioning events in which the annihilation photon deposits all its energy across multiple detector elements. Reconstructing the complete sequence of interactions of each photon provides a reliable way to select the earliest interaction because it ensures that all the interactions are consistent with one another. Bayesian estimation forms a natural framework to maximize the consistency of the sequence with the measurements while taking into account the physics of gamma-ray transport. An inherently statistical method, it accounts for the uncertainty in the measured energy and position of each interaction. An algorithm based on maximum a posteriori (MAP) was evaluated for computer simulations. For a high-resolution PET system based on cadmium zinc telluride detectors, 93.8% of the recorded coincidences involved at least one photon multiple-interactions event (PMIE). The MAP estimate of the first interaction was accurate for 85.2% of the single photons. This represents a two-fold reduction in the number of mispositioned events compared to minimum pair distance, a simpler yet efficient positioning method. The point-spread function of the system presented lower tails and higher peak value when MAP was used. This translated into improved image quality, which we quantified by studying contrast and spatial resolution gains.
View details for DOI 10.1088/0031-9155/54/17/001
View details for Web of Science ID 000269074500002
View details for PubMedID 19652293
Fast, Accurate and Shift-Varying Line Projections for Iterative Reconstruction Using the GPU
IEEE TRANSACTIONS ON MEDICAL IMAGING
2009; 28 (3): 435-445
List-mode processing provides an efficient way to deal with sparse projections in iterative image reconstruction for emission tomography. An issue often reported is the tremendous amount of computation required by such algorithm. Each recorded event requires several back- and forward line projections. We investigated the use of the programmable graphics processing unit (GPU) to accelerate the line-projection operations and implement fully-3D list-mode ordered-subsets expectation-maximization for positron emission tomography (PET). We designed a reconstruction approach that incorporates resolution kernels, which model the spatially-varying physical processes associated with photon emission, transport and detection. Our development is particularly suitable for applications where the projection data is sparse, such as high-resolution, dynamic, and time-of-flight PET reconstruction. The GPU approach runs more than 50 times faster than an equivalent CPU implementation while image quality and accuracy are virtually identical. This paper describes in details how the GPU can be used to accelerate the line projection operations, even when the lines-of-response have arbitrary endpoint locations and shift-varying resolution kernels are used. A quantitative evaluation is included to validate the correctness of this new approach.
View details for DOI 10.1109/TMI.2008.2006518
View details for Web of Science ID 000263920500012
View details for PubMedID 19244015