Maximum Likelihood Expectation Maximization Research Articles

Evaluation of lesion contrast in the walk-through long axial FOV PET scanner simulated with XCAT anthropomorphic phantoms

BackgroundThis study evaluates the lesion contrast in a cost-effective long axial field of view (FOV) PET scanner, called the walk-through PET (WT-PET). The scanner consists of two flat detector panels covering the entire torso and head, scanning patients in an upright position for increased throughput. High-resolution, depth-of-interaction capable, monolithic detector technology is used to provide good spatial resolution and enable detection of smaller lesions.MethodsMonte Carlo GATE simulations are used in conjunction with XCAT anthropomorphic phantoms to evaluate lesion contrast in lung, liver and breast for various lesion diameters (10, 7 and 5 mm), activity concentration ratios (8:1, 4:1 and 2:1) and patient BMIs (18–37). Images were reconstructed iteratively with listmode maximum likelihood expectation maximization, and contrast recovery coefficients (CRCs) were obtained for the reconstructed lesions.ResultsResults shows notable variations in contrast recovery coefficients (CRC) across different lesion sizes and organ locations within the XCAT phantoms. Specifically, our findings reveal that 10 mm lesions consistently exhibit higher CRC compared to 7 mm and 5 mm lesions, with increases of approximately 54% and 330%, respectively, across all investigated organs. Moreover, high contrast recovery is observed in most liver lesions regardless of diameter or activity ratio (average CRC = 42%), as well as in the 10 mm lesions in the lung. Notably, for the 10 mm lesions, the liver demonstrates 42% and 62% higher CRC compared to the lung and breast, respectively. This trend remains consistent across lesion sizes, with the liver consistently exhibiting higher CRC values compared to the lung and breast: 7 mm lesions show an increase of 96% and 41%, while 5 mm lesions exhibit approximately 294% and 302% higher CRC compared to the lung and breast, respectively.ConclusionA comparison with a conventional pixelated LSO long axial FOV PET shows similar performance, achieved at a reduced cost for the WT-PET due to a reduction in required number of detectors.

Anatomy-guided multi-resolution image reconstruction in PET

Objective. In this paper, we propose positron emission tomography image reconstruction using a multi-resolution triangular mesh. The mesh can be adapted based on patient specific anatomical information that can be in the form of a computed tomography or magnetic resonance imaging image in the hybrid imaging systems. The triangular mesh can be adapted to high resolution in localized anatomical regions of interest (ROI) and made coarser in other regions, leading to an imaging model with high resolution in the ROI with clearly reduced number of degrees of freedom compared to a conventional uniformly dense imaging model. Approach. We compare maximum likelihood expectation maximization reconstructions with the multi-resolution model to reconstructions using a uniformly dense mesh, a sparse mesh and regular rectangular pixel mesh. Two simulated cases are used in the comparison, with the first one using the NEMA image quality phantom and the second the XCAT human phantom. Main results. When compared to the results with the uniform imaging models, the locally refined multi-resolution mesh retains the accuracy of the dense mesh reconstruction in the ROI while being faster to compute than the reconstructions with the uniformly dense mesh. The locally dense multi-resolution model leads also to more accurate reconstruction than the pixel-based mesh or the sparse triangular mesh. Significance. The findings suggest that triangular multi-resolution mesh, which can be made patient and application specific, is a potential alternative for pixel-based reconstruction.

Stochastic model for enhanced PET image reconstruction

In this paper, we present a detailed stochastic model of crystal-to-crystal response for PET scanners, which is crucial for addressing their physical constraints. We present both a closed-form solution and several accurate approximations to manage the complexity of the exact mathematical expressions. Through experimental and analytical validation, we demonstrate that the disparities between the best approximations and actual crystal-to-crystal response are negligible. These responses are utilized to develop a compensation model termed a white image. This model is expressed in a single closed-form equation, simplifying its integration into established reconstruction methods. We modify the maximum likelihood expectation maximization (MLEM) algorithm and integrate our white image model into it. Unlike the conventional approach that relies on the system matrix, the modified MLEM algorithm operates on ray-driven projections and back-projections. The system matrix, being large and requiring complex methods and simulations for evaluation, contrasts with the white image, which is smaller and enables efficient computation. Our compensation model provides comprehensive information about the PET system. We validate our approach with both synthetic and real data. For real-world data acquisition, we utilize the Raytest ClearPET camera for small animals and the NEMA NU 4-2008 phantom. The proposed modification outperforms competing uncompensated reconstruction methods, delivering superior Peak Signal-to-Noise Ratio (PSNR) and reduced execution time. This research deepens our understanding of real PET measurement systems, providing improved input for subsequent reconstruction algorithms and ultimately producing high-quality images.

Split Bregman quantum noise removal algorithm for 3D reconstruction of neutron computed tomography image

The low intensity of the neutron source for neutron computed tomography (CT) results in a long acquisition time for a single projection, which causes the neutron projection data to contain a large amount of quantum noise. Quantum noise will degrade the quality of neutron CT reconstruction images. Therefore, an efficient quantum noise removal algorithm must be used in CT reconstruction. In this paper, an efficient quantum noise removal algorithm for neutron CT 3D image reconstruction is proposed by analysing classical image processing algorithms and quantum image processing algorithms, which employs the maximum likelihood expectation maximization to reconstruct the image and split Bregman algorithm to solve for the total variation (MLEM-SBTV). Experimental results show that MLEM-SBTV performs well in removing quantum noise and reconstructing the detailed structure of images.

Assessment of the axial resolution of a compact gamma camera with coded aperture collimator

PurposeHandheld gamma cameras with coded aperture collimators are under investigation for intraoperative imaging in nuclear medicine. Coded apertures are a promising collimation technique for applications such as lymph node localization due to their high sensitivity and the possibility of 3D imaging. We evaluated the axial resolution and computational performance of two reconstruction methods.MethodsAn experimental gamma camera was set up consisting of the pixelated semiconductor detector Timepix3 and MURA mask of rank 31 with round holes of 0.08 mm in diameter in a 0.11 mm thick Tungsten sheet. A set of measurements was taken where a point-like gamma source was placed centrally at 21 different positions within the range of 12–100 mm. For each source position, the detector image was reconstructed in 0.5 mm steps around the true source position, resulting in an image stack. The axial resolution was assessed by the full width at half maximum (FWHM) of the contrast-to-noise ratio (CNR) profile along the z-axis of the stack. Two reconstruction methods were compared: MURA Decoding and a 3D maximum likelihood expectation maximization algorithm (3D-MLEM).ResultsWhile taking 4400 times longer in computation, 3D-MLEM yielded a smaller axial FWHM and a higher CNR. The axial resolution degraded from 5.3 mm and 1.8 mm at 12 mm to 42.2 mm and 13.5 mm at 100 mm for MURA Decoding and 3D-MLEM respectively.ConclusionOur results show that the coded aperture enables the depth estimation of single point-like sources in the near field. Here, 3D-MLEM offered a better axial resolution but was computationally much slower than MURA Decoding, whose reconstruction time is compatible with real-time imaging.

Design and performance evaluation of a compact thermal and fast neutron spectrometer

Measurement of the space neutron radiation environment has put forward new requirements for the miniaturization and lightweight design of neutron spectrometers. In this paper, a compact broad-spectrum neutron spectrometer was designed, in which the fast neutron detector was based on the EJ-254 scintillator-coupled with SiPMs array, and the thermal and epithermal neutron detectors were based on the NaI(Tl + Li) (NaIL) scintillator coupled with SiPMs array encapsulated in Al and Cd shells, respectively. The overall weight of the neutron spectrometer was only 2.4 kg, and its dimensions were 154 mm × 123 mm × 115 mm. The fast neutron energy spectrum was obtained based on the capture-gated method and the Maximum Likelihood Expectation Maximization (MLEM) unfolding algorithm; the counts of thermal and epithermal neutrons were obtained based on the optimized frequency-domain Pulse Shape Discrimination (PSD) method and cadmium difference method. The proposed neutron spectrometer with all applied methods was tested as a whole system with a252Cf and a241Am–Be source. The spectrum result was compared to an ISO standard source. The accuracy of fast part of 0.5–10 MeV was 5.9% in definition of Relative Root Mean Squared Error (RRMSE). The thermal and epithermal neutron counts were far higher than standard spectrum, and we proved it being effected by laboratory introduced neutron scattering through a simulation, thus the thermal and epithermal flux measurements was believed to be decent. The Figure of Merit (FoM) of the NaIL's PSD neutron gamma discrimination were 2.57 and 2.90 respectively.

Setup and characterisation according to NEMA NU 4 of the phenoPET scanner, a PET system dedicated for plant sciences

Objective. The phenoPET system is a plant dedicated positron emission tomography (PET) scanner consisting of fully digital photo multipliers with lutetium–yttrium oxyorthosilicate crystals and located inside a custom climate chamber. Here, we present the setup of phenoPET, its data processing and image reconstruction together with its performance. Approach. The performance characterization follows the national electrical manufacturers association (NEMA) standard for small animal PET systems with a number of adoptions due to the vertical oriented bore of a PET for plant sciences. In addition temperature stability and spatial resolution with a hot rod phantom are addressed. Main results. The spatial resolution for a 22Na point source at a radial distance of 5 mm to the center of the field-of-view (FOV) is 1.45 mm, 0.82 mm and 1.88 mm with filtered back projection in radial, tangential and axial direction, respectively. A hot rod phantom with 18F gives a spatial resolution of up to 1.6 mm. The peak noise-equivalent count rates are 550 kcps @ 35.08 MBq, 308 kcps @ 33 MBq and 45 kcps @ 40.60 MBq for the mouse, rat and monkey size scatter phantoms, respectively. The scatter fractions for these phantoms are 12.63%, 22.64% and 55.90%. We observe a peak sensitivity of up to 3.6% and a total sensitivity of up to S A,tot = 2.17%. For the NEMA image quality phantom we observe a uniformity of %STD = 4.22% with ordinary Poisson maximum likelihood expectation-maximization with 52 iterations. Here, recovery coefficients of 0.12, 0.64, 0.89, 0.93 and 0.91 for 1 mm, 2 mm, 3 mm, 4 mm and 5 mm rods are obtained and spill-over ratios of 0.08 and 0.14 for the water-filled and air-filled inserts, respectively. Significance. The phenoPET and its laboratory are now in routine operation for the administration of [11C]CO2 and non-invasive measurement of transport and allocation of 11C-labelled photoassimilates in plants.

Quantum annealing-based computed tomography using variational approach for a real-number image reconstruction

Objective. Despite recent advancements in quantum computing, the limited number of available qubits has hindered progress in CT reconstruction. This study investigates the feasibility of utilizing quantum annealing-based computed tomography (QACT) with current quantum bit levels. Approach. The QACT algorithm aims to precisely solve quadratic unconstrained binary optimization problems. Furthermore, a novel approach is proposed to reconstruct images by approximating real numbers using the variational method. This approach allows for accurate CT image reconstruction using a small number of qubits. The study examines the impact of projection data quantity and noise on various image sizes ranging from 4 × 4 to 24 × 24 pixels. The reconstructed results are compared against conventional reconstruction algorithms, namely maximum likelihood expectation maximization (MLEM) and filtered back projection (FBP). Main result. By employing the variational approach and utilizing two qubits for each pixel of the image, accurate reconstruction was achieved with an adequate number of projections. Under conditions of abundant projections and lower noise levels, the image quality in QACT algorithm outperformed that of MLEM and FBP algorithms. However, in situations with limited projection data and in the presence of noise, the image quality in QACT was inferior to that in MLEM. Significance. This study developed the QACT reconstruction algorithm using the variational approach for real-number reconstruction. Remarkably, only 2 qubits were required for each pixel representation, demonstrating their sufficiency for accurate reconstruction.

Exploring the Influence of Initial Estimates on Iterative Maximum Likelihood Expectation–maximization in Tomographic Reconstruction

Purpose: To explore the influence of initial guess or estimate (uniform as “ones” and “zeros” vs. filtered back projection [FBP] image) as an input image for maximum likelihood expectation–maximization (MLEM) tomographic reconstruction algorithm and provide the curves of error or convergence for each of these three initial estimates. Methods: Two phantoms, created as digital images, were utilized: one was a simple noiseless object and the other was a more complicated, noise-degraded object of the section of lower thorax in a matrix of 256 × 256 pixels. Both underwent radon transform or forward projection process and the corresponding sinograms were generated. For filtering during tomographic image reconstruction, ramp and Butterworth filters, as high-pass and low-pass ones, were applied to images. The second phantom (lower thorax) was radon-transformed and the resulting sinogram was degraded by noise. As initial guess or estimate images, in addition to FBP tomographic image, two uniform images, one with all pixels having a value of 1 (“ones”) and the other with all having zero (“zeros”), were created. The three initial estimates (FBP, ones, and zeros) were reconstructed with iterative MLEM tomographic reconstruction (with 1, 2, 4, 8, 16, 32, and 64 iterations). The difference between the object and the updated slice was calculated at the end of each iteration (as error matrix), and the mean squared error (MSE) was computed and plotted separately or in conjunction with the MSE curves of other initial estimates. All computations were implemented in MATLAB software. Results: The results of ones and zeros seemed strikingly similar. The curves of uniform ones and uniform zeros were so close to each other that overlap near-perfectly. However, in the FBP slice as an initial estimate, the resulting tomographic slice was similar with a much higher extent to the object even after 1 or 2 iterations. The pattern of convergence for all three curves was roughly similar. The normalized MSE decreased sharply up to 5 iterations and then, after 10 iterations, the curves reached a plateau until 32 iterations. For the phantom of the lower thorax section with its noise-degraded sinogram, similar to the pattern observed for simple disk-shaped phantom, the curves (normalized MSE) fell sharply up to 10 iterations and then rapidly converged thereafter until 64 iterations. Conclusion: Similar results are observed when choosing different initial guesses or estimates (uniform vs. FBP) as the starting point, based on the error calculation using MSE. The algorithm converges almost similarly for all initial estimates. Therefore, selecting a uniform initial guess image can be an appropriate choice and may be preferred over an FBP image. Reducing the processing time can be a valid reason for this choice.

Positronium imaging in J-PET with an iterative activity reconstruction and a multistage fitting algorithm

Positronium imaging is a new technique complementary to positron emission tomography (PET) based on the histogramming of time delay between the emission of a de-excitation photon, and a consequent electron-positron annihilation, to estimate the mean lifetime of orthopositronium (o-Ps), which depends on the local size of the voids, concentration of oxygen and bioactive molecules. We improve the resolution and reduce noise in positronium imaging by building time-delay spectra from the PET activity reconstructed by a 3-photon time-of-flight maximum likelihood expectation maximisation. The method was tested on the data measured for four human-tissue samples injected by 22Na and put in the Jagiellonian PET “Big barrel” scanner. Due to an ill-posed problem of fitting time-delay histograms, a multistage optimisation procedure was explored along with inferential analysis of the solution space. Run in parallel for multiple sets of initial guesses, we compared the second-order LevenbergMarquardt algorithm (LMA) and the direct search Nelder-Mead simplex (NMS) method. The LMA proved to be faster and more precise, but the NMS was more stable with a higher convergence rate. The estimated mean o-Ps lifetimes in the 1.9 ns – 2.6 ns range were consistent with the reference results, while other fitting parameters allowed differentiation between the two patients who provided the tissue samples.

Near-field coded-mask technique and its potential for proton therapy monitoring

Objective. Prompt-gamma imaging encompasses several approaches to the online monitoring of the beam range or deposited dose distribution in proton therapy. We test one of the imaging techniques - a coded mask approach - both experimentally and via simulations. Approach. Two imaging setups have been investigated experimentally. Each of them comprised a structured tungsten collimator in the form of a modified uniformly redundant array mask and a LYSO:Ce scintillation detector of fine granularity. The setups differed in detector dimensions and operation mode (1D or 2D imaging). A series of measurements with radioactive sources have been conducted, testing the performance of the setups for near-field gamma imaging. Additionally, Monte Carlo simulations of a larger setup of the same type were conducted, investigating its performance with a realistic gamma source distribution occurring during proton therapy. Main results. The images of point-like sources reconstructed from two small-scale prototypes’ data using the maximum-likelihood expectation maximisation algorithm constitute the experimental proof of principle for the near-field coded-mask imaging modality, both in the 1D and the 2D mode. Their precision allowed us to calibrate out certain systematic offsets appearing due to the limited alignment accuracy of setup elements. The simulation of the full-scale setup yielded a mean distal falloff retrieval precision of 0.72 mm in the studies for beam energy range 89.5–107.9 MeV and with 1 × 108 protons (a typical number for distal spots). The implemented algorithm of image reconstruction is relatively fast—a typical procedure needs several seconds. Significance. Coded-mask imaging appears a valid option for proton therapy monitoring. The results of simulations let us conclude that the proposed full-scale setup is competitive with the knife-edge-shaped and the multi-parallel slit cameras investigated by other groups.

A generalization of the maximum likelihood expectation maximization (MLEM) method: Masked-MLEM

Objective. In our previous work on image reconstruction for single-layer collimatorless scintigraphy, we developed the min–min weighted robust least squares (WRLS) optimization algorithm to address the challenge of reconstructing images when both the system matrix and the projection data are uncertain. Whereas the WRLS algorithm has been successful in two-dimensional (2D) reconstruction, expanding it to three-dimensional (3D) reconstruction is difficult since the WRLS optimization problem is neither smooth nor strongly-convex. To overcome these difficulties and achieve robust image reconstruction in the presence of system uncertainties and projection noise, we propose a generalized iterative method based on the maximum likelihood expectation maximization (MLEM) algorithm, hereinafter referred to as the Masked-MLEM algorithm. Approach. In the Masked-MLEM algorithm, only selected subsets (‘masks’) from the system matrix and the projection contribute to the image update to satisfy the constraints imposed by the system uncertainties. We validate the Masked-MLEM algorithm and compare it to the standard MLEM algorithm using experimental data obtained from both collimated and uncollimated imaging instruments, including parallel-hole collimated SPECT, 2D collimatorless scintigraphy, and 3D collimatorless tomography. Additionally, we conduct comprehensive Monte Carlo simulations for 3D collimatorless tomography to further validate the effectiveness of the Masked-MLEM algorithm in handling different levels of system uncertainties. Main results. The Masked-MLEM and standard MLEM reconstructions are similar in cases with negligible system uncertainties, whereas the Masked-MLEM algorithm outperforms the standard MLEM algorithm when the system matrix is an approximation. Importantly, the Masked-MLEM algorithm ensures reliable image reconstruction across varying levels of system uncertainties. Significance. With a good choice of system uncertainty and without requiring accurate knowledge of the actual system matrix, the Masked-MLEM algorithm yields more robust image reconstruction than the standard MLEM algorithm, effectively reducing the likelihood of erroneously reconstructing higher activities in regions without radioactive sources.

Study on the PET image quality according to various scintillation detectors based on the Monte Carlo simulation

Purpose: Positron emisson tomography (PET) is a crucial medical imaging scanner for the detection of cancer lesions. In order to maintain the improved image quality, it is crucial to apply detectors of superior performance. Therefore, the purpose of this study was to compare PET image quality using Monte Carlo simulation based on the detector materials of BGO, LSO, and LuAP. Materials and Methods: The Geant4 Application for Tomographic Emission (GATE) was used to design the PET detector. Scintillations with BGO, LSO and LuAP were modelled, with a size of 3.95 × 5.3 mm2 (width × height) and 25.0 mm (thickness). The PET detector consisted of 34 blocks per ring and a total of 4 rings. A line source of 1 MBq was modelled and acquired with a radius of 1 mm and length of 20 mm for 20 seconds. The acquired image was reconstructed maximum likelihood expectation maximization with 2 iteration and 10 subsets. The count comparison was carried out. Results and Discussion: The highest true, random, and scatter counts were obtained from the BGO scintillation detector compared to LSO and LuAP. Conclusion: The BGO scintillation detector material indicated excellent performance in terms of detection of gamma rays from emitted PET phantom.

A Portable Three-Layer Compton Camera for Wide-Energy-Range Gamma-ray Imaging: Design, Simulation and Preliminary Testing

(1) Background: The imaging energy range of a typical Compton camera is limited due to the fact that scattered gamma photons are seldom fully absorbed when the incident energies are above 3 MeV. Further improving the upper energy limit of gamma-ray imaging has important application significance in the active interrogation of special nuclear materials and chemical warfare agents, as well as range verification of proton therapy. (2) Methods: To realize gamma-ray imaging in a wide energy range of 0.3~7 MeV, a principle prototype, named a portable three-layer Compton camera, is developed using the scintillation detector that consists of an silicon photomultiplier array coupled with a Gd3Al2Ga3O12:Ce pixelated scintillator array. Implemented in a list-mode maximum likelihood expectation maximization algorithm, a far-field energy-domain imaging method based on the two interaction events is applied to estimate the initial energy and spatial distribution of gamma-ray sources. The simulation model of the detectors is established based on the Monte Carlo simulation toolkit Geant4. The reconstructed images of a 133Ba, a 137Cs and a 60Co point-like sources have been successfully obtained with our prototype in laboratory tests and compared with simulation studies. (3) Results: The proportion of effective imaging events accounts for about 2%, which allows our prototype to realize the reconstruction of the distribution of a 0.05 μSv/h 137Cs source in 10 s. The angular resolution for resolving two 137Cs point-like sources is 15°. Additional simulated imaging of the 6.13 MeV gamma-rays from 14.1 MeV neutron scattering with water preliminarily demonstrates the imaging capability for high incident energy. (4) Conclusions: We conclude that the prototype has a good imaging performance in a wide energy range (0.3~7 MeV), which shows potential in several MeV gamma-ray imaging applications.

Simulation of unfolding fast neutron spectrum from CR-39 track morphology

The aim of this study is to investigate the impact of various factors on neutron spectrum unfolding, including track parameters (major axis, minor axis and depth), polyethylene (PE) thickness, and etching time. The Monte Carlo code Geant4 and proton track parameter calculation program TRACK_P were applied to simulate the response matrix and track parameters spectrum. Four test spectra—252Cf spectrum, Gaussian broadened deuterium–deuterium fusion neutron spectrum, 241Am–Be spectrum, and Gaussian broadened deuterium–tritium fusion neutron spectrum—were studied using the maximum-likelihood expectation maximization algorithm for neutron spectrum unfolding. The unfolded spectra were in good agreement with the reference spectra. Results showed that the major and minor axes were more suitable than the track depth for neutron spectrum unfolding. For high-energy neutrons, a thick PE layer is required. Moreover, etching time does not significantly affect the unfolded spectra. This study explores the factors influencing the unfolding of fast neutron spectrum based on simulation calculations and preliminary experiments that can be useful for applying CR-39 to neutron spectrum measurements.

A stochastic alternative technique for Compton Maximum Likelihood Expectation-Maximization (MLEM) reconstruction

Even though iterative methods and particularly Maximum Likelihood Expectation-Maximization (MLEM) algorithms have been established in reconstruction with Compton data, their detailed design with respect to physical rules and processes dominate their plain implementation in the form of a system matrix. A new elementary but efficient alternative for the well-known system matrix with respect to Compton Camera image reconstruction is presented in this work. For each detected event there is a weighting factor inserted as an accumulated probability which carries all the necessary information. This probability which involves only the Compton scattering angle of the incident photon corresponds to a map that correlates all events within all possible source origins. Based on maximizing likelihood principles, the proposed model weights in a stochastic way the difference of the scatterer-to-source angle θ0, as it is determined by the deposited energy on the absorber, and any other potential scattering angle θJ, specified by the position coordinates on the reconstruction matrix. Obtained image spatial resolution, angular distortions and response to focal length determination are a few of the studied cases for the algorithms’ evaluation via simulations in GEANT4/GATE with a set of radioactive sources and phantoms with in- and out-of-plane arrangement.

Machine Learning-based reconstruction of single and double radiation source distribution using plastic scintillation optical fiber

In this study, we propose a novel method for reconstructing the radiation distribution of a one-dimensional radioactive source using Machine Learning (ML) algorithms and plastic scintillation optical fiber. The wavelength spectrum unfolding technique is used to estimate the source position accurately. We compare the accuracy and time efficiency of three different algorithms, namely, Generalized reduced gradient (GRG), Maximum likelihood expectation maximization (MLEM), and ML, in the single-source and dual-source cases. Our results demonstrate that MLEM algorithm has a shorter reconstruction time with comparable accuracy of position and intensity compared to GRG algorithm for single-source case. For dual-source case, ML algorithm provides real-time estimation of position and intensity with acceptable errors, while GRG algorithm has a larger error in intensity estimation and longer computation time. Our proposed ML algorithms offer useful guidance for practical applications in radiation source location.

Fully 3D implementation of the end-to-end deep image prior-based PET image reconstruction using block iterative algorithm

Objective. Deep image prior (DIP) has recently attracted attention owing to its unsupervised positron emission tomography (PET) image reconstruction method, which does not require any prior training dataset. In this paper, we present the first attempt to implement an end-to-end DIP-based fully 3D PET image reconstruction method that incorporates a forward-projection model into a loss function. Approach. A practical implementation of a fully 3D PET image reconstruction could not be performed at present because of a graphics processing unit memory limitation. Consequently, we modify the DIP optimization to a block iteration and sequential learning of an ordered sequence of block sinograms. Furthermore, the relative difference penalty (RDP) term is added to the loss function to enhance the quantitative accuracy of the PET image. Main results. We evaluated our proposed method using Monte Carlo simulation with [18F]FDG PET data of a human brain and a preclinical study on monkey-brain [18F]FDG PET data. The proposed method was compared with the maximum-likelihood expectation maximization (EM), maximum a posteriori EM with RDP, and hybrid DIP-based PET reconstruction methods. The simulation results showed that, compared with other algorithms, the proposed method improved the PET image quality by reducing statistical noise and better preserved the contrast of brain structures and inserted tumors. In the preclinical experiment, finer structures and better contrast recovery were obtained with the proposed method. Significance. The results indicated that the proposed method could produce high-quality images without a prior training dataset. Thus, the proposed method could be a key enabling technology for the straightforward and practical implementation of end-to-end DIP-based fully 3D PET image reconstruction.

Study of Alternative Imaging Methods for In Vivo Boron Neutron Capture Therapy.

Boron Neutron Capture Therapy (BNCT) is an innovative and highly selective treatment against cancer. Nowadays, in vivo boron dosimetry is an important method to carry out such therapy in clinical environments. In this work, different imaging methods were tested for dosimetry and tumor monitoring in BNCT based on a Compton camera detector. A dedicated dataset was generated through Monte Carlo tools to study the imaging capabilities. We first applied the Maximum Likelihood Expectation Maximization (MLEM) iterative method to study dosimetry tomography. As well, two methods based on morphological filtering and deep learning techniques with Convolutional Neural Networks (CNN), respectively, were studied for tumor monitoring. Furthermore, clinical aspects such as the dependence on the boron concentration ratio in image reconstruction and the stretching effect along the detector position axis were analyzed. A simulated spherical gamma source was studied in several conditions (different detector distances and boron concentration ratios) using MLEM. This approach proved the possibility of monitoring the boron dose. Tumor monitoring using the CNN method shows promising results that could be enhanced by increasing the training dataset.

MLEM Neutron Spectra Unfolding in a Radiotherapy Bunker Using Bonner Sphere Spectrometer

High-energy radiotherapy treatments of a medical Linear Accelerator (LinAc) generate secondary neutrons that can produce health damage on the human body as the induction of secondary cancers. The energy spectrum of these neutrons must be determined to estimate the extra dose received by patients inside a radiotherapy room during radiotherapy treatment. To quantify the neutron production, a Ludlum Bonner sphere spectrometer (BSS) is used for measurement at different points of a LinAc bunker at the Hospital Universitari i Politècnic La Fe de València. With the neutron measured data and a set of response detector curves obtained by Monte Carlo simulations with MCNP6.1.1, the Maximum Likelihood Expectation Maximization unfolding method is used to unfold the energy neutron spectrum. Unfolded neutron spectra at different locations were compared to those obtained by Monte Carlo simulation of the same setup, showing the same energetic behavior. The fluence rate decreases with source distance, and the shape changes from a fast neutron peak in the nearest LinAc head location to a prominent thermal neutron peak in the bunker maze region. Moreover, the neutron ambient equivalent dose was obtained from the unfolded spectra and compared to Berthold detector measurements, being consistent.