A. Iborra

Calibration and Performance Tests of Detectors for Laser-Accelerated Protons

We present the calibration and performance tests carried out with two detectors for intense proton pulses accelerated by lasers. Most of the procedures were realized with proton beams of 0.46-5.60 MeV from a tandem accelerator. One approach made use of radiochromic films, for which we calibrated the relation between optical density and energy deposition over more than three orders of magnitude. The validity of these results and of our analysis algorithms has been confirmed by controlled irradiation of film stacks and reconstruction of the total beam charge for strongly non-uniform beam profiles. For the spectral analysis of protons from repeated laser shots, we have designed an online monitor based on a plastic scintillator. The resulting signal from a photomultiplier directly measured on a fast oscilloscope is especially useful for time-of-flight applications. Variable optical filters allow for suppression of saturation and an extension of the dynamic range. With pulsed proton beams we have tested the detector response to a wide range of beam intensities from single particles to 3 ×105 protons per 100 ns time interval.

3-D photon impact determination using fitting approaches to the Light Distribution

In Positron Emission Tomography (PET) detectors based in monolithic scintillators, the spatial resolution is limited by the accuracy in the determination of the interaction coordinates from the 511 keV photons. When linear algorithms, such as Center of Gravity (CoG) are used a poor estimation of the interaction positions, specially towards the edges is the major limitation in spatial resolution. A novel PET detector block, where complete information of Light Distribution (LD) for each event is available, allows to fit each event to a theoretical model, improving the estimation of the interaction coordinates, and minimizing border effects. In this work, by means of the LD fitting approach, we were able to obtain an average spatial resolution of 1.2 mm in the entire scintillator volume and an average depth of interaction (DOI) resolution of 1.5 mm. Moreover, splitting the data in three DOI regions, we obtained an average spatial resolution of 1.0 mm at the DOI region closer to the photodetectors. Finally, it is remarkable that the implementation of the LD fitting approach is capable of processing up to 50 kcps in a octacore system.

Expectation maximization (EM) algorithms using polar symmetries for computed tomography (CT) image reconstruction

We suggest a symmetric-polar pixellation scheme which makes possible a reduction of the computational cost for expectation maximization (EM) iterative algorithms. The proposed symmetric-polar pixellation allows us to deal with 3D images as a whole problem without dividing the 3D problem into 2D slices approach. Performance evaluation of each approach in terms of stability and image quality is presented. Exhaustive comparisons between all approaches were conducted in a 2D based image reconstruction model. From these 2D approaches, that showing the best performances were finally implemented and evaluated in a 3D based image reconstruction model. Comparison to 3D images reconstructed with FBP is also presented. Although the algorithm is presented in the context of computed tomography (CT) image reconstruction, it can be applied to any other tomographic technique as well, due to the fact that the only requirement is a scanning geometry involving measurements of an object under different projection angles. Real data have been acquired with a small animal (CT) scanner to verify the proposed mathematical description of the CT system.

EM tomographic image reconstruction using polar voxels

The splitting of the field of view (FOV) in polar voxels is pro posed in this work in order to obtain an efficient description of a cone-beam compu ted tomography (CT) scanner. The proposed symmetric-polar pixelation makes it possible to deal with the 3D iterative reconstruction considering a number of projections and voxel sizes typical in CT preclinical imaging. The performance comparison, between the filtered backproje ction (FBP) and 3D maximum likelihood expectation maximization (MLEM) reconstruction algorithm for CT, is presented. It is feasible to achieve the hardware spatial resolution limi t with the considered pixelation. The image quality achieved with MLEM and FBP have been analyzed. The results obtained with both algorithms in clinical images have been compared too. Although the polar-symmetric pixelation is presented in the context of CT imaging, it can be applied to any other tomographic technique as long as the scan comprises the measurement of an object under several projection angles.

Characterization of protons accelerated from a 3 TW table-top laser system

This project has been funded by Centro para el Desarrollo Tecnologico Industrial (CDTI, Spain) within the INNPRONTA program, grant no. IPT-20111027, by EUROSTARS project E9113, and by the Spanish Ministry for Economy and Competitiveness within the Retos-Colaboracion 2015 initiative, ref. RTC-2015-3278-1.

Detailed requirements for a laser-based proton/ion accelerator for radioisotope production

Background and objectives: Laser-plasma acceleration of protons and ions is often considered a promising technique for compact applications of highly intense beams of multi-MeV particles. A remarkable example is the on-site production of short-lived radioisotopes for medical and preclinical interventions. We study quantitatively the activity of four important PET isotopes which may be obtained by irradiation of suitable target nuclei with laser-accelerated protons and deuterons. These simulations allow for confining the range of useful parameters of a laser-based production system. Methods: We choose a total of ten p- and d-induced reaction channels for the production of F-18, C-11, O-15, and N-13 from suitable target nuclei. We calculate the activity yield as a function of projectile energy starting from the corresponding, known cross sections. In order to simulate typical laser-plasma particle spectra we generate exponentially decaying distributions spread over a wide range, up to a maximum energy between 6 and 16 MeV. From the yield curves and the spectra we obtain the single-shot activation and the total activity after a realistic production time at 100 Hz pulse rate, taking into account saturation effects due to decay during irradiation. Results: We present numerical results for ten reaction channels and six realistic projectile spectra. With single laser shots, the highest activities are generated for O-15 (up to 11 (20) kBq for 16 MeV maximum p (d) energy). After prolonged irradiation at 100 Hz pulse rate, useful quantities of C-11 and O-15 may be obtained from spectra with 10 MeV maximum energies. The production of N-13 and F-18, to the contrary, requires higher energies and/or shot rates. Conclusions: 10 MeV particle energy and 100 Hz pulse rate are realistic benchmarks for a laser-based PET isotope production system. Experimental work to achieve these demanding objectives is in progress.

Sparse Givens resolution of large system of linear equations: Applications to image reconstruction

In medicine, computed tomographic images are reconstructed from a large number of measurements of X-ray transmission through the patient (projection data). The mathematical model used to describe a computed tomography device is a large system of linear equations of the form AX=B. In this paper we propose the QR decomposition as a direct method to solve the linear system. QR decomposition can be a large computational procedure. However, once it has been calculated for a specific system, matrices Q and R are stored and used for any acquired projection on that system. Implementation of the QR decomposition in order to take more advantage of the sparsity of the system matrix is discussed.

A brain PET insert MR compatible: Final design and first results

A whole-body PET device is sometimes not suitable for brain studies because the achieved image resolution is typically not sufficient to investigate small size structures. Thus, a dedicated brain PET insert system with high performance would overcome such limitations. Moreover, these functional studies lack of anatomical information. It is shown elsewhere the convenience of simultaneously acquisition of PET and MR data. In this work we show the final design and first pilot evaluation tests of a novel brain PET insert. Each detector block is based on a monolithic scintillation crystal, an array of SiPMs and a readout allowing characterizing the scintillation light distribution in the X and Y detector axes. The scintillators have a parallelepiped geometry with dimensions of 50×50×20 mm3. Their lateral walls are black painted and with the entrance face coupled to a retroreflector optical layer. We have determined an average (XYZ) detector spatial resolution through the FWHM of 1.2 mm (whole scintillator volume). The DOI resolution was measured with lateral incidence experiments and found to be about 3.5 mm, also on average for all photons depth of interactions and crystal positions. Thanks to the retroreflector, the energy resolution improves when compared to a case with all surfaces black painted, resulting on an average value of 13%. The tomographic reconstruction of the data was evaluated using different algorithms, including analytical (FBP STIR-3D), iterative (MLEM and List Mode OS) and a novel method that provides images by directly tracing the measured LORs. The minimum pixel/voxel sizes that were tried are 0.8/0.4 mm, 1.0/0.5 mm and 0.16/0.16 mm, respectively. All methods made it possible to show the PET system capabilities to resolve 1.6 mm rods in a Derenzo-like phantom filled with 150 uCi and scanned for 20 minutes. Pilot tests of the PET insert inside a clinical 3T MR showed a good system performance for most of the sequences typically used for brain imaging.

Determination of the Interaction Position of Gamma Photons in Monolithic Scintillators Using Neural Network Fitting

In Positron Emission Tomography (PET) detectors based on monolithic scintillators, the photon interaction position needs to be estimated from the light distribution (LD) on the photodetector pixels. Due to the finite size of the scintillator volume, the symmetry of the LD is truncated everywhere except for the crystal center. This effect produces a poor estimation of the interaction positions towards the edges, an especially critical situation when linear algorithms, such as Center of Gravity (CoG), are used. When all the crystal faces are painted black, except the one in contact with the photodetector, the LD can be assumed to behave as the inverse square law, providing a simple theoretical model. Using this LD model, the interaction coordinates can be determined by means of fitting each event to a theoretical distribution. In that sense, the use of neural networks (NNs) has been shown to be an effective alternative to more traditional fitting techniques as nonlinear least squares (LS). The multilayer perceptron is one type of NN which can model non-linear functions well and can be trained to accurately generalize when presented with new data. In this work we have shown the capability of NNs to approximate the LD and provide the interaction coordinates of γ-photons with two different photodetector setups. One experimental setup was based on analog Silicon Photomultipliers (SiPMs) and a charge division diode network, whereas the second setup was based on digital SiPMs (dSiPMs). In both experiments NNs minimized border effects. Average spatial resolutions of 1.9 ±0.2 mm and 1.7 ±0.2 mm for the entire crystal surface were obtained for the analog and dSiPMs approaches, respectively.

Analysis of the Statistical Moments of the Scintillation Light Distribution With dSiPMs

In γ-ray detectors, monolithic scintillation crystals offer the possibility of preserving the scintillation light distribution especially when painted black. The statistical moments of this distribution provide accurate information on the three photon impact coordinates, including their depth of interaction (DOI). Digital SiPMs (dSiPMs) return digital information based on pixels about the collected light distribution, since the signal is a digital sum of the trigger bins. In this work we present, for the first time, an accurate analysis of the statistical moments of the light distribution using monolithic painted black crystals and state-of-the-art dSiPMs. Two 32.6 ×32.6 mm2 monolithic LYSO crystals covering the entire photodetectors area have been used in coincidence with 10 mm in thickness. The photosensor tiles were kept at a stable temperature of T = 20 °C. Energy resolution of about 18% was reached in relation to the zeroth moment. The first moment, related to the impact position, determined a spatial resolution of about 3 mm near the crystal center, but quadratically degrading towards the crystal borders. The DOI resolution, measured by means of the second moment, was found to be nearing 4 mm in the crystal center region. The third order moment, the so-called skewness, is related to the degree of truncation and once calibrated minimizes the compression effects. A corrected spatial resolution of about 3 mm was then measured for the entire crystal surface. DOI resolution improved at the crystals center, reaching 3.5 mm, but a degradation towards the borders remained due to truncation of the scintillation light distribution.