Time Of Flight PET (TOF-PET) is a transformative technology for PET systems, but reaping its fullest potential requires achieving very high spatial and timing resolution, and overcoming dependency on signal variability at the level of the individual pixels. In this study, we try to quantify the advantage of using custom waveform sampling devices versus TDC methodology using realistic estimates of noise and other non-idealities. With the use of DNN methodology we show achievable gain for current and future acquisition systems, and also demonstrate the feasibility of DOI estimation from single side readout. We conclude by arguing for the scalability of such a system based on our experience with compact waveform digitizer design.
Recent advances in deep learning (DL) have greatly improved the performance of positron emission tomography (PET) denoising performance. However, DL model performance can vary a lot across subjects, due to the large variability of the count levels and spatial distributions. A generalizable DL model that mitigates the subject-wise variations is highly expected toward a reliable and trustworthy system for clinical application. In this work, we propose a contrastive adversarial learning framework for subject-wise domain generalization (DG). Specifically, we configure a contrastive discriminator in addition to the UNet-based denoising module to check the subject-related information in the bottleneck feature, while the denoising module is adversarially trained to enforce the extraction of subject-invariant features. The sampled low-count realizations from the list-mode data are used as anchor-positive pairs to be close to each other, while the other subjects are used as negative samples to be distributed far away. We evaluated on 97 18F-MK6240 tau PET studies, each having 20 noise realizations with 25% fractions of events. Training, validation, and testing were implemented using 1400, 120, and 420 pairs of 3D image volumes in a subject-independent manner. The proposed contrastive adversarial DG demonstrated superior denoising performance than conventional UNet without subject-wise DG and cross-entropy-based adversarial DG.
Diffusion models (DM) built from a hierarchy of denoising autoencoders have achieved remarkable progress in image generation, and are increasingly influential in the field of image restoration (IR) tasks. In the meantime, its backbone of autoencoders also evolved from UNet to vision transformer, e.g. Restormer. Therefore, it is important to disentangle the contribution of backbone networks and the additional generative learning scheme. Notably, DM shows varied performance across IR tasks, and the performance of recent advanced transformer-based DM on PET denoising is under-explored. In this study, we further raise an intuitive question, "{if we have a sufficiently powerful backbone, whether DM can be a general add-on generative learning scheme to further boost PET denoising}". Specifically, we investigate one of the best-in-class IR models, i.e., DiffIR, which is a latent DM based on the Restormer backbone. We provide a qualitative and quantitative comparison with UNet, SR3 (UNet+pixel DM), and Restormer, on the 25% low dose 18F-FDG whole-body PET denoising task, aiming to identify the best practices. We trained and tested on 93 and 12 subjects, and each subject has 644 slices. It appears that Restormer outperforms UNet in terms of PSNR and MSE. However, additional latent DM over Restormer does not contribute to better MSE, SSIM, or PSNR in our task, which is even inferior to the conventional UNet. In addition, SR3 with pixel space DM is not stable to synthesize satisfactory results. The results are consistent with the natural image super-resolution tasks, which also suffer from limited spatial information. A possible reason would be the denoising iteration at latent feature space cannot well support detailed structure and texture restoration. This issue is more crucial in the IR tasks taking inputs with limited details, e.g., SR and PET denoising.
Delineating lesions and anatomical structure is important for image-guided interventions. Point-supervised medical image segmentation (PSS) has great potential to alleviate costly expert delineation labeling. However, due to the lack of precise size and boundary guidance, the effectiveness of PSS often falls short of expectations. Although recent vision foundational models, such as the medical segment anything model (MedSAM), have made significant advancements in bounding-box-prompted segmentation, it is not straightforward to utilize point annotation, and is prone to semantic ambiguity. In this preliminary study, we introduce an iterative framework to facilitate semantic-aware point-supervised MedSAM. Specifically, the semantic box-prompt generator (SBPG) module has the capacity to convert the point input into potential pseudo bounding box suggestions, which are explicitly refined by the prototype-based semantic similarity. This is then succeeded by a prompt-guided spatial refinement (PGSR) module that harnesses the exceptional generalizability of MedSAM to infer the segmentation mask, which also updates the box proposal seed in SBPG. Performance can be progressively improved with adequate iterations. We conducted an evaluation on BraTS2018 for the segmentation of whole brain tumors and demonstrated its superior performance compared to traditional PSS methods and on par with box-supervised methods.
SymPET is a low-power, high channel density, waveform-digitizing readout microchip under development at Nalu Scientific for SiPM-based TOF PET applications. Our "System on Chip" waveform digitizing architecture includes features such as fully random accessible analog storage, input triggering, and on-chip biasing, control and waveform feature extraction capabilities, enabling many effective mechanisms to optimize features such as e.g., throughput, speed, and buffer length while simultaneously allowing excellent control of systematic effects which typically significantly affect the timing precision of time-over-threshold based readouts. Preliminary results show that SymPET can provide less than 10 ps timing jitter at a reasonable power budget which is substantially better compared to existing solutions. This will allow for high-channel density and/or limited angle applications such as ultra high resolution brain TOF PET designs that require proper management of heat dissipation, while necessitating high spatial and timing resolution.
New modular data acquisition electronics for cross-strip cadmium zinc telluride (CZT) detector readout were developed to lower the electronic noise and increase the applied voltage bias capabilities compared to previous designs. The modular data acquisition electronics will be utilized to develop a two-panel head and neck dedicated positron emission tomography (PET) system. Each panel will consist of 150 CZT crystals (4 × 4 × 0.5cm 3 ) covering an area of 20x15 cm2 in an edge-on configuration to achieve high detector efficiency at 511 keV. This paper presents the first characterization results of CZT detectors based on the new readout electronics system. Ge-68 and Cs-137 were used as the point source for measuring the energy spectra. Three individual CZT detectors (117 anode channels) were tested together with a full data acquisition chain. The mean FWHM energy resolution across all 117 anode channels is 8.36% ± 0.52% at 511 keV without any correction and the lower keV threshold of the energy spectra resides between 0 keV to 50 keV. For reference, the previous small animal system had a mean FWHM energy resolution of 10.78% ±1.45% at 511 keV before correction and a lower keV threshold of 100 keV to 150 keV.
Instrumentation research in small animal Positron Emission Tomography (PET) imaging is driven by improving timing, spatial resolution and sensitivity. Conventional PET scanners are built of multiple detectors placed in a cylindrical geometry with gaps between them in both the transaxial and axial planes. These gaps decrease sensitivity and degrade spatial resolution towards the edges of the system field of view (FOV). To mitigate these problems, we have designed and validated an edgeless pre-clinical PET system based on a single LYSO annulus with an inner diameter of 62 mm and 10 outer facets of 26 × 52 mm2 each. The scintillation light is read out using the row and columns of Silicon Photomultipliers (SiPMs) mounted in magnetic-field compatible PCBs. The objective of this work is to provide a calibration method for this system. The particular design of the annulus produces some undesirable effects in the light distributions (LD) at the module joints, which needs to be addressed. Nevertheless, after calibration, the system allows one to properly retrieve both, the energy and 3D photon impact positions.
The time response of a Silicon Photomultiplier (SiPM) depends on some of the intrinsic parameters of the sensor. Combining multiple and small SiPM instead of one with larger area will reduce the detector capacitance at electronic level, which can be translated into a lower signal jitter. This will improve the Single Photon Time Resolution (SPTR) of the sensor and the electronics and thus the Coincidence Time Resolution (CTR) of a PET system. This effect is studied using a combination of a physics simulation environment (GATE) and a commercial electrical simulator. GATE is an advanced opensource software developed by the international collaboration OpenGATE, dedicated to numerical simulations in medical imaging and radiotherapy. Accurate modelling of photon interactions with crystal surfaces is essential in optical simulations, but the existing UNIFIED model in GATE is often inaccurate, especially for rough surfaces. A new approach has been developed for GATE, named Davis Model. This method calculates the reflectance properties from the crystal topography previously measured with a surface scan device (AFM, CONFOCAL...). In combination with GATE, an electrical simulator is being used on this work to simulate the response of the SiPM and the front-end readout electronics. The aim of this work is to provide a framework that will enable a global optimization of the PET system that consider the scintillator, the sensor (sensor size, pixel pitch, dead area, capacitance) and the readout electronics (input impedance, noise, bandwidth, summation).
The Strontium Iodide Radiation Instrument (SIRI) is a single detector, gamma-ray spectrometer designed to space-qualify the new scintillation detector material europium-doped strontium iodide (SrI2:Eu) and new silicon photomultiplier (SiPM) technology. SIRI covers the energy range from.04 - 8 MeV and was launched into 600 km sun-synchronous orbit on Dec 3, 2018 onboard STPSat-5 with a one-year mission to investigate the detector's response to on-orbit background radiation. The detector has an active volume of 11.6 cm(3) and a photo fraction efficiency of 50% at 662 keV for gamma-rays parallel to the long axis of the crystal. Its spectroscopic resolution of 4.3% was measured by the fullwidth-half-maximum of the characteristic Cs-137 gammaray line at 662 keV. Measured background rates external to the trapped particle regions are 40- 50 counts per second for energies greater than 40 keV and are largely the result of short- and long-term activation products generated by transits of the South Atlantic Anomaly (SAA) and the continual cosmic-ray bombardment. Rate maps determined from energy cuts of the collected spectral data show the expected contributions from the various trapped particle regions. Early spectra acquired by the instrument show the presence of at least 10 characteristic gamma-ray lines and a beta continuum generated by activation products within the detector and surrounding materials. As of April 2019, the instrument has acquired over 1000 hours of data and is expected to continue operations until the space vehicle is decommissioned in Dec. 2019. Results indicate SrI2:Eu provides a feasible alternative to traditional sodium iodide and cesium iodide scintillators, especially for missions where a factor- of-two improvement in energy resolution would represent a significant difference in scientific return. To the best of our knowledge, SIRI is the first on-orbit use of SrI2:Eu scintillator with SiPM readouts.
This paper describes a deep learning method that provides an improved image quality for single focal spot CT scan. An experiment with convolutional neural network learning is based on 30 clinical head datasets of flying focal spot images and the corresponding single focal spot images. The generated anti-aliasing images reduce streak artifacts and noise granularity, and improve the contrast of bony structure, which potentially meet with the high diagnostic criteria in clinical applications. At the same time, the CT system simplicity and stability properties are easier to maintain compared to a system with the flying focal spot system. Further, results for applications such as inner ear and extremities are also of diagnostic quality.
The ATLAS experiment at the Large Hadron Collider (LHC) is currently preparing for major detector upgrades for the Phase-II of the LHC operation, scheduled to start in 2026. In order to achieve an integrated luminosity up to 4000 fb -1 within the High Luminosity LHC (HL-LHC) phase, the instantaneous luminosity is expected to reach unprecedented values. To cope with the expected radiation damage and high density of tracks per bunch crossing, a complete replacement of the existing Inner Detector of ATLAS is required.An all-silicon Inner Tracker (ITk) is under development with a pixel detector subsystem surrounded by the strip detector subsystem at larger radii, aiming to provide an increased tracking coverage up to pseudorapidities of four.In this report an overview of the strip detector system design of the ITk is given as well as the results from an extensive prototyping effort are presented. The main focus is on recent testbeam results of the validation of the performance of un-irradiated and irradiated strip detector modules. These are able to reach the specified performance requirements in the HL-LHC environment. The next step of the ITk project is the transition to the production phase to provide a functional detector by 2026.
The improvement of manufacturing technologies is the key aspect in nuclear science experiments. In this work we present the results of the development aimed to improve gamma detection modules based on the use of Silicon Drift Detectors (SDD). Previous attempts to build this type of detector showed issues in terms of reliability and robustess which limit the use of this solution in experiments. The introduction of a passivation layer based on deposition of optical resin on the light entrance window of the SDD allows to enhance robustness and to couple the Silicon detector to scintillators for gamma-ray spectroscopy. Experimental results show multiple times that no worsening of electronics noise is observed after resin deposition.
The goal of this work is to design a detector with ≤ 2 mm intrinsic resolution and multiple levels of depth of interaction (DOI), suitable for brain dedicated Positron Emission Tomography (PET). Our starting point is a 2 cm thick LYSO:Ce crystal with single-side readout, containing laser induced optical barriers (LIOB) in a simple pixel like pattern extending half-way through the crystal thickness. While this design is characterized by depth variation of the scintillation light spread in the detector, this dependence is complex due to the discontinuities in the design. Furthermore, the transversal resolution is limited in the unprocessed part of the scintillator where no optical barriers exist. In this work we are therefore exploring the possibility of enhancing performance beyond the basic design, by adding single photodetector elements to the sides and/or top of the scintillator. Our results show that the addition of side and/or top photodetectors can indeed improve DOI performance by providing more linear DOI curves, and the side photodetectors will provide additional XY information that can be exploited for improved positioning.
The present work describes a new channel architecture suited to reading out Silicon Photomultipliers (SiPM). The aim is to develop a multichannel Application-Specific Integrated Circuit (ASIC) in a standard CMOS 130 nm technology that achieves excellent timing accuracy while fully exploiting the dynamic range of large area SiPMs. The sensor is AC-coupled to the differential front-end, which features two separated timing and charge signal processing paths. The first one exploits a differential input current-mode preamplifier and an embedded Time-to-Digital Converter (TDC) that exhibits a 10 ps binning. Accurate timing and low jitter are also attained by means of a high-speed discriminator with threshold adjustable in small steps just above the baseline, which operates on the very steep edge of the output signal produced by the preamplifier. The charge measurement signal path is based on an active-RC integrator, a peak detector and an Analog-to-Digital Converter (ADC). Operating the front-end at 1.2 V supply voltage, with a power consumption of 10 mW, a simulated Single-Photon Time Resolution (SPTR) of 78 ps at Full-Width Half Maximum (FWHM) is achieved, linearly covering an input dynamic range of 1280 pC that corresponds to ~8000 photoelectrons with a SiPM gain of ~10 6 .
The new Digital Readout Module, named DRM2, designed for the upgrade of the readout of the ALICE TOF detector at CERN, has been produced and is ready to face the commissioning stage. It is a narrow 9U VME card (16 cm x 33 cm) with the goal to read out the TDC (Time to Digital Converter) chips on 9 or 10 VME boards, named TRMs (TDC Readout Module). A test setup for the DRM2 production card validation was designed and built, to check and validate the correct behavior of the produced boards. The test setup and related validation results will be described in detail in this paper.
Compared to traditional cylindrical ring PET, spherical PET (S-PET) can obtain more information which perform better in reducing parallax error and improving geometrical sensitivity. A polyhedral brain-dedicated PET with neck opening reserved, which was assembled by 11 pentagon detectors and 15 hexagon detectors, was preliminarily designed with the advantages mentioned above kept. Our simulation study for single-layer hexagon detector is aimed at finding out decoding methods with lower position error. The neural network (NN) method and some analytical methods with center-of-gravity (COG) or maximum-value (MMV) have been conducted. The results of simulations with the NN method showed that: (1) The NN method get higher resolution in edged area than central area. (2) Reducing reflectivity of the film can improve resolution around central area with the cost of less received photons. (3) The resolution in central area is below 0.9 mm and the resolution in edged area is below 0.5 mm. However, the simulations with analytical methods haven't achieved satisficing resolution yet. The methods adopted have shown obvious drifts with multiple sides' COG or MMV, and methods with single side's COG or MMV have shown dead zones which leads to terrible result. Our next step is to optimize the simulation with analytical methods. And the experiments will be carried out to verify the simulation and evaluate the performance of polyhedral PET.
Accurate cone-beam computed tomography (CBCT) reconstruction requires knowing the true geometric parameters of the scanner, obtained from calibrating the system. Positioning the detector panel offset to the source-detector axis is a technique used by linac-mounted CBCTs in the context of radiation therapy in order to image the full width of the patient over the 360-degree scan. However, offsetting the detector panel increases the difficulty of calibration as the resulting projections will be truncated along their width. We present an extended version of an existing phantom-based geometric calibration method that we have adapted for use in offset detector systems. This calibration method extracts the required geometric parameters from measured projection images of a specific calibration phantom we designed for use in the method. The calibration has been implemented in tandem with a variant of the Feldkamp-DavisKress (FDK) reconstruction algorithm which we have modified to both integrate our calibration and reconstruct width-truncated projections. The calibration and modified FDK algorithm are validated by the successful reconstruction of a simulated SheppLogan phantom generated with misaligned detector geometry (detector panel shifting, and variable detector panel rotation along the 360-degree scan). The method also applies to other types of detector misalignment. A physical version of the phantom has been constructed to validate the method with real data in the future.
We present the fabrication and the first functional tests on a new class of silicon devices: the AC-coupled Low-Gain Avalanche Diodes (AC-LGAD). Because of its good timing performance, the LGAD was originally developed to support the silicon tracker at High-Energy Physics experiments for the reconstruction of pile-up events originating from a same bunch crossing. Pile-up events happen in fact at slightly different times and adding a fourth dimension ( the time) is crucial to resolve them. On the other hand, LGADs suffers from poor spatial resolution. AC-LGADs offer a solution to this drawback, while retaining the good timing performance. Test structures have been fabricated at BNL, using the internal silicon processing facility, and have been tested with fast electronics and radioactive sources. Large gains and fast signal have been measured, while the noise stays low and comparable to the standard LGAD case. Other wafer fabrications are ongoing to further improving performances and correct pitfalls in the very first fabrication.
CT Perfusion (CTP) imaging is one of the most common regimes for evaluation of acute ischemic stroke patients. The CTP imaging protocol typically involves the rapid acquisition of several frames of the brain volume over ~1 minute, following contrast administration. Therefore, it is associated with a relatively high radiation dose. The ability to reduce this dose, while maintaining the accuracy of image-based stroke analysis is highly desirable. However, a reduction in dose is accompanied by an increase in noise, which can compromise computation of important haemodynamic parameters during stroke analysis. In this paper, we investigate the feasibility of using 3D conditional generative adversarial networks (3D c-GANs) to achieve CTP dose reduction while preserving image quality. We simulated low-dose CTP images corresponding to tube currents of 100 mAs and 45 mAs for 18 positive acute stroke subjects and applied a 3D c-GANs model to estimate the standard dose CTP images from the simulated low-dose images. We also compared two different strategies for handling the 4D nature of the CTP data in the 3D c-GANs model. Qualitatively, the results showed excellent agreement between the estimated low-noise images and the true images. Quantitative assessment also showed good performance of the model associated with high peak signal-to-noise ratio (PNSR) around 40 dB, normalized mean squared error (NMSE) close to zero, and structural similarity index (SSIM) close to 1. By stacking the original data rather than concatenating all volumes, the results were improved by 1.05 dB PNSR, 0.005 NMSE, and 0.01 SSIM at the simulated exposure of 100mAs, and by 0.93dB PNSR, 0.019 NMSE at the simulated tube current of 45 mAs. The results show good promise for dose reduction in CTP, however we are currently performing full stroke modelling analysis on the synthetic images to validate the method.
A prototype handheld dual particle imager composed of stilbene bars coupled to silicon photomultipliers was used to measure a 4.5 kg sphere of alpha-phase weapons-grade plutonium and a canister containing 3.4 kg of plutonium oxide (7% 240P u and 93% 239P u). Each object was measured independently and both objects were measured in the same field of view separated by 50 cm. Experimental results of the measurements conducted are presented. These results demonstrate the ability of the handheld dual particle imager to image and obtain neutron spectroscopic information for sources that would be of great interest for nuclear safeguards and emergency response applications.
