Intercrystal scatter studies for a 1 mm3 resolution clinical PET system prototype

Strategies for mitigating inter-crystal scattering effects in positron emission tomography: a comprehensive review

Inter-crystal scattering (ICS) events in Positron Emission Tomography (PET) present challenges affecting system sensitivity and image quality. Understanding the physics and factors influencing ICS occurrence is crucial for developing strategies to mitigate its impact. This review paper explores the physics behind ICS events and their occurrence within PET detectors. Various methodologies, including energy-based comparisons, Compton kinematics-based approaches, statistical methods, and Artificial Intelligence (AI) techniques, which have been proposed for identifying and recovering ICS events accurately are introduced. Energy-based methods offer simplicity by comparing energy depositions in crystals. Compton kinematics-based approaches utilize trajectory information for first interaction position estimation, yielding reasonably good results. Additionally, statistical approach and AI algorithms contribute by optimizing likelihood analysis and neural network models for improved positioning accuracy. Experimental validations and simulation studies highlight the potential of recovering ICS events and enhancing PET sensitivity and image quality. Especially, AI technologies offers a promising avenue for addressing ICS challenges and improving PET image accuracy and resolution. These methods offer promising solutions for overcoming the challenges posed by ICS events and enhancing the accuracy and resolution of PET imaging, ultimately improving diagnostic capabilities and patient outcomes. Further studies applying these approaches to real PET systems are needed to validate theoretical results and assess practical implementation feasibility.

Self-normalization for a 1 mm3resolution clinical PET system using deep learning

Objective.This work proposes, for the first time, an image-based end-to-end self-normalization framework for positron emission tomography (PET) using conditional generative adversarial networks (cGANs).Approach.We evaluated different approaches by exploring each of the following three methodologies. First, we used images that were either unnormalized or corrected for geometric factors, which encompass all time-invariant factors, as input data types. Second, we set the input tensor shape as either a single axial slice (2D) or three contiguous axial slices (2.5D). Third, we chose either Pix2Pix or polarized self-attention (PSA) Pix2Pix, which we developed for this work, as a deep learning network. The targets for all approaches were the axial slices of images normalized using the direct normalization method. We performed Monte Carlo simulations of ten voxelized phantoms with the SimSET simulation tool and produced 26,000 pairs of axial image slices for training and testing.Main results.The results showed that 2.5D PSA Pix2Pix trained with geometric-factors-corrected input images achieved the best performance among all the methods we tested. All approaches improved general image quality figures of merit peak signal to noise ratio (PSNR) and structural similarity index (SSIM) from ∼15 % to ∼55 %, and 2.5D PSA Pix2Pix showed the highest PSNR (28.074) and SSIM (0.921). Lesion detectability, measured with region of interest (ROI) PSNR, SSIM, normalized contrast recovery coefficient, and contrast-to-noise ratio, was generally improved for all approaches, and 2.5D PSA Pix2Pix trained with geometric-factors-corrected input images achieved the highest ROI PSNR (28.920) and SSIM (0.973).Significance.This study demonstrates the potential of an image-based end-to-end self-normalization framework using cGANs for improving PET image quality and lesion detectability without the need for separate normalization scans.

Crystal scatter effects in a large-area dual-panel Positron Emission Mammography system

Positron Emission Mammography (PEM) is a valuable molecular imaging technique for breast studies using pharmaceuticals labeled with positron emitters and dual-panel detectors. PEM scanners normally use large scintillation crystals coupled to sensitive photodetectors. Multiple interactions of the 511 keV annihilation photons in the crystals can result in event mispositioning leading to a negative impact in radiopharmaceutical uptake quantification. In this work, we report the study of crystal scatter effects of a large-area dual-panel PEM system designed with either monolithic or pixelated lutetium yttrium orthosilicate (LYSO) crystals using the Monte Carlo simulation platform GATE. The results show that only a relatively small fraction of coincidences (~20%) arise from events where both coincidence photons undergo single interactions (mostly through photoelectric absorption) in the crystals. Most of the coincidences are events where at least one of the annihilation photons undergoes a chain of Compton scatterings: approximately 79% end up in photoelectric absorption while the rest (<1%) escape the detector. Mean positioning errors, calculated as the distance between first hit and energy weighted (assigned) positions of interaction, were 1.70 mm and 1.92 mm for the monolithic and pixelated crystals, respectively. Reconstructed spatial resolution quantification with a miniDerenzo phantom and a list mode iterative reconstruction algorithm shows that, for both crystal types, 2 mm diameter hot rods were resolved, indicating a relatively small effect in spatial resolution. A drastic reduction in peak-to-valley ratios for the same hot-rod diameters was observed, up to a factor of 14 for the monolithic crystals and 7.5 for the pixelated ones.

Technical opportunities and challenges in developing total-body PET scanners for mice and rats

Positron emission tomography (PET) is the most sensitive in vivo molecular imaging technique available. Small animal PET has been widely used in studying pharmaceutical biodistribution and disease progression over time by imaging a wide range of biological processes. However, it remains true that almost all small animal PET studies using mouse or rat as preclinical models are either limited by the spatial resolution or the sensitivity (especially for dynamic studies), or both, reducing the quantitative accuracy and quantitative precision of the results. Total-body small animal PET scanners, which have axial lengths longer than the nose-to-anus length of the mouse/rat and can provide high sensitivity across the entire body of mouse/rat, can realize new opportunities for small animal PET. This article aims to discuss the technical opportunities and challenges in developing total-body small animal PET scanners for mice and rats.

Inter-crystal scattering recovery of light-sharing PET detectors using convolutional neural networks

Inter-crystal scattering (ICS) is a type of Compton scattering of photons from one crystal to adjacent crystals and causes inaccurate assignment of the annihilation photon interaction position in positron emission tomography (PET). Because ICS frequently occurs in highly light-shared PET detectors, its recovery is crucial for the spatial resolution improvement. In this study, we propose two different convolutional neural networks (CNNs) for ICS recovery, exploiting the good pattern recognition ability of CNN techniques. Using the signal distribution of a photosensor array as input, one network estimates the energy deposition in each crystal (ICS-eNet) and another network chooses the first-interacted crystal (ICS-cNet). We performed GATE Monte Carlo simulations with optical photon tracking to test PET detectors comprising different crystal arrays (8 × 8 to 21 × 21) with lengths of 20 mm and the same photosensor array (3 mm 8 × 8 array) covering an area of 25.8 × 25.8 mm². For each detector design, we trained ICS-eNet and ICS-cNet and evaluated their respective performance. ICS-eNet accurately identified whether the events were ICS (accuracy > 90%) and selected interacted crystals (accuracy > 60%) with appropriate energy estimation performance (R² > 0.7) in the 8 × 8, 12 × 12, and 16 × 16 arrays. ICS-cNet also exhibited satisfactory performance, which was less dependent on the crystal-to-sensor ratio, with an accuracy enhancement that exceeds 10% in selecting the first-interacted crystal and a reduction in error distances compared when no recovery was applied. Both ICS-eNet and ICS-cNet exhibited consistent performances under various optical property settings of the crystals. For spatial resolution measurements in PET rings, both networks achieved significant enhancements particularly for highly pixelated arrays. We also discuss approaches for training the networks in an actual experimental setup. This proof-of-concept study demonstrated the feasibility of CNNs for ICS recovery in various light-sharing designs to efficiently improve the spatial resolution of PET in various applications.

Hybrid model of photon propagation based on the analytical and Monte Carlo methods for a dual-head PET system

The construction of photon propagation has a close relationship with the quality of reconstructed images. The classical Monte Carlo (MC) based method can model the photon propagation precisely, but it is time-consuming. The analytical method can often quickly construct a model, but its precision is a problem. How to fully exploit the advantages of the MC simulation and analytical model is an open problem. Inspired by the characteristics of the depth of interaction (DOI) detectors, which can help confirm the deposited position of a photon with DOI-encoding technology, we virtually discretize each crystal into several subcrystals to obtain the statistical distribution by MC-based simulation. Then, the statistical distribution is combined with a spatially variant solid-angle model. This combination strategy provides a hybrid model to describe photon propagation with relatively high accuracy and low computational cost. Three discretization schemes are compared to optimize the constructed photon propagation model. Several experiments are carried out to evaluate the performance of the proposed hybrid method. The metrics of full width at half maximum (FWHM), contrast recovery (CR), and coefficient of variation (COV) are adopted to quantitate the imaging results. The classical MC-based method is compared as a gold-standard reference. When a crystal is divided into two discretized positions, the convergent tendencies of CRs and COVs are consistent with that based on MC simulation method, respectively. In terms of FWHMs, the resolutions of using the MC-based model and the proposed hybrid model are 0.71 mm and 0.68 mm in the direction parallel to the detector head, respectively. This indicates the potential of the proposed method in positron emission tomography imaging.

High-Resolution Depth-Encoding PET Detector Module with Prismatoid Light-Guide Array

Depth-encoding detectors with single-ended readout provide a practical, cost-effective approach for constructing high-resolution and high-sensitivity PET scanners. However, the current iteration of such detectors uses a uniform glass light-guide to achieve depth encoding, resulting in nonuniform performance throughout the detector array due to suboptimal intercrystal light sharing. We introduce Prism-PET, a single-ended-readout PET detector module with a segmented light-guide composed of an array of prismatoids that introduce enhanced, deterministic light sharing. Methods: High-resolution PET detector modules were fabricated with single-ended readout of polished multicrystal lutetium yttrium orthosilicate scintillator arrays directly coupled 4-to-1 and 9-to-1 to arrays of 3 × 3 mm silicon photomultiplier pixels. Each scintillator array was coupled at the nonreadout side to a light-guide (one 4-to-1 module with a uniform glass light-guide, one 4-to-1 Prism-PET module, and one 9-to-1 Prism-PET module) to introduce intercrystal light sharing, which closely mimics the behavior of dual-ended readout, with the additional benefit of improved crystal identification. Flood histogram data were acquired using a 3-MBq ²²Na source to characterize crystal identification and energy resolution. Lead collimation was used to acquire data at specific depths to determine depth-of-interaction (DOI) resolution. Results: The flood histogram measurements showed excellent and uniform crystal separation throughout the Prism-PET modules, whereas the uniform glass light-guide module had performance degradation at the edges and corners. A DOI resolution of 5.0 mm full width at half maximum (FWHM) and an energy resolution of 13% FWHM were obtained in the uniform glass light-guide module. By comparison, the 4-to-1 coupled Prism-PET module achieved a DOI resolution of 2.5 mm FWHM and an energy resolution of 9% FWHM. Conclusion: PET scanners based on our Prism-PET modules with segmented prismatoid light-guide arrays can achieve high and uniform spatial resolution (9-to-1 coupling with ∼1-mm crystals), high sensitivity (20-mm-thick detectors and intercrystal Compton scatter recovery), good energy and timing resolutions (using polished crystals and after applying DOI correction), and compact size (depth encoding eliminates parallax error and permits smaller ring-diameter).