Search

Stationary single photon emission computed tomography (SPECT) systems offer numerous advantages over rotating-head systems, but have one notable drawback in that the array of relatively smaller cameras contains more edges and gaps than clinical two headed systems. Scintillation events occurring at the edges of traditional photomultiplier-tube (PMT)-based SPECT cameras lose spatial resolution due to loss of light-sampling. Simulations using customized non-sequential raytracing scripts to model and analyze mean detector response functions (MDRFs) showed significant improvement in spatial resolution for hybrid-sensor cameras employing both silicon photomultipliers (SiPMs) and PMTs. The results inform the hardware design of AdaptiSPECT-C: a stationary clinical whole-brain SPECT imager with adaptive apertures for selective dynamic or high spatial resolution imaging. Its modular hybrid cameras use SiPMs to augment the PMTs and improve spatial resolution for position estimation tasks. SiPMs, having a small pitch and efficient fill factor, are employed as a border around the edges of each detector area. PMTs, being low cost and reliable, are packed in the center. The front end electronics are split into two main boards: one to drive and provide signal conditioning for the PMTs, and the other performing a similar function for the SiPMs. Ultimately, 81 total signal channels leave each camera as negative voltage pulses. AdaptiSPECT-C will have two equatorial rings of 10 cameras each and a quasi-vertex ring of 4 cameras, totaling 24. Modularity is the guiding design principle for the mechanical components of the cameras and ensures ease of assembly and field service in the completed system.

Single-photon emission computed tomography (SPECT) can be used with a wide variety of radioligands for drug discovery and pharmacokinetic studies of promising drugs for neurodegenerative diseases. We are developing a human brain SPECT system with a stationary array of detectors that will provide dynamic high-resolution, high-sensitivity imaging. We are assessing the benefits of incorporating cylindrically curved scintillation detectors, which - due primarily to significant reduction in depth of interaction uncertainty - have resolution advantages over planar detectors at the edges. We are studying the use of a cylindrically curved to planar fiber optic plate to transfer the scintillation light from the curved crystal and light guide to a planar surface for photodetection using conventional methods. Another design component being evaluated is a novel light-sensor configuration combining photomultiplier tubes (PMTs) and silicon photomultipliers (SiPMs). Simulation methods were used to predict performance of a variety of detector layouts. The purpose of the study was to balance the tradeoff between detector cost and performance, as the final imager will be comprised of 24 camera modules. We demonstrate that combining PMTs and SiPMs for electronic readout achieves a spatial resolution advantage at the edges while maintaining a lower cost than a full SiPM readout or a curved detector.

An adaptive-stationary-modular multi-pinhole (MPH) brain SPECT, AdaptiSPECT-C is being developed by the University of Arizona and University of Massachusetts Medical School to meet static and dynamic brain SPECT imaging needs. Salient features of the ASC include the use of adjustable pinhole apertures to dynamically adapt to imaging task needs, improved light measuring around the edge of the scintillator crystal, plus motion tracking and correction with attenuation correction enabled by usage of depth-sensing (DS)-cameras. For a target system spatial resolution of 8 mm at the focal point of the apertures, selected to enable comparison to current 2-headed commercial (2HC) SPECT imaging, we report investigation of aperture layout designs for a system with 3 rings of 18.4 cm flat square detector modules. We investigated sensitivity at the focal point in comparison to 2HC for usage of 1 versus 5 apertures per module, and variation in the extent of truncation and multiplexing of the irradiation fields by adjustment of the aperture location between the detector and focal point. For a system with one aperture per module and minor truncation we determined a sensitivity of 2.7x that of 2HC; whereas, with use of 4 oblique apertures with minor truncation and moderate multiplexing we determined the sensitivity was 4.6x, and with all 5 apertures resulting in significant multiplexing the sensitivity was 5.7x. We also determined through simulation better visualization of the rods of a Derenzo phantom, and perfusion distribution of XCAT brain phantom with the 5 pinhole design, using solely the 4 oblique pinholes. We thus believe that this design with 5 pinholes per detector module is an excellent candidate for use in construction of the AdaptiSPECT-C system.

We are designing AdaptiSPECT-C, a novel, multiple-detector, adaptive-aperture single-photon emission, tomographic (SPECT) system dedicated to brain imaging. This system is designed to change sensitivity and spatial resolution in realtime to address the needs of dynamic imaging[1]. The aim of this work is to document the creation of a manufacturable aperture design including hardware, electronics and software that effectively adapt in real-time spatial resolution and sensitivity functions to the needs of a dynamically changing subject. We accomplish these goals through metal printing of apertures and with custom control boards based on the Arduino microcontroller. With this design we are able to precisely control each aperture motion with a step resolution of 0.20 millimeters, which is within our required tolerances.

A next-generation multi-pinhole system dedicated to brain SPECT imaging is being constructed by our research team, which we call AdaptiSPECT-C. The system prototype used herein consists of 25 square detector modules and a total of 100 apertures grouped by 4 per module. The system is specifically designed for multi-purpose brain imaging and capable of adapting in real-time each aperture size and whether it is open or shuttered closed. The use of such system would provide optimum high-performance patient-personalized imaging for a wide range of brain imaging tasks. In this work we investigated the effect of pinhole diameter variation on spherical tumor quantification for the promising brain tumor imaging agent 123I-CLINDE. To assess the quality of the images reconstructed for the different aperture sizes, we used a customized multiple-sphere tumor phantom derived from the XCAT software with a tumor size of 1 cm in diameter. Our results suggest through quantification and visual inspection that an aperture diameter in the range of 2 to 5 mm in diameter for the adaptive AdaptiSPECT-C system is likely the most suited for high performance brain tumor 123I-CLINDE imaging. In addition, our study concludes that a 4 mm pinhole diameter given its excellent spatial-resolution-to-sensitivity trade-off is promising for scout acquisition in localizing target tumor regions within the brain. We have initiated a task-based performance on the tumor detection and localization accuracy for a range of simulated tumor sizes using the channelized non-pre-whitening (CNPW) matched-filter scanning-observer.

AdaptiSPECT-C is a dedicated human brain SPECT system currently under development. The system utilizes hybrid PMT/SiPM modular gamma-ray scintillation cameras with 81 photosensor channels for each camera. Here we present the read-out system design for the modular cameras. A front-end board inside the camera enclosure performs the current-to-voltage conversion for the photosensor signals. The back-end board receives the 81 analog signals and estimates time of interaction and amount of light acquired by each photosensor. The energy estimation is performed by sigma-delta modulation (SDM). A non-uniform 2-bit SDM is utilized for triggering and timing. The SDM approach utilized in the front-end board reduces the complexity of the system compared to conventional ADC methods, while maintaining system performance. The digital side of the SDMs is implemented on the FPGA of a system-on-chip, which also contains an ARM-based processor. A direct memory access (DMA) controller implemented on the FPGA writes the data to on-board memory. An embedded lightweight TCP/IP stack reads the data from memory and streams it to the system computer. The back-end board also monitors and controls camera temperature, and supplies power to the camera.

Patient motion and its deteriorating effects in medical imaging is well known. Likewise, head rigid-body motion degrades the image quality in brain SPECT. We developed an algorithm to compensate the head motion in multi-pinhole SPECT systems within a statistical iterative image reconstruction algorithm. Previously, volunteer’s head motion was recorded by Vicon MX visual tracking system for 10 minutes while laying inside a SPECT/CT gantry. We then divided the motion into 120 intervals, each 5 seconds long. AdaptiSPECT-C, a multi-pinhole multi-detector stationary SPECT system, we are developing for dedicated brain imaging was used for this study. We generated an XCAT voxelized brain phantom emulating the activity distribution of Iodine-123 N-isopropyl-4-iodoamphetamine (IMP) for brain perfusion scan. To simulate the data acquisition with head motion, we used generic analytic simulation software we developed for multi-pinhole SPECT systems. The 6-degrees-offreedom (6-DOF) motion was incorporated into the simulation software to realistically simulate the data acquisition with motion. Our previously developed graphics-processing-unit (GPU)-based iterative reconstruction software was augmented to incorporate motion compensation using 3D Gaussian interpolation. The rigidbody (i.e. 6-DOF) head motion was input to the reconstruction software through 120 motion intervals. For comparison, we reconstructed the motion corrupted SPECT data without motion compensation and a motion-free acquisition as ground truth. The results show that our proposed motion compensation method provides a significantly better SPECT reconstruction when compared to no motion compensation. The developed software can be applied for any scan duration with any number of motion intervals.

AdaptiSPECT-C (ASC) is being designed as an adaptive multi-pinhole SPECT imaging system with multiple detectors arranged in sphere-like geometry in three rings (Caudal, Middle, and Quasi-Vertex (QV)). Herein we investigate the usage of rotated square detectors for the first two rings, and hexagonal detectors for QV ring. We compare designs for different levels of temporal shuttering of the multiple apertures irradiating each detector. We assess image quality for a variety of phantoms assessing uniformity, axial sampling, and brain perfusion imaging. We determine that rotated square detectors can provide good uniformity, the axial sampling we have observed thus far based on reconstruction of a tailored Defrise Phantom and reconstructions of the brain phantom modeling perfusion distribution.

Machine learning (ML) is increasingly used in medical imaging. This paper provides pilot data of decision trees and random forests (RFs) to predict if a 18F-florbetapir brain positron emission tomography (PET) is positive or negative for amyloid deposition based on quantitative data analysis. The dataset included 55 18F-florbetapir brain PETs in participants with severe white matter disease and mild cognitive impairment (MCI), early Alzheimer's disease (AD) or transient ischemic events. The Montreal Cognitive Assessment (MoCA) score was known for each participant. All PET images were processed using the MINC toolkit to extract standardized uptake value ratios (SUVRs) for 59 regions of interest (features). Each PET was clinically read by 2 dual certified radiology/nuclear medicine physicians with final interpretation based on consensus. An initial study of RFs using conventional binary decision trees and PET quantitation suggests this is a powerful algorithm for PET classification as positive or negative for amyloid deposition. Preliminary data did not show improved results when a ternary RF approach was used. Finally, a soft-decision approach may be helpful to predict the $\mathbf{MoCA}$ score.