Leo H. C. Braga

A CMOS mini-SiPM detector with in-pixel data compression for PET applications

A new architecture for PET photodetectors based on small-area SiPMs (mini-SiPMs) with individual SPAD digitization and in-pixel data compression is presented. The main goal of this architecture is to reduce the electronics area occupation per SPAD and thus improve the fill factor. Two compression schemes are described, spatial and temporal compression, and a combination of both is implemented in the pixel. Our first chip using this architecture is described, where each pixel contains a mini-SiPM with 32 SPADs, a 4-bit digital counter and individual SPAD SRAMs for disabling high DCR devices. The sensor contains a 14 × 10 pixel array for a total of 4480 SPADs, and the achieved fill factor is about 29%. The sensor also contains column-level TDCs, which acquire the time of arrival of the first photon in each column. The expected compression response of this sensor is presented.

An 8×16-pixel 92kSPAD time-resolved sensor with on-pixel 64ps 12b TDC and 100MS/s real-time energy histogramming in 0.13µm CIS technology for PET/MRI applications

Positron-Emission Tomography (PET) is a nuclear imaging technique that provides functional 3-dimensional images of the body, finding its key applications in clinical oncology and brain-function analyses. The typical PET scanner is composed of a ring of scintillator crystals that absorb gamma rays and emit photons as a result, coupled to photon-sensing devices. The photons hit the sensors with a certain spread in space and time, depending on the material and geometry of the crystals. The sensors must then estimate the energy, the time of arrival (ToA), and the axial position of incoming gamma rays. Most commercially available scanners use photomultiplier tubes (PMTs), which are sensitive to magnetic fields, as the sensing element, making the integration of these systems with Magnetic-Resonance Imaging (MRI) impossible. A significant amount of research has focused on replacing PMTs with solid-state detectors, such as Silicon photomultipliers (SiPMs) [1], which can be integrated with MRI while maintaining the high-sensitivity of PMTs.

A time of arrival estimator based on multiple timestamps for digital PET detectors

Recent fully digital CMOS detectors suggested for PET can feature multiple on-chip TDCs. In these detectors, the timestamps of up to virtually all detected scintillation photons can be stored for later processing. However, the improvement in timing resolution that these multiple timestamps can provide strongly depends on the estimator (i.e. the post-processing algorithm) used. In this work, we propose an estimator that utilizes the timestamps of the first few photons to obtain a single time of arrival for the gamma event. The estimator is to be implemented in an FPGA for real-time coincidence detection, and thus aims at low computational complexity. The estimator performance was evaluated through Monte Carlo simulations of a scintillation with double exponential pulse shape and a photodetector with Gaussian jitter. The results of a LYSO scintillator coupled to a 250 ps jitter FWHM detector with 1000 detected photons per scintillation show that our estimator provides around 10% improvement over a single photon estimator.

SPADnet: A fully digital, networked approach to MRI compatible PET systems based on deep-submicron CMOS technology

This paper is the first comprehensive presentation of the SPADnet concept. SPADnet is a fully digital, networked MRI compatible time-of-flight PET system, exploiting the speed and integration density of deep-submicron CMOS technologies. The core enabling technologies of SPADnet are a sensor device comprising an array of 8×16 pixels, each composed of 4 mini-SiPMs with in situ time-to-digital conversion, a multi-ring network to filter, carry, and process data produced by the sensor devices at 2Gbps, and a 130nm CMOS process enabling mass-production of photonic modules that are optically interfaced to scintillator crystals. The SPADnet photonic modules comprise a matrix of tightly packed sensor devices; each module is networked in multiple rings, where coincidence pairs are identified and readily used in reconstruction algorithms, enabling scalable, MRI compatible pre-clinical PET systems for multi-modal imaging.

First characterization of the SPADnet sensor: a digital silicon photomultiplier for PET applications

Silicon Photomultipliers have the ability to replace photomultiplier tubes when used as light sensors in scintillation gamma-ray detectors. Their timing properties, compactness, and magnetic field compatibility make them interesting for use in Time-of-Flight Magnetic Resonance Imaging compatible Positron Emission Tomography. In this paper, we present a new fully digital Single Photon Avalanche Diode (SPAD) based detector fabricated in CMOS image sensor technol- ogy. It contains 16x8 pixels with a pitch of 610x571.2 mm 2 .

Complete characterization of SPADnet-I - a digital 8×16 SiPM array for PET applications

This paper presents the complete gamma characterization of the SPADnet-I sensor, a novel, fully digital SiPM array for PET applications. Each SPADnet-I pixel contains 720 SPADs and counting and timestamping logic, resulting in pixel dimensions of 0.6 × 0.6 mm2 and 42.6% fill-factor. Moreover, the sensor provides a real-time output of the total detected energy at up to 100 Msamples/s, which is internally used by a discriminator to distinguish gamma events from background noise. Results for the pile-up identification, crystal image and discriminator efficiency are presented. An energy resolution of 10.8% and a coincidence resolution time of 288 ps are reported.

Characterization of single-photon avalanche photodiodes in CMOS 150nm technology

The characterization of two Single-Photon Avalanche Diodes (SPADs) structures fabricated in CMOS 150nm technology is reported in this paper. The structures are based on a pwell/n-iso junction and differ only for the presence of a polysilicon layer above the guard ring. Each structure is implemented in two different shapes (circular and square) and four sizes (5,10,15 and 20μm). Measurement results show that both average breakdown voltage and non-uniformity decrease with SPAD sizes. The statistical variation of Photon Detection Efficiency (PDE) and its dependence on device size are also reported and discussed. For all the considered device sizes, a PDE non-uniformity lower than 0.5% was measured.

Characterization of Single-Photon Avalanche Diode arrays in 150nm CMOS technology

In this paper, a summary of characterization results from a SPAD test chip is reported. The chip includes test arrays based on two different SPAD structures, having p+/n-well and p-well/n-iso active area. Devices with different shapes and sizes as well as arrays dedicated to cross-talk extraction are present. Measurement results show that SPADs of the second type have a slightly lower Dark Count Rate. The peak Photon Detection Probability (PDP) is larger than 20% at 3V excess bias voltage for both structures. The average cross-talk probability between neighbors in SPAD arrays with 15.6μm pitch and 40% fill factor is less than 0.6% at 3V excess bias voltage for the first structure, while it is around 2.3% for the second one.

SPADnet: a fully digital, scalable, and networked photonic component for time-of-flight PET applications

The SPADnet FP7 European project is aimed at a new generation of fully digital, scalable and networked photonic components to enable large area image sensors, with primary target gamma-ray and coincidence detection in (Time-of- Flight) Positron Emission Tomography (PET). SPADnet relies on standard CMOS technology, therefore allowing for MRI compatibility. SPADnet innovates in several areas of PET systems, from optical coupling to single-photon sensor architectures, from intelligent ring networks to reconstruction algorithms. It is built around a natively digital, intelligent SPAD (Single-Photon Avalanche Diode)-based sensor device which comprises an array of 8×16 pixels, each composed of 4 mini-SiPMs with in situ time-to-digital conversion, a multi-ring network to filter, carry, and process data produced by the sensors at 2Gbps, and a 130nm CMOS process enabling mass-production of photonic modules that are optically interfaced to scintillator crystals. A few tens of sensor devices are tightly abutted on a single PCB to form a so-called sensor tile, thanks to TSV (Through Silicon Via) connections to their backside (replacing conventional wire bonding). The sensor tile is in turn interfaced to an FPGA-based PCB on its back. The resulting photonic module acts as an autonomous sensing and computing unit, individually detecting gamma photons as well as thermal and Compton events. It determines in real time basic information for each scintillation event, such as exact time of arrival, position and energy, and communicates it to its peers in the field of view. Coincidence detection does therefore occur directly in the ring itself, in a differed and distributed manner to ensure scalability. The selected true coincidence events are then collected by a snooper module, from which they are transferred to an external reconstruction computer using Gigabit Ethernet.

Characterizing single- and multiple-timestamp time of arrival estimators with digital SiPM PET detectors

In this paper, we make use of a recently proposed fully digital 8×16 SiPM array with per-pixel time-to-digital converters (TDCs), the SPADnet-I, to compare hardware-friendly, single- and multiple-timestamp time-of-arrival (ToA) estimators for PET. Two SPADnet-I sensors are used in coincidence, each coupled to a 3×3×5 mm3 LYSO crystal wrapped in Teflon. We acquired and processed 100k coincidence events to extract the coincidence resolving time (CRT) of the system. Pre-processing steps were applied to the data to remove low-energy events and correct the TDC values. Results show a 8% improvement in CRT of the suggested multiple timestamp estimator with respect to the best performing single-timestamp estimator (from 313 ps to 288 ps).