Pith. sign in

REVIEW 2 major objections 4 minor 1 references

108 ps coincidence time resolution through optimized scintillators, photodetectors, readout electronics, and DOI-based timing correction in orthogonally stacked detector configurations

T0 review · 2 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read Orthogonally stacked scintillator detectors reach 108.6 ps coincidence timing in TOF-PET.

desk verdict Credible component-optimization result, but the 108 ps headline depends on an unrealized DOI estimator; the realistic number is 110.7 ps. read the letter →

arxiv 2608.06746 v1 pith:DNG4HVKZ submitted 2026-08-07 physics.med-ph physics.ins-det

classification physics.med-phphysics.ins-det PACS 87.57.uk
keywords positronemissiontomographytime-of-flightPETcoincidencetimeresolutiondepthofinteractionxDetectorsiliconphotomultiplierLYSOscintillatorDOI-basedtimingcorrection
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that a PET detector geometry called xDetector can overcome the usual trade-off between timing resolution and detection efficiency. Two groups of thin scintillator crystals are stacked orthogonally and read by separate silicon photomultipliers, keeping the module thick enough for efficient 511 keV gamma detection while reducing photon transport time spread and encoding interaction depth in the light-sharing pattern. After switching to faster crystals, higher-photon-detection-efficiency SiPMs, high-frequency readout electronics, and applying depth-of-interaction timing correction, the authors measured 105.3 ± 0.6 ps FWHM in coincidence with a reference detector and calculated 108.6 ± 1.9 ps FWHM for two xDetectors facing each other. That is an average 10.3% improvement over a conventional 20.0 mm single-ended LYSO detector and brings system-level time-of-flight PET close to the 100 ps FWHM target.

What carries the argument

The mechanism is the xDetector: a module in which two groups of three thin LYSO scintillator crystals, each group coupled to its own SiPM, are stacked orthogonally with their bare sides facing each other through an air gap, so that interaction depth is encoded in how scintillation light is shared between channels. The paper's timing correction then shifts the coincidence-time histogram for each depth position to a common reference before fitting, removing the position-dependent optical path length variation (photon transport time spread) that otherwise broadens the CTR. Supporting that correction are the component choices: faster 36.4 ns decay-time LYSO crystals, SiPMs with roughly 10% higher PDE at 420 nm, and high-frequency readout built on two cascaded BGA616 amplifiers.

What would settle it

Measure the xDetector's CTR under uniform irradiation with interaction depth estimated internally from the channel light-sharing ratios instead of set by manual translation. If the internally estimated DOI has about 6 mm resolution, the paired CTR should land near 110.7 ± 2.0 ps; a measured paired CTR above about 111 ps would indicate that the 108.6 ps four-position result is not achievable as a system-level value.

Watch

Extended reading notes

Core claim

The central discovery is that the timing penalty of long scintillators can be removed without sacrificing detection efficiency by replacing a single 20 mm crystal with two orthogonally stacked detector groups read out at their ends, using the light-sharing pattern to sense depth of interaction. With optimized components, the best single-channel CTR of the xDetector in coincidence with a reference detector was 107.8 ± 1.2 ps FWHM at 63 V bias. Applying DOI-based timing correction over four positions spaced 3.15 mm along the longitudinal axis improved the measured CTR to 105.3 ± 0.6 ps FWHM; error propagation for two xDetectors in coincidence gives 108.6 ± 1.9 ps FWHM, versus 113.5 ± 2.7 ps without correction. The authors conclude that the xDetector offers a realistic route toward the 100 ps FWHM system-level CTR goal, with the remaining gap coming mainly from finite longitudinal position resolution.

Load-bearing premise

The load-bearing assumption is that interaction depth is known to about 3.15 mm in real operation: in the experiment the xDetector was manually slid to set the interaction position and the time shifts were derived from Gaussian fits to the same data, while the detector's intrinsic longitudinal resolution is about 6 mm, and recalculating with only two positions raises the paired CTR to 110.7 ± 2.0 ps.

Editorial extensions

If this is right

  • The paired xDetector CTR of 108.6 ± 1.9 ps FWHM places a detector built from clinical-thickness scintillators within reach of the 100 ps system-level TOF-PET target.
  • Because the xDetector keeps scintillator thickness and detection efficiency while improving CTR by about 10% over a conventional 20.0 mm single-ended detector, it offers an alternative to side- or dual-ended readout that uses fewer SiPMs.
  • The SNR gain of TOF-PET scales roughly as the inverse square root of CTR, so moving from about 200 ps to about 108 ps would translate into a meaningful improvement in image signal-to-noise ratio, shorter scan times, or reduced dose.
  • The two-position DOI correction result of 110.7 ± 2.0 ps paired CTR shows that improving longitudinal resolution alone should push the concept below 108 ps and closer to 100 ps.
  • The authors identify packing fraction, assembly reproducibility, and the number of SiPMs as practical issues for translating the measured module CTR into a full PET ring.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: since the four-position correction was applied with the interaction position set manually rather than estimated by the detector, the headline 108.6 ps paired CTR should be read as the upper bound of what is achievable; with the detector's own roughly 6 mm longitudinal resolution, the realistic paired CTR is closer to the two-position value of 110.7 ± 2.0 ps.
  • Editorial inference: combining the xDetector geometry with further DOI-resolution improvements, such as finer channel segmentation or signal-shape-based depth estimation, is the most direct path below 100 ps, because the paper's own analysis shows the residual spread is dominated by position resolution rather than by single-photon timing.
  • Editorial inference: a testable prediction is that an xDetector with continuously estimated DOI from channel ratios will land between 108.6 and 110.7 ps paired CTR; if it lands above about 111 ps, the manual-position result overstates the achievable system performance.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 4 minor

Summary. The manuscript reports an experimental study of a TOF-PET detector module called xDetector, which consists of orthogonally stacked groups of thin LYSO scintillators read out by SiPMs. The authors optimize the scintillator decay time, SiPM PDE, and readout electronics, and then apply a depth-of-interaction (DOI) based timing correction. Measured in coincidence with a reference detector, the best corrected CTR is 105.3 ± 0.6 ps FWHM, and by quadrature error propagation the authors estimate a paired-xDetector CTR of 108.6 ± 1.9 ps FWHM. They also compare against conventional single detectors with 12.6 mm and 20.0 mm scintillators and report an average CTR improvement of 10.3% over the 20 mm device. The abstract claims that the xDetector offers potential to reach a 100 ps system-level CTR.

Significance. If the raw measurements are taken at face value, the work provides a credible, well-characterized demonstration of a TOF-PET detector with sub-110 ps CTR while retaining a 20-mm-equivalent detection thickness, which is a meaningful step toward the 100 ps TOF-PET goal. The strengths of the paper include direct measurements of the component improvements (faster LYSO, improved SiPM PDE, high-frequency readout), explicit Gaussian fitting with reported uncertainties, and a direct comparison against conventional detectors under the same conditions. The key limitation is that the DOI-based timing correction is performed with externally imposed interaction positions rather than with a signal-derived DOI estimator, and the paper itself acknowledges that the demonstrated longitudinal resolution of the xDetector is only ~6 mm. With that realistic resolution, the corrected CTR degrades to 110.7 ± 2.0 ps. The central claim should therefore be reframed so that the headline value reflects the achievable system projection, not the idealized 3.15 mm DOI case.

major comments (2)
  1. [Abstract and Section 3.2 (Figure 8)] The headline result of 108.6 ± 1.9 ps paired CTR is presented in the abstract and conclusion as an achieved system-level value, but it is conditional on knowing the interaction DOI to 3.15 mm. As stated in Section 2.3, the interaction position was controlled by manually sliding the xDetector rather than estimated from detector signals. Section 4 then acknowledges that the demonstrated longitudinal resolution of the xDetector is ~6 mm and reports that with two positions the corrected CTR becomes 110.7 ± 2.0 ps. The 108.6 ps value is therefore an idealized upper-bound projection, not a realistic system CTR. The abstract and conclusions should present the two-position value (110.7 ± 2.0 ps) as the primary projection, or clearly label the 108.6 ps value as conditional on an as-yet-undemonstrated DOI estimator.
  2. [Section 2.3] The DOI-based timing correction is an in-sample calibration rather than a predictive correction. For each position, the time-difference histogram is shifted by the mean value obtained from a Gaussian fit to that same histogram; the shifted histograms are then superimposed and re-fitted. This procedure removes position-dependent mean offsets by construction and does not validate the correction on independent data or against DOI values estimated from the detector signals. The small observed improvement (~2-3 ps) should be interpreted with this circularity in mind. I recommend either demonstrating the correction with a signal-based DOI estimate or reporting the uncorrected and realistic-two-position results as the central evidence.
minor comments (4)
  1. [References] In the reference for Pagano et al. 2024, the author name appears as 'Cate J W' but should be 'Cates J W' to match the other citations in the text.
  2. [Introduction] The citation 'van Suluis et al. 2019' contains a typo; the correct name is 'van Sluis' as in the reference list.
  3. [Section 2.3] The text states that the shifted histograms are referenced to 'an arbitrary reference value.' While this does not affect the FWHM, it would be clearer to specify that only the relative positions of the shifts matter, and that the fitted FWHM is invariant under a constant offset.
  4. [Section 3.1] The report of 'the average CTR of the xDetector over all channels was 110.0 ps FWHM' would be clearer if the standard deviation or range across channels were also given, since Figure 6 presumably contains that information.

Circularity Check

1 steps flagged · score 6.0 of 10

The 108.6 ps DOI-corrected CTR is an in-sample alignment: each position histogram is shifted by its own fitted mean, so the correction removes inter-position shifts by construction; the realistic two-position value is 110.7 ps.

  1. fitted input called prediction [Section 2.3 (DOI-based timing correction); result reported in Section 3.2 and Figure 8]
    "the interaction position was controlled by manually moving the xDetector instead of estimating it from the signals of channels aligned orthogonally to the longitudinal axis. ... time difference histograms corresponding to each position were shifted to an arbitrary reference value based on the mean value obtained from each Gaussian fitting and superimposed."

    The corrected CTR is produced by taking each measured position histogram, fitting its mean, and shifting that same histogram by its own fitted mean before combining. Hence the inter-position peak shifts that DOI correction is meant to remove are removed exactly, by construction; the reported improvement (~3 ps) is the guaranteed variance reduction from aligning group means, not a demonstration that an estimated DOI can predict the correction for unseen events. Since the experiment manually controlled the interaction position, the 105.3/108.6 ps values are in-sample calibrations conditional on 3.15 mm position bins, not achieved system CTRs with a real DOI estimator.

full rationale

The only load-bearing circularity is the DOI-based timing correction. Section 2.3 states that the interaction position was manually set rather than estimated from detector signals, and that each position's time-difference histogram was shifted by the mean obtained from a Gaussian fit to that same histogram before superposition and re-fitting. The corrected FWHM is therefore the pooled within-position width after the between-position means have been removed by construction; the ~3 ps gain from 'DOI-based timing correction' is a mathematical consequence of aligning group means, not a validated prediction from an estimated DOI. This makes the headline 108.6 ± 1.9 ps paired-xDetector CTR an in-sample calibration result. The paper partially mitigates the concern by reporting that the demonstrated longitudinal resolution is ~6 mm and that using two positions gives 110.7 ± 2.0 ps paired, but the abstract and title still lead with the 108.6 ps four-position value. The error-propagation step that converts measured single-xDetector CTRs to paired-xDetector CTRs is standard and not circular, and the self-citation to Onishi and Ota (2025) for the xDetector concept and ~6 mm longitudinal resolution is background evidence, not a load-bearing circular argument.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The central claim rests on measurements rather than on a formal derivation. The main extra assumptions are standard quadrature CTR combination and the equivalence of manual sliding to real DOI information. No new particles or entities are introduced.

free parameters (4)
  • Timing pick-off threshold = not reported (optimized)
    The detection timing threshold was optimized to obtain the best CTR (Section 2.2, Figure 6). This is a data-dependent tuning parameter.
  • Bias voltage operating point = 63 V (overvoltage 12 V)
    CTR was measured at 62, 63, and 64 V; the headline results use 63 V, the best voltage. This is a selected operating point from a small scan.
  • Per-position DOI time shifts = four values, not tabulated
    In Section 2.3, each position histogram is shifted by the mean value obtained from its Gaussian fit. These shifts are fitted to the same data used to evaluate the corrected CTR.
  • Energy threshold = ~420 keV
    Set at the valley between the 511 keV photopeak and Compton region; a standard but manually chosen analysis setting.
assumptions (3)
  • domain assumption CTR variances add in quadrature and the reference detector's paired CTR can be used to extract the xDetector's single-detector CTR.
    Used in Sections 3.1 and 3.2 to convert the measured x-reference CTR to a paired xDetector CTR via sqrt(2*m^2 - r_pair^2). Standard in TOF-PET.
  • domain assumption For the uncorrected CTR measurement, the 22Na source irradiates uniformly along the full scintillator length.
    Section 2.2 states the source is placed at sufficient distance to ensure uniform gamma-ray irradiation; this makes the position-averaged CTR representative.
  • domain assumption Manually sliding the xDetector reproduces the same time behavior as interactions at those depths.
    Section 2.3 replaces DOI estimation with external position control; this assumes the optical path variation is equivalent to real DOI.

how reviews work

0 comments
Cite this review

Pith. "Pith review of 108 ps coincidence time resolution through optimized scintillators, photodetectors, readout electronics, and DOI-based timing correction in orthogonally stacked detector configurations." pith.science (2026). https://pith.science/paper/DNG4HVKZ

@misc{pith2026260806746,
  author       = {Pith},
  title        = {Pith review of: 108 ps coincidence time resolution through optimized scintillators, photodetectors, readout electronics, and DOI-based timing correction in orthogonally stacked detector configurations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DNG4HVKZ}},
  note         = {Machine review of arXiv:2608.06746}
}
abstract

Objective. Existing commercial time-of-flight positron emission tomography (TOF-PET) systems yield a coincidence time resolution (CTR) of ~200 ps or less full width at half maximum (FWHM). Recently, there has been a challenge to achieve a CTR of 100 ps FWHM at the system level. However, current silicon photomultipliers (SiPMs) and 20-mm-thick scintillators in conventional single-ended readout scheme is difficult to achieve 100 ps CTR; the photon transport time spread (PTS) within the scintillator crystal is a major barrier. Differences in the interaction position result in variations in PTS on the order of several tens of ps, thereby degrading the CTR. A shorter scintillator can improve CTR; however, this can degrade detection efficiency. Approach. To overcome this trade-off between the CTR and detection efficiency, we previously proposed xDetector, an orthogonally stacked configuration along the longitudinal axis of scintillator crystals. We investigated the CTR potential of the xDetector by improving the scintillator, photodetector, and readout electronics, and by applying CTR correction based on a three-dimensional interaction within the scintillator. Main results. Based on error propagation, the CTR of the paired xDetector was calculated as 113.5 $\pm$ 2.7 ps FWHM. Furthermore, the CTR of the xDetector was measured at four positions along the longitudinal axis by manually sliding the xDetector, and the corrected achieved CTR was 108.6 $\pm$ 1.9 ps FWHM. Moreover, compared with the conventional single detector using a 20.0 mm scintillator, CTR improved by an average of 10.3%. Significance. The xDetector offers potential as a PET detector concept to achieve a CTR of 100 ps FWHM. Such timing performance is expected to improve TOF-PET image quality and quantitative accuracy, contributing to more reliable disease detection and diagnosis than current PET detectors.

Figures

Figures reproduced from arXiv: 2608.06746 by the authors.

Figure 1
Figure 1. Schematic of the xDetector: (a) A detector consisting of LYSO crystals coupled to an SiPM using Meltmount, with the four sides covered by ESR. (b) A detector group formed by aligning three detectors along the short axis. (c) xDetector formed by orthogonally stacking two detector groups with their bare sides facing each other via air coupling [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 5
Figure 5. In this measurement, owing to the limited number of input channels of the oscilloscope, the interaction position was controlled by manually moving the xDetector instead of estimating it from the signals of channels aligned orthogonally to the longitudinal axis. The measurements were conducted using the Ch1 of the xDetector across four positions separated by 3.15 mm. For the sake of convenience, these positions were … view at source ↗
Figure 8
Figure 8. Based on error [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

1 extracted references

  1. [1]

    Cates J W, Vinke R, and Levin C S 2015 Analytical calculation of the lower bound on timing resolution for PET scintillation detectors comprising high-aspect-ratio crystal elements Phys. Med. Biol. 60 5141 (doi: 10.1088/0031-9155/60/13/5141) Cates J W and Levin C S 2018 Evaluation of a clinical TOF-PET detector design that achieves ⩽ 100 ps coincidence tim...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.