Pith. sign in

REVIEW 4 major objections 6 minor 21 references

Position-Sensitive Silicon Photomultiplier Array with Enhanced Position Reconstruction by means of a Deep Neural Network

T0 review · 4 major / 6 minor · reviewed 2026-08-03 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A neural network trained on six channel charges from a position-sensitive silicon photomultiplier tile reconstructs light-spot positions with roughly 0.2 mm granularity, a 2.4–3.5x improvement over the standard linear formula.

desk verdict Solid DNN calibration study, but the granularity and resolution claims outrun what the experiment actually measures. read the letter →

arxiv 2512.02771 v2 pith:IFDFBZAF submitted 2025-12-02 physics.ins-det hep-exphysics.data-an

Cyril Alispach , Fabio Acerbi , Hossein Arabi , Domenico della Volpe , Alberto Gola , Aramis Raiola , Habib Zaidi This is my paper · ORCID
classification physics.ins-dethep-exphysics.data-an
keywords SiPMlinearly-gradedposition-sensitivedetectordeepneuralnetworkpositionreconstructiongammacamerachargesharingsub-millimeterresolution
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

This paper sets out to show that a deep neural network can replace the standard linear readout formula for a 2x2 array of linearly-graded silicon photomultipliers (LG-SiPMs) and recover nearly distortion-free position maps. The motivation is practical: such detectors read out a 16x16 mm2 imaging tile with only six channels, and if the DNN approach holds, compact gamma cameras could achieve sub-millimeter position resolution without dense pixelation. The authors report that the DNN reduces the average position error from about 0.68 mm to about 0.20 mm, mainly by removing a systematic shift while keeping the statistical noise nearly unchanged. They argue this makes the effective pixel count of the tile an order of magnitude larger, which matters for medical imaging applications such as scintillation cameras.

What carries the argument

The central mechanism is the trained deep neural network mapping from six channel amplitudes to planar coordinates. The network has a 6-unit input layer (charge amplitudes normalized by total charge), one or more hidden layers of 64 units with hyperbolic tangent activations, and a two-unit linear output layer for the reconstructed coordinates. It is trained by minimizing the mean squared error between reconstructed and motor-stage positions. The network's role is to learn and compensate the nonlinear, position-dependent distortions of the LG-SiPM tile, including inter-tile gain variations and charge-sharing asymmetries, which the hand-derived linear formula (a center-of-gravity calculation w

What would settle it

A collimated gamma-ray source scanned across the full tile, including the gaps between the four chips, would directly test transferability: if the DNN's mean position error near edges and gaps exceeds the linear formula's, or if retraining on scintillation light fails to reproduce the 0.2 mm granularity, the central claim fails.

Watch

Extended reading notes

Core claim

The paper claims that a deep neural network taking the six charge amplitudes from a 2x2 array of linearly-graded SiPMs (normalized by total charge) as input can reconstruct the coordinates of a light spot on the 16x16 mm2 tile with an average granularity of about 0.2 mm, compared with about 0.68 mm for the standard linear center-of-gravity formula. The improvement comes almost entirely from correcting a systematic spatial shift, which drops from roughly 317 µm to 41–93 µm depending on the train/test split, while the random noise component (resolution) stays nearly unchanged at 66–68 µm versus 78–79 µm for the linear model. Across three different splitting strategies — random, chessboard, and

Load-bearing premise

The measured improvement applies to small LED spots on interior grid positions after a total-charge cut; the claim that the same accuracy holds for the broader light distributions, edges, and gaps encountered with real gamma-ray scintillation events is assumed, not demonstrated.

Editorial extensions

If this is right

  • If the result transfers to gamma scintillation detection, a compact handheld gamma camera could achieve sub-millimeter intrinsic resolution with only six readout channels per 16x16 mm2 tile.
  • Because the DNN mainly removes systematic shifts, detector noise remains the fundamental resolution limit; improving the sensor itself would further push the achievable granularity.
  • The same training procedure can be applied to larger arrays or other position-sensitive detector geometries, replacing hand-tuned linear gain matrices with a learned calibration.
  • The demonstrated 0.2 mm granularity is finer than the pixel pitch of many scintillator arrays, suggesting the detector, not the crystal, could set the imaging resolution in future systems.

Reading between the lines

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

  • A natural next test is a gamma-irradiated scintillator read out by the same tile; if the DNN trained on LED spots does not transfer to the broader scintillation light profile, a domain-adapted training set will be needed — this is not addressed in the paper.
  • The total-charge cut used to remove edge and inter-tile-gap events implies the reported 6400 regions is an upper bound for the interior field of view; a real camera would need a strategy to handle or reject events near gaps, which the paper does not specify.
  • The paper's zero-hidden-layer model already shows that allowing a full 6x2 weight matrix (14 free parameters) captures a large fraction of the linear correction; a closed-form least-squares fit might recover much of the gain without a deep network.
  • The stability of the learned mapping over temperature, bias voltage, and count rate is untested; if it drifts, periodic recalibration would be required in the field.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper characterizes a 2x2 array of linearly-graded SiPMs read out through six channels and compares two position reconstruction methods: the standard linear formula of Section 2.2 (Eqs. 1–4) and a fully connected DNN with 6 inputs, 64-unit tanh hidden layers, and a 2-unit linear output (Section 2.3.2). The DNN is trained on 10k-event LED scans on a 37x37 grid with 0.5 mm steps, using the mean squared error to motor positions as the loss. Three train/test splittings are investigated (random, chessboard, and random-position, Section 2.1). Table 1 reports that the DNN reduces the mean systematic shift from ~310–317 µm to 41–93 µm and the quantity 2⟨d⟩ from ~675–688 µm to 198–284 µm, while the mean statistical resolution σ remains similar (~66–79 µm). The conclusion claims a granularity of about 0.2 mm and an increase in distinguishable regions by a factor 10.24.

Significance. If the central claim holds, the work is significant for compact gamma cameras and other position-sensitive photon detectors: a neural calibration would remove the geometric and gain nonlinearities of a 6-channel LG-SiPM tile, achieving sub-millimeter positioning without increasing channel count. The paper has real strengths: it uses three data splittings, including B and C which exclude training grid positions and therefore provide genuine generalization evidence; it reports quantitative per-position metrics from Rice fits; and it compares directly against the standard linear reconstruction. The main caveat is that the headline 'granularity' / 'pixel factor' claim rests on an indirect metric that has not been validated as a two-point resolution measure.

major comments (4)
  1. [§3.4 and Table 1] The 'granularity' quantity 2⟨d⟩ is not a two-point resolution. ⟨d⟩ in Eq. (7) is the mean Euclidean distance between reconstructed and motor positions for isolated LED spots on a 0.5 mm grid. Multiplying by two and converting to '6400 distinguishable regions' over the 16x16 mm2 tile (Section 3.4, conclusion) assumes that a calibration error of ⟨d⟩ implies resolvability of two spots separated by 2⟨d⟩, but no two-spot or sub-grid measurement is performed. The improvement in 2⟨d⟩ is dominated by the reduction of the systematic shift ν (317→41–93 µm), not of σ (78→66–68 µm), so it primarily demonstrates that a smooth distortion map is learned. Please either provide a direct two-point resolution measurement (e.g., two sources or sub-0.5-mm stepping) or rephrase the claims in terms of position reconstruction accuracy rather than resolved pixels.
  2. [§2.1, Fig. 3, and Table 1] The total-charge quality cut in Figure 3 removes fiber positions near the edges and inter-tile gaps, and the table caption states that the outer rows and columns are excluded from the averages. The abstract and conclusion nevertheless quote '16x16 mm2' as the field of view for the 6400-region count. The tested sensitive area is smaller than the nominal window. Please report the effective area used after the cuts and avoid extrapolating pixel counts to the full 16x16 mm2 unless the edge/gap regions are included in the measurements.
  3. [§3, Eq. (7), and Section 2.3.2] The evaluation metric d is the same mean squared error used as the DNN loss, and in splitting A the test events share exact motor coordinates with training. This makes gains in A partly a measure of memorization of per-position output biases. Splittings B and C, which use held-out grid positions, provide the true generalization evidence and still show improvements (2⟨d⟩ = 228.7 µm and 283.6 µm vs 674.8–687.6 µm for the linear model). The abstract's upper factor of 12.1 comes from A; the paper should present B and C as the primary evidence for generalization, or explicitly weight the splittings when summarizing the improvement range.
  4. [Conclusion vs. Abstract and Table 1] The conclusion states that the DNN 'reduces the granularity to about 0.2 mm thus increasing the number of distinguishable regions by a factor 10.24.' This number does not match Table 1 or the abstract's range. From Table 1, the squared ratios 2⟨d⟩_lin / 2⟨d⟩_DNN are about 12.1 (A), 9.0 (B), and 5.7 (C). The abstract says 5.7 to 12.1, but the conclusion's 10.24 is unexplained. Please reconcile these numbers or specify which splitting (or average) the conclusion refers to.
minor comments (6)
  1. [§2.3] Typo: 'The the performance' should be 'The performance'.
  2. [Fig. 4 caption] 'DDNs' should be 'DNNs' for consistency with the text.
  3. [§2.3.2] In the text, 'a system of to 2 linear equations' should be 'a system of two linear equations.'
  4. [Eqs. (3)–(6)] The linear transformation parameters lx, ly, x0, y0, φ are introduced in (3)–(4) and reused in the matrix A of (6), but the relation between (x, y) in the relative frame and the motor positions is not fully stated. An explicit definition of the relative coordinates would improve readability.
  5. [Fig. 7 caption] The caption says 'run splitting' for (C); this should be 'random position splitting' or 'random splitting' as in the text.
  6. [Data/code availability] The paper does not mention whether the acquisition code, trained models, or dataset are available. Please add a data/code availability statement, or note that they are available from the authors on request.

Circularity Check

2 steps flagged · score 4.0 of 10

Granularity gain is the training objective restated; split-A headline value is partly memorization, while held-out splittings give partial independent support.

  1. self definitional [Section 2.3.2 (loss), Section 3.3/Eq. 7, Section 3.4, Table 1]
    "The loss function was set as the mean squared error between the reconstructed positions (xreco, yreco) and motor positions (xmotor, ymotor). ... This quantity estimates the average size of a distinguishable region in the sensor. It is computed as twice the mean of the distance distribution 2⟨d⟩ which corresponds to when two neighboring regions are overlapping and thus are indistinguishable."

    The headline metric 'granularity' is 2⟨d⟩, where d (Eq. 7) is the Euclidean distance between reconstructed and motor positions. The DNN is trained to minimize exactly the squared version of this same distance. Therefore the reported improvement in granularity, and the derived factor of 5.7-12.1 increase in 'resolved areas' (area/(2⟨d⟩)^2), is a restatement of the training objective rather than an independent measurement of two-point resolution. Splittings B and C show that the reduction generalizes to held-out grid positions, so the circularity is partial: the gain is real interpolation, but the quoted granularity is defined as the optimized error.

  2. fitted input called prediction [Section 2.1 (splitting A), abstract, Table 1]
    "We split the data set into two samples of the same size ... (A) a random split across data set; ... the DNN-based reconstruction boosts the number of resolved areas (‘pixels’) by a factor of at least 5.7"

    In splitting A the test sample is drawn randomly from the same motor positions used for training, so the DNN can memorize the position-dependent systematic shift. The largest claimed factor (12.1 in the abstract range) corresponds to this split: Table 1 gives (687.6 µm / 198.0 µm)^2 ≈ 12.1. Thus the upper headline value is partly memorization rather than prediction on unseen positions. The held-out splittings B and C provide legitimate generalization evidence and yield the smaller factor (~5.7), which is why this is partial circularity rather than a fully forced result.

full rationale

The core experiment is a fair calibration comparison: both the linear map and the DNN are fit to the same charge-to-motor-position objective and evaluated on held-out positions (splittings B and C), giving genuine evidence that the DNN interpolates the sensor nonlinearity better than the 5-parameter linear formula. No load-bearing argument reduces to a self-citation: [12] supplies the device formula and [20] the Rice fitting, but the DNN result is an independent measurement on this tile. The circularity is limited to the interpretation layer: 'granularity' is defined as 2⟨d⟩, the same error the network minimizes, so the factor-of-5.7-12.1 'resolved areas' claim is partly the training loss renamed as a resolution. In addition, the upper end of the claimed range comes from split A, where test positions coincide with training positions, and the Q-threshold cut excludes edge and inter-tile-gap events, so the 6400-region extrapolation over the full 16x16 mm² field is not directly measured. These are metric-definition and extrapolation issues rather than a derivation that reduces entirely to its inputs.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The paper's position estimate is entirely determined by parameters fitted to the same scan because all free parameters (linear alignment, A-matrix, DNN weights, threshold, hyperparameters) are calibrated on motor-position labels from the device being characterized. No independent external calibration dataset or first-principles constraint fixes these numbers. The DNN contributes no new physical entity, but its generalization outside the scanned interior grid is assumed.

free parameters (5)
  • Linear reconstruction alignment parameters (lx, ly, x0, y0, phi) = not reported (minimized on training sample)
    Eqs. 3-4 map relative LG-SiPM coordinates to motor frame; fitted by minimizing squared distance to motor positions.
  • Zero-hidden-layer A matrix entries and bias (14 parameters) = not reported
    Section 2.3.1 frees all 12 linear weights plus 2 offsets to motivate the DNN; fit to training data.
  • DNN weights and biases (6-64*N-2 MLP) = not reported
    Section 2.3.2; trained with Adam for 40 epochs on motor-position labels; exact weights not released.
  • Total-charge quality cut threshold = not reported (red line in Fig. 3)
    Section 2.1; removes events at edges/gaps, changing the evaluated sample and hence granularity.
  • DNN hyperparameters (number of hidden layers, batch size, learning rate, seeds) = not fully reported; 40 epochs, 64 units, tanh
    Section 2.3.2; model selection among Nlayers=0..5 is done on the same data; batch size and learning rate are omitted.
assumptions (5)
  • domain assumption Motor-stage coordinates equal the true optical spot position and the ~2 mm LED spot's center of gravity corresponds to those coordinates.
    Used as ground truth labels for linear calibration and DNN training (Section 2, Eqs. 3-4, 7).
  • domain assumption The LG-SiPM charge-to-position response is stable during the scan and the mapping learned from 2 mm LED spots transfers to gamma scintillation light distributions.
    No gamma-source data are presented; the DNN is trained and evaluated on the same LED scan of one device (Sections 2, 3).
  • domain assumption The 6-channel 'smart channel' wiring and Qi ordering are as described in prior LG-SiPM publications [12,14,16].
    The paper uses Eqs. 1-2 without redefining which physical outputs correspond to Q1..Q6.
  • standard math Distances d between reconstructed and motor positions follow a Rice distribution, so sigma and nu can be extracted from fits.
    Section 3.3 follows [19,20] to estimate resolution and shift from Eq. 7.
  • domain assumption Held-out grid positions in splittings B and C are sufficiently separated from training positions for the test metrics to represent generalization rather than interpolation with near-zero distance.
    Chessboard split has train/test positions only 0.5-1.0 mm apart, so spatial correlation may still favor the DNN (Section 2.1).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Position-Sensitive Silicon Photomultiplier Array with Enhanced Position Reconstruction by means of a Deep Neural Network." pith.science (2026). https://pith.science/paper/IFDFBZAF

@misc{pith2026251202771,
  author       = {Pith},
  title        = {Pith review of: Position-Sensitive Silicon Photomultiplier Array with Enhanced Position Reconstruction by means of a Deep Neural Network},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IFDFBZAF}},
  note         = {Machine review of arXiv:2512.02771}
}
read the original abstract

Single-photon sensitive detectors like Silicon Photomultipliers are widely used in many medical imaging applications. By using detectors with position resolutions, it is possible to build compact photodetector readouts with reduced number of channels, but still preserving position resolution and gamma-rays imaging capabilities. In this work, we present the advantage of using a Deep Neural Networks (DNNs) light position reconstruction applied to a 2x2 array of linearly-graded SiPMs (LG-SiPMs), to minimize the distortions on the reconstructed event maps. Our approach significantly enhances both the resolution and linearity of position detection compared to the nominal reconstruction formula based on the device architecture. Remarkably, the DNN-based reconstruction boosts the number of resolved areas (pixels) by a factor of 5.7 to 12.1 (depending the training splitting used) allowing for a higher level of precision and performance in light detection.

Figures

Figures reproduced from arXiv: 2512.02771 by the authors.

Figure 1
Figure 1. Left: front view of the SiPM tile, mounted in a compact module including signal [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the 3 splittings used for training and testing. The gray scale indicates [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Total charge distribution (black) of the test and train sample. A lower threshold [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Schematic of the Deep Neural Network architecture used. The DDNs consist of an [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Reconstructed position of the test sample for the splitting technique [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Loss of each DNN during training as function of the epoch for the training sample [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Distribution of the distance d between the reconstructed positions and motor posi￾tions for the random splitting (A) (left), the chessboard splitting (B) (center) and the run splitting (C) (right) techniques on the test sample. The distributions are shown for all 0 to …
Figure 8
Figure 8. Figure 8: Mean resolution (left), shift (center) and distances (right) as function of the number [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 9 canonical work pages

  1. [1]

    H. O. Anger, Scintillation camera, Review of Scientific Instruments 29 (1958) 27–33. URL:https://doi.org/10.1063/1.1715998. doi:10.1063/ 1.1715998

    work page doi:10.1063/1.1715998 1958

  2. [2]

    S. R. Cherry, J. A. Sorenson, M. E. Phelps, chapter 14 - the gamma camera: Performance characteristics, in: S. R. Cherry, J. A. Sorenson, M. E. Phelps (Eds.), Physics in Nuclear Medicine (Fourth Edition), fourth edition ed., W.B. Saunders, Philadelphia, 2012, pp. 209–231. URL:https://www. 10 sciencedirect.com/science/article/pii/B9781416051985000149. doi:...

    work page doi:10.1016/b978-1-4160-5198-5.00014-9 2012

  3. [3]

    T. E. Peterson, L. R. Furenlid, Spect detectors: the anger cam- era and beyond, Physics in Medicine and Biology 56 (2011) R145– R182. URL:https://doi.org/10.1088/0031-9155/56/17/R01. doi:10. 1088/0031-9155/56/17/R01

    work page doi:10.1088/0031-9155/56/17/r01 2011

  4. [4]

    Tsuchimochi, K

    M. Tsuchimochi, K. Hayama, Intraoperative gamma cameras for radio- guided surgery: Technical characteristics, performance parameters, and clinical applications, Physica Medica 29 (2013) 126–138. URL:https: //www.sciencedirect.com/science/article/pii/S1120179712000397. doi:https://doi.org/10.1016/j.ejmp.2012.05.002

    work page doi:10.1016/j.ejmp.2012.05.002 2013

  5. [5]

    A. L. Farnworth, S. L. Bugby, Intraoperative gamma cameras: A review of development in the last decade and future outlook, Journal of Imaging 9 (2023). doi:10.3390/jimaging9050102

    work page doi:10.3390/jimaging9050102 2023

  6. [6]

    M. G. Bisogni, A. Del Guerra, N. Belcari, Medical applications of silicon photomultipliers, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associ- ated Equipment 926 (2019) 118–128. URL:https://www.sciencedirect. com/science/article/pii/S016890021831489X. doi:https://doi.org/ 10.1016/j.nima.2018.1...

    work page doi:10.1016/j.nima.2018.10.175 2019

  7. [7]

    Gundacker, A

    S. Gundacker, A. Heering, The silicon photomultiplier: fundamentals and applications of a modern solid-state photon detector, Physics in Medicine and Biology 65 (2020) 17TR01. URL:https://dx.doi.org/10. 1088/1361-6560/ab7b2d. doi:10.1088/1361-6560/ab7b2d

    doi:10.1088/1361-6560/ab7b2d 2020

  8. [8]

    Acerbi, A

    F. Acerbi, A. Raiola, C. Alispach, H. Arabi, H. Zaidi, A. Gola, D. Della Volpe, Compact and handheld sipm-based gamma cam- era for radio-guided surgery and medical imaging, Instruments 9 (2025). URL:https://www.mdpi.com/2410-390X/9/2/14. doi:10.3390/ instruments9020014

    2025

Show all 21 references
  1. [9]

    della Volpe, A

    D. della Volpe, A. Gola, A. Raiola, C. Alispach, F. Acerbi, H. Zaidi, H. Arabi, The POSiCS handheld gamma-ray camera for radio-guided surgery, EPJ Web Conf. 338 (2025) 09009. URL: https://www.epj-conferences.org/articles/epjconf/abs/2025/ 23/epjconf_animma2025_09009/epjconf_an...

    2025

  2. [10]

    Zhang, Y

    X. Zhang, Y. Liu, Y. Shao, Y. Yan, R. Yang, K. Liang, D. Han, One- dimensional position-sensitive lgad with a dc coupled and resistive charge division mechanism, IEEE Transactions on Nuclear Science 70 (2023) 853–

    2023

  3. [11]

    Schaefer, R

    P. Schaefer, R. D. Williams, G. K. Davis, R. A. Ross, Accuracy of posi- tion detection using a position-sensitive detector, IEEE Transactions on Instrumentation and Measurement 47 (1998)

    1998

  4. [12]

    A. Gola, A. Ferri, A. Tarolli, N. Zorzi, C. Piemonte, A novel ap- proach to position-sensitive silicon photomultipliers: First results, 2013 IEEE Nuclear Science Symposium and Medical Imaging Conference (2013 NSS/MIC) (2013) 1–4. URL:https://api.semanticscholar.org/ CorpusID:24489596

    2013

  5. [13]

    Acerbi, S

    F. Acerbi, S. Merzi, A. Gola, Large area tiles of position-sensitive silicon photomultipliers, in: Proceedings of the 13th Interna- tional Conference on Position Sensitive Detectors (PSD13), Fon- dazione Bruno Kessler (FBK), Oxford, United Kingdom, 2023. URL:https://indico.glo...

    2023

  6. [14]

    Acerbi, S

    F. Acerbi, S. Merzi, A. Gola, Position-sensitive silicon photomultiplier arrays with large-area and sub-millimeter resolution, Sensors 24 (2024). doi:10.3390/s24144507

    2024 doi

  7. [15]

    A. Gola, K. Majumdar, G. Casse, K. Mavrokoridis, S. Merzi, L. P. Franca, First demonstration of the use of lg-sipms for optical readout of a tpc (2020). URL:http://arxiv.org/abs/ 2009.05086http://dx.doi.org/10.1088/1748-0221/15/12/P12017. doi:10.1088/1748-0221/15/12/P12017

    2020 arXiv

  8. [16]

    Ferri, F

    A. Ferri, F. Acerbi, A. Gola, G. Paternoster, C. Piemonte, N. Zorzi, Char- acterization of linearly graded position-sensitive silicon photomultipliers, EJNMMI Physics 1 (2014) A14. doi:10.1186/2197-7364-1-S1-A14

    2014 doi

  9. [17]

    Jaliparthi, P

    G. Jaliparthi, P. Martone, A. V. Stolin, R. R. Raylman, Deep residual- convolutional neural networks for event positioning in a monolithic annular pet scanner, Physics in Medicine & Biology 66 (2021). URL:https://api. semanticscholar.org/CorpusID:235595794

    2021

  10. [18]

    Chollet, et al., Keras,https://keras.io, 2015

    F. Chollet, et al., Keras,https://keras.io, 2015

    2015

  11. [19]

    K. K. Talukdar, W. D. Lawing, Estimation of the parameters of the rice distribution, The Journal of the Acoustical Society of America 89 (1991) 1193–1197. URL:https://doi.org/10.1121/1.400532. doi:10.1121/1. 400532

    1991 doi

  12. [20]

    Raiola, F

    A. Raiola, F. Acerbi, C. Alispach, H. Arabi, D. della Volpe, A. Gola, H. Zaidi, Quantitative determination of spatial resolution and linear- ity of position-sensitive lg-sipms at sub-millimeter scale via ricean dis- tribution fitting, Nuclear Instruments and Methods in Physics...

    2025

  13. [858]

    doi:10.1109/TNS.2023.3264661. 11

    2023

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.