Calibration of DOI-Capable PET Detector Panels Using Uncollimated Front-Face Irradiation

Andre Z. Kyme; Francisco E. Enr\'iquez-Mier-y-Ter\'an; Steven R. Meikle

arxiv: 2607.01592 · v1 · pith:GGTNRIHCnew · submitted 2026-07-02 · ⚛️ physics.ins-det

Calibration of DOI-Capable PET Detector Panels Using Uncollimated Front-Face Irradiation

Francisco E. Enr\'iquez-Mier-y-Ter\'an , Andre Z. Kyme , Steven R. Meikle This is my paper

Pith reviewed 2026-07-03 03:26 UTC · model grok-4.3

classification ⚛️ physics.ins-det

keywords PET detectorsdepth of interactionDOI calibrationuncollimated irradiationmultilayer perceptronenergy resolutiondetector panels

0 comments

The pith

MLP calibration from uncollimated irradiation achieves 2.0 mm DOI resolution in PET detector panels.

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

The paper establishes that an MLP-based method can calibrate depth-of-interaction and energy in a PET detector panel using only uncollimated front-face irradiation. It demonstrates this by calibrating three blocks with the gold-standard collimated method, comparing physics-informed and MLP approaches on uncollimated data, and then applying the MLP to the full 4x4 panel. The calibrated panel reaches 15.6% energy resolution and 2.0 mm DOI resolution. A reader would care because this replaces time-consuming collimated calibration with a simple single-irradiation procedure suitable for system-level use.

Core claim

The MLP-based approach to DOI calibration using uncollimated irradiation generalizes from three gold-standard calibrated blocks to the full panel, with RMSEs of 0.36-0.61 mm versus gold-standard, and the panel achieves a mean energy resolution of 15.6% and a DOI resolution of 2.0 mm after saturation correction.

What carries the argument

The multilayer perceptron that predicts crystal depth from photopeak location in uncollimated irradiation data, trained on three blocks and applied to the panel.

If this is right

Uncollimated irradiation yields photopeak-to-depth mappings with median relative RMSE of 1% to collimated second-order polynomial fits.
The MLP approach outperforms the physics-informed model in DOI resolution estimates.
The full panel can be calibrated using only a single uniform 511-keV irradiation.
After saturation correction, the panel has 15.6% mean energy resolution and 2.0 mm DOI resolution.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

In situ calibration of complete PET systems becomes feasible without collimators or disassembly.
The approach may generalize to other dual-ended readout scintillator configurations.
Block-to-block manufacturing variations are apparently small enough for the model to handle.
Independent validation on panels with different crystal sizes would test the method's robustness.

Load-bearing premise

The photopeak-to-depth mapping learned from three collimated-calibrated blocks will generalize accurately to the remaining blocks in the panel under uncollimated irradiation without additional depth-specific constraints or overfitting.

What would settle it

An independent collimated-beam measurement of DOI resolution on the MLP-calibrated full panel would directly test if the 2.0 mm resolution holds.

Figures

Figures reproduced from arXiv: 2607.01592 by Andre Z. Kyme, Francisco E. Enr\'iquez-Mier-y-Ter\'an, Steven R. Meikle.

**Figure 1.** Figure 1: Experimental setups for the single-detector irradiation. (a) Gold-standard irradiation using a slab detector for electronic collimation. (b) Front-face irradiation using an uncollimated source. 2.1.1. Electronically Collimated Irradiation A Teflon-wrapped LYSO slab (0.5 mm × 20 mm × 20 mm), coupled to a 1 × 6 MicroFJ30035 SiPM array, was assembled to electronically collimate a 22Na point source (Fig. 1a).… view at source ↗

**Figure 3.** Figure 3: Detector panel irradiation setups. Detectors are numbered from 01 to 16 for clarity. Detector panel calibration (see Section 2.3.2) was performed using uncollimated irradiation from a 22Na and 137Cs source, respectively, centred with respect to the detector panel front face and positioned 60 mm from it in the 𝑧 direction. The accuracy of the proposed methods for DOI calibration parameters and resolution (s… view at source ↗

**Figure 4.** Figure 4: (a) The DOI ratio histogram arising from uncollimated irradiation (red curve) is modelled as the convolution of the attenuation of the 511-keV photons in the crystal (grey line) and the intrinsic DOI response function of the detector, assumed to be Gaussian distributed. (b) The MLP is fed with crystal-wise DOI ratio histograms to extract the DOI calibration parameters (𝑘 𝑀𝐿𝑃 ,𝑖 𝑀𝐿𝑃) and the DOI resolution … view at source ↗

**Figure 5.** Figure 5: (a) Photopeak location (in ASIC units) as a function of the detector depth for the uncollimated (blue) and collimated (red) irradiation methods. Box locations correspond to the depth centre of the virtual crystal. (b) RRMSE between the quadratic fits to the crystal-wise photopeak locations measured in each irradiation method. (c) Energy resolution (in %) for both irradiation methods. 3.2. Detector Panel Ir… view at source ↗

**Figure 6.** Figure 6: (top) DOI calibration for the central crystal in detector 01 (left), 02 (centre) and 03 (right) using ground-truth collimated irradiation at known depths (grey), and uncollimated irradiation using the DOI model (red) and MLP model (blue). (bottom) RMSE between the ground-truth DOI calibration and DOI model (red) or MLP model (blue) [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗

**Figure 8.** Figure 8: Relative percentage error in DOI resolution from the MLP [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗

read the original abstract

Objective. Gold-standard depth-of-interaction (DOI) calibration using collimated gamma-ray irradiation is time-consuming and impractical for system-level calibration of detector arrays. This work investigates an efficient DOI and energy calibration method for detector panels using uncollimated irradiation, with gamma rays incident nearly parallel to the crystal depth direction. Approach. The 511-keV photopeak location in a dual-ended readout PET detector block was evaluated as a function of crystal depth using collimated and uncollimated $^{22}$Na irradiation. A $4\times4$ dual-ended readout PET detector panel was then assembled. Three detector blocks were calibrated using the gold-standard method, and two uncollimated-irradiation DOI calibration approaches--a physics-informed model and a multilayer perceptron (MLP)--were compared against it. Finally, the full panel was calibrated for DOI and energy using the MLP-based approach. Main Results. The median relative RMSE between second-order polynomial fits from collimated and uncollimated irradiation was 1%, showing that uncollimated irradiation can provide reliable estimates when accurate DOI calibration parameters are available. Compared with gold-standard DOI calibration, the physics-informed and MLP-based approaches achieved RMSEs of 0.38-0.58 mm and 0.36-0.61 mm, respectively. The MLP-based approach provided better DOI resolution estimates and was therefore used for full-panel calibration. After saturation correction, the panel achieved a mean energy resolution of 15.6% and a DOI resolution of 2.0 mm. Significance. The proposed MLP-based calibration requires only a single uniform 511-keV irradiation, making it simple to implement and suitable for in situ calibration of DOI-capable PET detector arrays.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

MLP matches gold-standard RMSE on three blocks but the panel-wide 2.0 mm claim rests on untested transfer to the other thirteen.

read the letter

The main things to know are that their MLP learns a photopeak-to-depth mapping from uncollimated irradiation and hits 0.36-0.61 mm RMSE against collimated gold-standard data on three blocks, close to their physics-informed model, then they apply the MLP to the full 4x4 panel for a reported 2.0 mm DOI resolution and 15.6% energy resolution.

What is actually new is the direct use of an MLP for this specific uncollimated front-face setup as a practical alternative to collimated calibration. The paper does well on the three-block comparison: it gives concrete RMSE numbers, shows the uncollimated photopeak shift tracks depth with only 1% median relative error to the collimated polynomial fits when parameters are known, and picks the MLP because it produced better resolution estimates than the physics model.

The soft spot is the jump to the remaining blocks. They calibrate three blocks with the gold-standard collimated method, compare both uncollimated approaches there, then calibrate the whole panel with the MLP. No held-out block, cross-validation, or separate check on the other thirteen is described, so any inter-block differences in light yield or coupling are untested. That makes the panel-level 2.0 mm figure rest on an assumption of transferability. It is a moderate rather than fatal issue if the blocks are similar, but it is the load-bearing step for the main claim.

The work is for people building or calibrating DOI PET detector arrays who want to avoid full collimated setups. A reader in detector instrumentation would get value from the reported metrics and the single-irradiation simplicity. It deserves a serious referee because the experimental comparison on the three blocks is direct and the practical motivation is clear, even though the generalization point needs attention in revision.

Referee Report

1 major / 0 minor

Summary. The paper claims that uncollimated front-face 511-keV irradiation can be used for efficient DOI and energy calibration of dual-ended readout PET detector panels. After showing that uncollimated photopeak-vs-depth data agree with collimated gold-standard data to ~1% relative RMSE on three blocks, the authors compare a physics-informed model (RMSE 0.38-0.58 mm) and an MLP (RMSE 0.36-0.61 mm) against the gold standard, select the MLP, and apply it to calibrate an entire 4×4 panel, obtaining 15.6% mean energy resolution and 2.0 mm DOI resolution after saturation correction.

Significance. If the MLP mapping generalizes, the method would enable practical, in-situ DOI calibration of large detector arrays with only a single uniform irradiation, removing the need for time-consuming collimated scans on every block. The concrete RMSE numbers and direct comparison to an independent gold-standard measurement on the training blocks constitute a clear strength.

major comments (1)

[Abstract / full-panel results] Abstract and full-panel calibration section: The MLP is trained exclusively on the three collimated-calibrated blocks and then applied to the remaining blocks of the 4×4 panel. No held-out block, cross-validation across blocks, or inter-block variability metric is reported, so the claimed panel-wide DOI resolution of 2.0 mm rests on the untested assumption that the learned photopeak-to-depth mapping transfers without degradation due to block-to-block differences in light yield, optical coupling, or crystal properties.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive review and the recognition of the method's potential significance. The major comment concerns the generalization of the MLP to the full panel without explicit cross-validation. We address this point below and indicate where revisions will be made to strengthen the manuscript.

read point-by-point responses

Referee: [Abstract / full-panel results] Abstract and full-panel calibration section: The MLP is trained exclusively on the three collimated-calibrated blocks and then applied to the remaining blocks of the 4×4 panel. No held-out block, cross-validation across blocks, or inter-block variability metric is reported, so the claimed panel-wide DOI resolution of 2.0 mm rests on the untested assumption that the learned photopeak-to-depth mapping transfers without degradation due to block-to-block differences in light yield, optical coupling, or crystal properties.

Authors: We agree that the manuscript does not include held-out block testing or formal cross-validation across the 16 blocks. The three blocks used for gold-standard calibration and MLP training were selected as representative of the panel manufacturing process. The 1% median relative RMSE agreement between collimated and uncollimated photopeak-vs-depth curves on these blocks provides indirect support for consistency, but does not directly quantify transfer to the other 13 blocks. In the revised version we will add (i) a table or figure showing the distribution of uncollimated photopeak positions and energy resolutions across all 16 blocks and (ii) a brief discussion of the assumption of block-to-block similarity, supported by the observed low variability in the uncollimated data. The panel-wide 2.0 mm DOI resolution figure will be qualified accordingly. This is a partial revision that directly responds to the concern without requiring new experiments. revision: partial

Circularity Check

0 steps flagged

No circularity: empirical validation against independent gold-standard measurements

full rationale

The paper compares two uncollimated calibration approaches (physics-informed model and MLP) to gold-standard collimated irradiation on three blocks, reporting RMSE values of 0.36-0.61 mm. The MLP is then applied to the remaining blocks in the 4x4 panel, with final panel metrics (15.6% energy resolution, 2.0 mm DOI resolution) presented after this step. No equations, self-definitions, or fitted quantities are shown to reduce to the model inputs by construction. The derivation chain consists of experimental comparisons and application of a trained model; the central claims rest on independent collimated reference data rather than tautological renaming or self-referential prediction. Generalization assumptions affect validity but do not constitute circularity per the enumerated patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on the domain assumption that photopeak position encodes depth information even under broad irradiation; no free parameters or new entities are introduced in the abstract.

axioms (1)

domain assumption Photopeak location in dual-ended readout detectors varies systematically with interaction depth.
Invoked to justify using uncollimated data for depth estimation.

pith-pipeline@v0.9.1-grok · 5870 in / 1172 out tokens · 25570 ms · 2026-07-03T03:26:15.284944+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

2 extracted references · 2 canonical work pages · 1 internal anchor

[1]

Allen M S, Scipioni M and Catana C 2024 New Horizons in Brain PET Instrumentation PET Clinics 19 25–36 Bircher C and Shao Y 2012 Use of internal scintillator radioactivity to calibrate DOI function of a PET detector with a dual -ended-scintillator readout: DOI function calibration with internal scintillator radioactivity Med. Phys. 39 777–87 Borghi G, Tab...

work page 2024
[2]

Accelerators, Spectrometers, Detectors and Associated Equipment 917 1–8 Kuhl Y, Naunheim S, Schug D, Schulz V and Mueller F 2023 Angular Irradiation Methods for DOI Calibration of Light-Sharing Detectors—A Perspective for PET In-System Calibration IEEE Trans. Radiat. Plasma Med. Sci. 7 673–83 Kyme A Z, Judenhofer M S, Gong K, Bec J, Selfridge A, Du J, Qi ...

work page internal anchor Pith review Pith/arXiv arXiv 2023

[1] [1]

Allen M S, Scipioni M and Catana C 2024 New Horizons in Brain PET Instrumentation PET Clinics 19 25–36 Bircher C and Shao Y 2012 Use of internal scintillator radioactivity to calibrate DOI function of a PET detector with a dual -ended-scintillator readout: DOI function calibration with internal scintillator radioactivity Med. Phys. 39 777–87 Borghi G, Tab...

work page 2024

[2] [2]

Accelerators, Spectrometers, Detectors and Associated Equipment 917 1–8 Kuhl Y, Naunheim S, Schug D, Schulz V and Mueller F 2023 Angular Irradiation Methods for DOI Calibration of Light-Sharing Detectors—A Perspective for PET In-System Calibration IEEE Trans. Radiat. Plasma Med. Sci. 7 673–83 Kyme A Z, Judenhofer M S, Gong K, Bec J, Selfridge A, Du J, Qi ...

work page internal anchor Pith review Pith/arXiv arXiv 2023