{"id":"2ed84e39-85d9-44ed-8a44-a7cca277a9c0","arxiv_id":"2506.13460","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"First positronium lifetime imaging with scandium-44 yields lifetimes consistent with water, but unresolved 1157 keV prompt photons degrade count statistics, making 44Sc inferior to 124I on this scanner.","lead":"A medical-imaging team tested scandium-44 for measuring positronium lifetimes in a water-filled phantom on a long-axial-field-of-view PET/CT scanner. The scanner's limited energy resolution caused excess random coincidences, making scandium-44 less useful than iodine-124 for this new three-photon imaging technique.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The comparative claim rests on unmatched 44Sc/124I acquisition conditions: the pSBR and error comparisons are not normalized for the ~6.2x activity-concentration difference, so the energy-resolution mechanism may be confounded by count rate.","rationale":"The reader's weakest assumption correctly flags the unmatched activity concentration and scan duration between the 44Sc and 124I scans. That is the most load-bearing point because the paper's central negative claim is explicitly comparative, and the only quantitative evidence for the comparison is the pSBR ratio and τ3 uncertainty ratio. I partially agree: the activity/scan-time mismatch is the core issue, but the concern is broader, since the two studies may also differ in prompt-photon energy-window selection and geometry, and the pSBR metric is count-rate dependent. A matched-condition reanalysis or simulation would settle whether the observed inferiority of 44Sc is due to the unresolved 1157 keV photopeak or simply to lower accumulated counts. Because the paper already emphasizes that its conclusion applies to the given methodology and the reader's verdict is CONDITIONAL, my stress-test does not move the verdict; it sharpens the condition under which the conclusion would be accepted.","tokens_in":11328,"tokens_out":15099,"duration_ms":158081,"concrete_test":"Reanalyze the 124I list-mode data from Ref. [49] (or a validated Quadra Monte Carlo simulation) under conditions matched to the 44Sc scan: same activity concentration (40.68 kBq/ml), same 20 min acquisition, same 476-546 keV annihilation window, same voxel size and phantom geometry, and each isotope's realistic prompt-photon selection (e.g., 603 keV window for 124I, 720-735 keV overflow bin for 44Sc). Recompute the pSBR and marginalized τ3 uncertainty in a central 4x4x4 mm³ voxel. If the matched 124I pSBR remains substantially above 12.6 and the τ3 error remains about four times smaller, the energy-resolution mechanism is confirmed; if the gap shrinks or reverses, the comparative conclusion is not supported and should be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central conclusion that 44Sc cannot outperform 124I is supported primarily by comparing the measured pSBR of 12.6 (44Sc) with the 55.5 reported in Ref. [49] for 124I, and by comparing the τ3 uncertainties. These comparisons are not apples-to-apples. The 44Sc scan used 40.68 kBq/ml in the spheres for 20 min, whereas the 124I reference used 252 kBq/ml for 15 min, a 6.2-fold activity-concentration difference with a different scan duration. The pSBR in a PAL spectrum is not isotope-intrinsic: the true 3γE rate scales roughly linearly with activity, while the accidental-coincidence background scales roughly with the product of singles rates, so pSBR can change strongly with count rate. The two studies may also differ in prompt-photon energy-window selection, because the 44Sc scan must use the 720-735 keV overflow bin for the unresolved 1157 keV line, whereas Ref. [49] could select a resolved 124I line, and the compared voxels sit in different geometries. The Discussion acknowledges the concentration difference but does not normalize for it. Without a matched-count-rate comparison, the observed gap in pSBR and τ3 uncertainty cannot be unambiguously attributed to the unresolved 1157 keV photopeak rather than to the different acquisition conditions. This is the load-bearing step for the paper's headline negative result.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports the first positronium lifetime imaging measurement with 44Sc on a commercial long-axial field-of-view PET/CT. A NEMA image-quality phantom filled with 41.7 MBq of 44Sc in water was scanned on a Biograph Vision Quadra for 20 minutes in singles mode. Three-photon events were selected with prototype software, and positron annihilation lifetime (PAL) spectra were fitted with a Bayesian procedure. The oPs lifetime is determined in the six phantom spheres and in 4x4x4 mm^3 voxels; the values are reported to be compatible with the literature water value (1.839 ns), though with large uncertainties. The main conclusion is that, because the Quadra cannot resolve the 1157 keV prompt photon of 44Sc, random three-photon coincidences dominate, and 44Sc does not outperform 124I in count statistics for oPs lifetime imaging under the present methodology.","tokens_in":11520,"tokens_out":8233,"duration_ms":79754,"significance":"The paper provides a useful and honest negative result for the emerging field of positronium lifetime imaging: it documents a scanner-specific limitation that will inform radionuclide selection and detector design. The strengths are the clear description of the measurement protocol, the external literature check on the oPs lifetime, the transparent reporting of large statistical uncertainties, and the explicit limitation of the conclusion to the given methodology. If the comparative claim is substantiated, the paper will help prevent futile clinical studies with 44Sc on this class of scanners.","major_comments":[{"comment":"The central conclusion that 44Sc cannot outperform 124I rests on a cross-paper comparison with Ref. [49] that is not normalized for acquisition conditions. The 44Sc measurement used 40.68 kBq/ml for 20 min in NEMA spheres, whereas the 124I reference used 252 kBq/ml for 15 min in water tubes. The pSBR and τ3 uncertainty are count-rate dependent quantities: the true 3γE rate scales approximately linearly with activity, while the accidental-coincidence background scales with a higher power of the singles rates. The manuscript acknowledges the concentration difference but does not correct for it, so the observed gap (pSBR 12.6 vs 55.5; τ3 error about four times larger) cannot be unambiguously attributed to the unresolved 1157 keV photopeak rather than to the different count rates and geometry. I request either a matched-count-rate comparison, a quantitative model of the activity and scan-time dependence, or an explicit sensitivity analysis.","section":"§4 Discussion, pSBR comparison"},{"comment":"The paper does not report the basic count-statistics quantities needed to evaluate the central claim: the number of selected 3γE events, the estimated random-coincidence fraction, or the pSBR for each ROI. Reporting these values for the spheres and voxels would allow the reader to see how the background grows with the prompt-photon window and to compare with Ref. [49] on a like-for-like basis. As written, the random-coincidence mechanism is inferred from a single pSBR comparison rather than demonstrated from measured singles or coincidence rates.","section":"§2 Method / §3 Results"},{"comment":"The statement that all measured oPs lifetimes are consistent with the literature water value deserves a more careful statistical treatment. The value for s2 is 1.39 ± 0.20 ns, about 2.25 standard deviations below the reference 1.839 ± 0.015 ns; with six spheres plus one voxel, this may still be a plausible fluctuation, but a multiple-comparison or systematic-uncertainty discussion would strengthen the validation claim.","section":"§3 Results, Table 2"}],"minor_comments":[{"comment":"Table 1 is difficult to read: the rows for 82Rb and 124I are split across lines, and the BRγ/β+ values for 82Rb are unclear. A single row per nuclide with an additional footnote would improve clarity.","section":"Table 1"},{"comment":"The Fig. 1 caption says 'The absence attenuation correction'; this should read 'The absence of attenuation correction'.","section":"Fig. 1 caption"},{"comment":"In the text, 'spheres1' (Fig. 1 legend) and 'the spheres4 has a volume' (Discussion) are grammatically awkward; these should be rephrased as 'sphere s1' and 'sphere s4 has a volume'.","section":"General text"},{"comment":"The typo '4.8, %' should be corrected to '4.8%'.","section":"§4 Discussion"},{"comment":"The fitting priors are taken from Ref. [49] without being restated; since the lifetime validation depends on the fit, a short statement of the priors or a prior-sensitivity check would improve reproducibility.","section":"§2 Method"},{"comment":"Ref. [25] is a preprint; if a peer-reviewed version exists, it should be cited instead.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The main risk is that the comparative claim depends on Ref. [49], an arXiv preprint by largely the same group. If that reference is not yet peer-reviewed, the reliability of the comparison is reduced. I would encourage the editor to ensure that the 124I data are publicly available or that the key 124I results are included as an appendix. The negative result, if supported, is of genuine value to the field."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First off, the useful thing here is the negative result: 44Sc's high prompt-photon branching ratio does not automatically translate into better oPs lifetime statistics on the Quadra, because the 1157 keV photopeak lands in the overflow bin and inflates random 3γE. The paper is honest about this, and the measured lifetimes agree with the literature water value, so the measurement itself looks clean.\n\nWhat's new: first 44Sc oPs lifetime imaging on a clinical LAFOV PET/CT, and first demonstration that the unresolved high-energy photopeak hurts count statistics. The method is clearly described and the error bars are realistic.\n\nThe soft spot is the comparison with 124I. The pSBR of 12.6 vs 55.5 and the τ3 errors are compared across studies with very different activity concentrations: 40.68 kBq/ml for 20 min vs 252 kBq/ml for 15 min. That's a 6.2x difference in start activity. The paper mentions the difference but doesn't normalize for it, so the gap could be partly due to count rate rather than energy resolution. The stress-test note is correct that this is load-bearing for the headline conclusion. I'd want to see a matched-activity comparison or a simulation that separates the two effects before fully believing the energy-resolution story.\n\nStill, the physical argument is plausible, and the authors explicitly limit the claim to 'the given methodology.' The paper gives the community a useful data point and a clear reason to think carefully about detector energy resolution when choosing radionuclides.\n\nThe citation pattern is fine, nothing egregious. Data availability is 'upon reasonable request' rather than open, which is a minor drawback.\n\nThis deserves peer review. It's a short, honest experimental report with a clear negative result. The comparison weakness is fixable with a revision or a follow-up, not a fatal flaw.","headline":"A clean first measurement of 44Sc positronium lifetimes on a clinical LAFOV PET/CT that delivers a useful negative result, though the isotope comparison driving the conclusion is confounded by unmatched activity concentrations.","tokens_in":12246,"tokens_out":1869,"would_cite":true,"duration_ms":18364,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Scandium-44 does not improve positronium lifetime imaging over iodine-124 on this scanner because its 1157 keV prompt photon is unresolved and random coincidences dominate.","keywords":["scandium-44","positronium lifetime imaging","ortho-positronium","long axial field-of-view PET/CT","three-photon events","prompt gamma","count statistics","NEMA phantom"],"falsifier":"Run $^{44}\\mathrm{Sc}$ in the same phantom at the same 252 kBq/ml activity concentration and 15-minute scan duration used for the $^{124}\\mathrm{I}$ reference: if the voxel pSBR stays near 12.6 and the $\\tau_3$ error remains several times larger, the unresolved 1157 keV photon is confirmed as the cause; if its statistics match $^{124}\\mathrm{I}$, the comparison is confounded by count rate.","tokens_in":11032,"feed_emoji":"⏱️","tokens_out":10721,"duration_ms":96189,"temperature":0.7,"pith_summary":"This paper asks whether scandium-44 ($^{44}\\mathrm{Sc}$), which emits a prompt gamma ray in nearly every positron decay, can supply the count statistics needed for ortho-positronium (oPs) lifetime imaging on a commercial long axial field-of-view PET/CT. The authors scanned a water-filled NEMA phantom containing $^{44}\\mathrm{Sc}$, selected three-photon events with the scanner's prototype feature and software, and fitted oPs lifetimes in the phantom spheres and in $4\\times4\\times4$ mm$^3$ voxels. All fitted lifetimes agree with the known water value, so the measurement chain works. The decisive result is statistical: a single-voxel peak signal-to-background ratio of about 12.6, against 55.5 reported for $^{124}\\mathrm{I}$, and background-region relative errors above 20% even inside the spheres. The paper concludes that $^{44}\\mathrm{Sc}$ cannot outperform $^{124}\\mathrm{I}$ for oPs lifetime imaging under this methodology, because the 1157 keV prompt photon is not resolved by the detector and creates excess random three-photon coincidences.","feed_headline":"Scandium-44 trails iodine-124 in positronium lifetime imaging","feed_subtitle":"Its 1157 keV prompt photon lands in an unresolved energy bin, inflating random coincidences and lifetime uncertainties.","key_machinery":"The load-bearing quantity is the positron annihilation lifetime (PAL) spectrum, in which the 1157 keV nuclear gamma of $^{44}\\mathrm{Sc}$ starts a clock and the two 511 keV annihilation photons stop it; the oPs lifetime $\\tau_3$ is extracted by a Bayesian fit. The spectrum's peak signal-to-background ratio (pSBR) carries the argument: about 12.6 for $^{44}\\mathrm{Sc}$ in a single voxel here, compared with 55.5 reported for $^{124}\\mathrm{I}$. The mechanism behind the low ratio is the scanner's energy discrimination, which collects every photon above 726 keV into one final bin, so the 1157 keV prompt photon cannot be rejected and unresolved hits swell the random background.","core_discovery":"This is the first oPs lifetime imaging study with $^{44}\\mathrm{Sc}$ on a long axial field-of-view PET/CT, and its central finding is that the isotope's favorable decay data do not translate into usable count statistics. In the three smallest phantom spheres the fitted oPs lifetimes are $2.65\\pm0.50$, $1.39\\pm0.20$, and $1.76\\pm0.18$ ns, and a single $4\\times4\\times4$ mm$^3$ voxel in the largest sphere gives $1.79\\pm0.57$ ns; all are consistent with the literature water value of $1.839\\pm0.015$ ns. The relative standard deviation in the time-difference background (time differences below $-2.7$ ns) stays above 20% even for voxels inside the spheres, and the average per-voxel uncertainty on $\\tau_3$ across a slice is 0.53 ns. The authors conclude that, given the scanner's limited energy resolution and the current three-photon event selection, $^{44}\\mathrm{Sc}$ does not seem able to outperform $^{124}\\mathrm{I}$ in count statistics for oPs lifetime imaging.","pith_inferences":["A matched-activity, matched-duration head-to-head of $^{44}\\mathrm{Sc}$ and $^{124}\\mathrm{I}$ on the same scanner would separate isotope physics from count-rate differences, since the comparison here uses 40.68 kBq/ml over 20 minutes versus 252 kBq/ml over 15 minutes.","If the detector were upgraded to resolve 1157 keV photons, $^{44}\\mathrm{Sc}$'s roughly 94% prompt-gamma-per-positron branching ratio would make it the strongest candidate for oPs lifetime imaging, reversing the present conclusion.","The practice of discarding voxels with more than 20% background error, used in the $^{124}\\mathrm{I}$ study, may need to be replaced with explicit random-coincidence subtraction when high-energy prompt emitters are used."],"forward_implications":["On this scanner and with this event selection, $^{124}\\mathrm{I}$ remains the better radionuclide for oPs lifetime imaging, in spite of its far lower prompt-gamma branching ratio per positron.","Clinical translation of $^{44}\\mathrm{Sc}$ for positronium imaging will require hardware or event selection that can reject the unresolved 1157 keV photon, rather than simply more activity.","$^{43}\\mathrm{Sc}$, whose 372.9 keV prompt photon falls inside the detector's energy range, avoids this specific random-coincidence mechanism but offers no count-statistics advantage over $^{124}\\mathrm{I}$.","Random-coincidence estimation or time-thresholding reconstruction, as already proposed for other protocols, is a plausible route to recovering precision from $^{44}\\mathrm{Sc}$ data."],"supporting_citations":[{"why":"Supplies the prototype three-photon event selection and the scanner's 726 keV energy-resolution limit used throughout the measurement.","marker":"[33]"},{"why":"Defines the Bayesian fitting priors and provides the 124I comparison values (pSBR 55.5, 1.1% error in a 5 ml tube) that anchor the negative conclusion.","marker":"[49]"},{"why":"Establishes the in vivo oPs lifetime measurement methodology on a LAFOV PET/CT and the same dedicated software workflow.","marker":"[25]"},{"why":"Gives the literature water oPs lifetime (1.839±0.015 ns) against which all measured phantom lifetimes are checked.","marker":"[53]"},{"why":"Provides an alternative random-coincidence estimation and time-thresholding reconstruction cited as a possible route to reduce uncertainty for high-energy prompt photons.","marker":"[54]"}],"fun_headline_variants":["Scandium-44 lags iodine-124 in positronium lifetime stats","Sc-44 positronium imaging hindered by random coincidences","Scandium-44 positronium lifetimes: poor count statistics","Prompt photon noise limits Sc-44 positronium imaging","Sc-44 fails to beat I-124 in positronium lifetime counts"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central comparison assumes the $^{44}\\mathrm{Sc}$ and $^{124}\\mathrm{I}$ scans are directly comparable, even though the $^{124}\\mathrm{I}$ reference used about six times higher activity concentration (252 kBq/ml versus 40.68 kBq/ml) and a different scan duration (15 minutes versus 20 minutes).","fun_headline_variants_meta":{"raw":{"variants":["Scandium-44 lags iodine-124 in positronium lifetime stats","Sc-44 positronium imaging hindered by random coincidences","Scandium-44 positronium lifetimes: poor count statistics","Prompt photon noise limits Sc-44 positronium imaging","Sc-44 fails to beat I-124 in positronium lifetime counts"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000736,"raw_usage":{"total_tokens":3384,"prompt_tokens":1131,"completion_tokens":2253,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":747,"completion_tokens_details":{"reasoning_tokens":2160}},"tokens_in":747,"tokens_out":2253,"duration_ms":15727,"temperature":1.0,"reasoning_tokens":2160,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:00:25.923175+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run $^{44}\\mathrm{Sc}$ in the same phantom at the same 252 kBq/ml activity concentration and 15-minute scan duration used for the $^{124}\\mathrm{I}$ reference: if the voxel pSBR stays near 12.6 and the $\\tau_3$ error remains several times larger, the unresolved 1157 keV photon is confirmed as the cause; if its statistics match $^{124}\\mathrm{I}$, the comparison is confounded by count rate.","supporting_citations":[{"cited_title":"Positronium Lifetime Imaging with the Biograph Vision Quadra using 124I","cited_arxiv_id":"2501.04145","evidence_quote":"Defines the Bayesian fitting priors and provides the 124I comparison values (pSBR 55.5, 1.1% error in a 5 ml tube) that anchor the negative conclusion."},{"cited_title":"In Vivo Positronium Lifetime Measurements with a Long Axial Field-of-View PET/CT","cited_arxiv_id":null,"evidence_quote":"Establishes the in vivo oPs lifetime measurement methodology on a LAFOV PET/CT and the same dedicated software workflow."},{"cited_title":"Measurement of positron lifetime to probe the mixed molecular states of liquid water","cited_arxiv_id":null,"evidence_quote":"Gives the literature water oPs lifetime (1.839±0.015 ns) against which all measured phantom lifetimes are checked."}],"review_version":2}