Pith. sign in

REVIEW 5 major objections 5 minor 36 references

Foundation Model-Enabled Efficient Data Sampling (FEEDS): A label-efficient training strategy for pan-cancer, multi-tracer PET/CT datasets

T0 review · 5 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A single farthest-first pass over unlabeled PET/CT scans, using DINOv2 embeddings, produces a lesion segmentation model that matches full-label training with 70% fewer annotations.

desk verdict A practical label-efficient sampling strategy for PET/CT that mostly delivers on false-positive reduction and non-inferiority, but its 'outperforms random' claim overreaches on Dice and the selection criterion is not validated as a diversity proxy. read the letter →

arxiv 2608.11076 v1 pith:XYJCZC5Z submitted 2026-08-11 cs.CV

Biratal Raj Wagle , Bashirul Azam Biswas , Grant Chau , Matthew E. Maeder , Muhammad Azeem Arshad , Michael S. Leapman , James B. Yu , Indrani Bhattacharya This is my paper · ORCID
classification cs.CV
keywords PET/CTlesionsegmentationlabel-efficienttrainingfarthest-firstsamplingDINOv2embeddingswhole-bodyimagingactivelearningsemi-supervisednnU-Net
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

FEEDS is a one-step method for deciding which unlabeled PET/CT scans to annotate when building a whole-body lesion segmentation model. It claims that selecting the 20% of scans farthest from a fixed 10% labeled pool, using DINOv2 embeddings of maximum-intensity projections, closes most of the gap between limited labels and full labeling. On the AutoPET-III test set and two unseen datasets, the resulting model is statistically non-inferior to a model trained on 100% of the labels while using 70% fewer annotations. The gains are most consistent in false-negative volume and lesion-level sensitivity, and FEEDS also beats random sampling, DPP sampling, and pseudolabel semi-supervision at the same label budget. If the claim holds, it gives clinical groups a practical way to prioritize annotation effort in large unannotated PET/CT archives.

What carries the argument

The load-bearing object is a tracer-stratified farthest-first sampling rule in DINOv2 feature space. DINOv2 is a self-supervised vision model pretrained on 142 million natural images; here it encodes z-score-normalized maximum-intensity projections of each 3D PET scan into 768-dimensional embeddings. For each unlabeled scan $j$ of a tracer type, the rule computes $d_j = 1 - \max_{i \in L_t} \frac{z_j \cdot z_i}{\|z_j\| \|z_i\|}$, the minimum cosine distance to the labeled set $L_t$ of the same tracer, and selects the scans with the largest $d_j$. This greedy procedure fills gaps in the training distribution by adding cases least represented by the existing labels, and it preserves the FDG:PSMA ratio because selection is done separately per tracer. The selected cases are annotated and combined with the fixed 10% labeled pool to train a single nnU-Net model, with no iterative retraining or pretraining loop.

What would settle it

Compare FEEDS against random sampling on a cohort where each scan's true lesion count and segmentation difficulty are known; if the farthest-first additions do not systematically lower false-negative volume relative to random additions across repeated random seeds, then embedding distance is not selecting the informative lesions.

Watch

Extended reading notes

Core claim

FEEDS' central discovery is that diversity in a self-supervised embedding space, measured by the minimum cosine distance to a small labeled set, is a usable proxy for training informativeness in whole-body PET/CT lesion segmentation. For each tracer type, the method embeds every scan's maximum-intensity projection with DINOv2, computes $d_j$ as the distance to the labeled set, and annotates the scans with the largest $d_j$. Trained with 30% of the AutoPET-III labels (the fixed 10% pool plus the farthest 20%), the nnU-Net model matches the fully supervised model under the paper's non-inferiority margin, with lower false-positive volume and comparable or better lesion-level positive predictive value. Raw Dice on the AutoPET-III test set is slightly lower for FEEDS than for full training (0.61 vs. 0.64), but the difference falls within the margin the paper defines, and the effect generalizes to held-out AutoPET-III, Deep-PSMA, and an internal clinical dataset across FDG and PSMA tracers and multiple cancer types.

Load-bearing premise

The method assumes that how different an unlabeled scan looks from the already-labeled scans, as measured by distance in a self-supervised image embedding, is the same as how much that scan will teach the segmentation model; if that link breaks, farthest-first selection adds outliers instead of informative lesions.

Editorial extensions

If this is right

  • A model trained with FEEDS-selected 30% labels is non-inferior in Dice to a model trained with 100% labels on the AutoPET-III test set, and it reduces false-positive volume relative to random sampling.
  • FEEDS generalizes to unseen Deep-PSMA and internal clinical scans across FDG and PSMA tracers and multiple cancer types, with lesion-level sensitivity near or above random sampling.
  • At the same 30% label budget, FEEDS yields lower false-negative volume than random sampling on the AutoPET-III validation and test sets, meaning fewer whole lesions are missed.
  • FEEDS outperforms one-iteration pseudolabel-based semi-supervised learning, which produced noisy labels and worse Dice than training on the same amount of strongly labeled data.
  • The selection step runs once before any training, so the annotation queue can be produced with a single pass of feature extraction, unlike active learning or semi-supervised loops.

Reading between the lines

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

  • A reader could test whether the same farthest-first rule improves other dense prediction tasks, such as organ or tumor segmentation in CT or MRI, whenever a small labeled pool and a large unlabeled archive exist.
  • Because the paper uses only PET maximum-intensity projections, replacing them with 3D PET or combined PET+CT embeddings is a natural extension; the paper itself notes the 2D projection can lose spatial information, so such a test would reveal how much of the gain depends on that projection.
  • The same ranking could also be used to build a prioritized annotation queue for a clinical archive, letting radiologists annotate only the most informative scans; the paper frames this as a deployment benefit but does not measure the time saved.
  • A fair reading of 'matching 100%' is statistical non-inferiority within pre-specified margins rather than equal performance, since the raw Dice on the AutoPET-III test set is 0.61 for FEEDS versus 0.64 for full training.
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

5 major / 5 minor

Summary. The paper proposes FEEDS, a one-shot label-efficient data sampling strategy for whole-body PET/CT lesion segmentation. FEEDS computes DINOv2 embeddings of 2D maximum-intensity projections of PET volumes, measures the minimum cosine distance from each unlabeled scan to a fixed 10% labeled AutoPET-III training pool (Eq. 1), and selects the farthest 20% of scans per tracer for expert annotation. These 30% of labeled data are then used to train an nnU-Net, which is evaluated on a held-out AutoPET-III test set, the Deep-PSMA dataset, and an internal DHMC dataset with voxel-, lesion-, and anatomical-region-level metrics. The paper claims that FEEDS outperforms random sampling and matches fully-labeled (100%) training performance with 70% less annotation burden.

Significance. If the claims are fully supported, FEEDS would be practically valuable because it requires no iterative active-learning or semi-supervised training, uses a frozen general-purpose foundation model, and is evaluated on multiple external datasets with clinically oriented metrics. The inclusion of lesion-level and anatomical-region-level analysis, the comparison against DPP-based sampling and pseudo-label SSL, and public code release are strengths. However, the evidence in the manuscript supports a more limited claim: consistent and statistically significant improvement over random sampling is shown for false-positive volume, whereas Dice improvements are absent or reversed on two of three datasets, and the causal role of the proposed selection rule is not directly validated.

major comments (5)
  1. [§3.1.2, Eq. (1)] The central claim that the largest dj identifies the most informative unlabeled cases for segmentation is not validated. No experiment correlates dj with any per-case measure of segmentation error or expected improvement of the 10%-labeled model, no control experiment selects the closest cases instead of the farthest, and the distance is computed only to the fixed labeled set, not to previously selected cases, so the top 20% can be a mutually redundant tail of the embedding distribution rather than a diverse informative set. Please add a closest-case control and/or an analysis showing that selected cases cover lesion patterns that cause errors in the 10% model.
  2. [§4.4, Table 6] The statistical results do not support the broad claim that FEEDS outperforms random sampling. On AutoPET, Dice is identical (0.613 vs 0.613); on DeepPSMA, FEEDS Dice is lower than random (0.645 vs 0.653); only DH shows a Dice improvement, with p=0.050. False-positive volume is the only metric with consistently significant improvement across all datasets, while false-negative volume differences are not significant. The paper should reframe its central claim to a false-positive-volume reduction with non-inferiority in Dice and false-negative volume, and discuss why the diversity mechanism specifically reduces false positives.
  3. [§4.4, Table 6] The non-inferiority margins stated in the text (0.05 Dice, 5 cc) disagree with those reported in Table 6 (0.02 Dice, 2 mL). This is not a cosmetic discrepancy: DeepPSMA FNVol differs by 4.35 cc between FEEDS and random, which is within 5 cc but outside 2 mL, and Table 6 marks the NI test as failed (×). Also, the text says non-inferiority passes in the pooled analysis for all three metrics, but Table 6 shows a failure for DeepPSMA FNVol. Please reconcile the margins, report both analyses, and justify the chosen margin.
  4. [§2.1, §4.4, Table 6] The reported sample sizes are inconsistent. Section 2.1 states that the AutoPET-III test set has 321 scans, but Table 6 reports n=207 for AutoPET Dice and FNVol. If the Dice/FN analysis is limited to diseased cases only, that filtering is not described in the statistical analysis section, and the pooled n=430 (207+200+23) should be explained. These decisions affect the paired Wilcoxon tests and the strength of the non-inferiority claims.
  5. [§5, Limitations] The authors concede that the 2D MIP projection can lose spatial information and that DINOv2 is a natural-image model, but no experiment quantifies whether the embedding ranking preserves the information needed for 3D lesion segmentation. Given that the only significant advantage over random sampling is lower false-positive volume, it remains plausible that the selection preferentially oversamples negative or atypical scans rather than error-prone lesion patterns. A focused analysis of what the selected cases actually contain—lesion counts, sizes, tracers, scanners—would help establish the mechanism claimed in the paper.
minor comments (5)
  1. [§3.1.1, Figure 2] The text states that the DINOv2 feature representation is f ∈ R^768, but Figure 2 refers to a 784-dimensional feature space; these numbers should be reconciled.
  2. [§4.1, Table 2] The narrative says FEEDS performs consistently better than random sampling, but at the 10+10% budget FEEDS Dice equals the random mean (0.637 vs 0.637), and at 10+20% FEEDS FPVol is higher than the random mean (18.98 vs 18.33); the phrasing should be softened to reflect the actual pattern.
  3. [§4.4, Table 6] The abstract and conclusion state that FEEDS outperforms random sampling on all evaluated metrics, but Table 6 shows no significant Dice or FNVol differences on AutoPET and DeepPSMA; the abstract and conclusion should be revised to match the evidence.
  4. [§4.3, Table 5] On the DHMC cohort, the model trained on only 10% fixed labeled data achieves the highest Dice and lowest FPVol, with FEEDS second. This counterintuitive result is not discussed beyond the 100% model; the authors should explain why the 10% model outperforms both FEEDS and random on this small cohort.
  5. [§4.4] The statistical analysis says a two-sided Wilcoxon signed-rank test with FDR-BH correction is used for FEEDS versus random comparisons, but the exact pairing structure and the handling of ties in Dice values (where many per-case scores may be equal) are not described; a sentence clarifying the test procedure would strengthen reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No material circularity: FEEDS's selection rule is defined on frozen DINOv2 embeddings, and the reported results come from held-out supervised segmentation, so the central claim is not an input to the method.

full rationale

The derivation chain in FEEDS is self-contained. Equation (1) computes a cosine distance between frozen DINOv2 embeddings of unlabeled cases and a fixed, randomly chosen 10% labeled pool; it contains no segmentation-model output, no fitted parameter, and no term derived from the Dice or false-positive metrics it is later compared against. The selected cases are then annotated and used to train an nnU-Net, and all headline claims are evaluated on held-out AutoPET-III, DeepPSMA, and DHMC test sets. There is therefore no fitted-input-called-prediction pattern and no self-definitional equivalence between the sampling criterion and the evaluation metrics. The only mild judgment call is that the 10+20% budget was selected using the AutoPET-III validation set, but this is model selection rather than circularity, and the held-out test results are reported separately. One self-citation appears in support of the general claim that DINOv2 represents medical images well [5], but that claim is corroborated by an independent citation [2] and is not load-bearing for the core derivation; even if removed, Eq. (1) and the training pipeline would be unchanged. The Section 5 limitation that 2D MIP projections may lose spatial information is an external validity concern about the informativeness proxy, not a circularity, because the proxy is assumed rather than fitted to the outcome. No uniqueness theorem, ansatz import, or renaming of a known result is used as evidence. Hence the paper does not reduce to its inputs or to a self-citation chain.

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

The method introduces no new physical or model entities. Its main free parameter is the labeled data budget selected on validation. The axioms are domain assumptions about transferability of DINOv2 embeddings, the value of farthest-first selection, and reliance on external label quality.

free parameters (3)
  • Labeled data budget X% (10+20%) = 30% total (10% fixed + 20% FEEDS-selected)
    Chosen from validation ablations (Section 4.1); the headline '70% less annotation burden' depends on this budget selection.
  • Non-inferiority margins = 0.05 Dice; 5 cc FP/FN
    Set by authors in Section 4.4 without clinical justification; they determine whether FEEDS 'matches' 100%.
  • Training epochs = 280
    Fixed per Rokuss et al. [26]; no sensitivity analysis in the paper.
assumptions (4)
  • domain assumption DINOv2 features from 2D PET MIPs are informative for selecting 3D segmentation-relevant diversity
    Core premise of Section 3.1.1; acknowledged in Section 5 as 2D MIPs may lose spatial information.
  • ad hoc to paper Farthest-first selection yields a representative and informative labeling queue
    No theoretical proof; indirect support only from the validation ablation (Section 4.1).
  • domain assumption Per-tracer sampling preserves the needed FDG:PSMA balance
    Based on the observation that FDG and PSMA embeddings form distinct clusters (Section 3.1.2).
  • domain assumption Dataset labels from AutoPET-III and DeepPSMA are reliable ground truth
    Labels produced or verified by specialists [10,17,21]; the DHMC labels were made by a trainee and reviewed by a radiation oncologist.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Foundation Model-Enabled Efficient Data Sampling (FEEDS): A label-efficient training strategy for pan-cancer, multi-tracer PET/CT datasets." pith.science (2026). https://pith.science/paper/XYJCZC5Z

@misc{pith2026260811076,
  author       = {Pith},
  title        = {Pith review of: Foundation Model-Enabled Efficient Data Sampling (FEEDS): A label-efficient training strategy for pan-cancer, multi-tracer PET/CT datasets},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XYJCZC5Z}},
  note         = {Machine review of arXiv:2608.11076}
}
read the original abstract

Automated lesion segmentation in whole-body PET/CT imaging can assist clinicians with cancer detection, staging, and treatment planning across radiotracers and cancer types. However, training lesion segmentation models that capture variations in lesion size, distribution, and appearance requires large annotated datasets, whose creation is both time- and expertise-intensive. As a result, models trained on limited labeled PET/CT data often lack the accuracy and generalizability needed for clinical use. We present FEEDS (Foundation model-Enabled Efficient Data Sampling), a label- and compute-efficient learning strategy that uses vision foundation model embeddings to select the most informative and diverse unlabeled cases for expert annotation. Unlike unsupervised, semi-supervised, and active learning approaches, FEEDS is a one-step training paradigm requiring only a limited, representative training set, making it label- and compute-efficient. We train and validate FEEDS using the AutoPET-III dataset. We test its accuracy and generalizability on three held-out sets: AutoPET-III, DeepPSMA, and an internal Dartmouth-Hitchcock Medical Center dataset. We evaluate clinical utility at the voxel, lesion, and anatomic region level to assess performance in high-risk areas and treatment planning utility. FEEDS outperforms random-sampling-based labeling, pseudolabel-based semi-supervised learning, and training with limited labeled data alone. It generalizes across all three test sets, FDG and PSMA tracers, and multiple diseases, matching fully-labeled (100\%) training performance with 70\% less annotation burden. FEEDS addresses the challenge of label scarcity in an automatic lesion segmentation framework by providing a practical approach for constructing representative and diverse annotation queues from large, unannotated clinical repositories.

Figures

Figures reproduced from arXiv: 2608.11076 by the authors.

Figure 1
Figure 1. Overview of the FEEDS: (1) foundation model (DinoV2) feature extraction from PET maximum intensity [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. FEEDS samples additional cases for labeling by selecting the farthest cases (circled) from the labeled training [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Whole-body PET/CT lesion segmentation performance improves with increasing percentage of labeled [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Voxel-level evaluation on the AutoPET-III test set, comparing 10% , (10+20)% randomly sampled (five [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Anatomical region-level comparison of lesion segmentation with AutoPET-III test set, comparing 10% , [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Coronal view of two PSMA images with the segmented lesion mask (green) overlaid. Yellow boxes highlight [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Qualitative evaluation across three representative lesion regions. Columns 1 and 2 show the CT and PET [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Qualitative evaluation of FEEDS on a whole-body PET/CT scan. Columns 1 and 2 show the CT and PET [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 31 canonical work pages

  1. [1]

    Role of psma pet-guided metastases-directed therapy in oligometastatic recurrent prostate cancer.Frontiers in oncology, 12:929444, 2022

    Matthew Alberto, Arthur Yim, Nathan Papa, Shankar Siva, Joseph Ischia, Karim Touijer, James A Eastham, Damien Bolton, and Marlon Perera. Role of psma pet-guided metastases-directed therapy in oligometastatic recurrent prostate cancer.Frontiers in oncology, 12:929444, 2022

    work page 2022

  2. [2]

    Evaluating general purpose vision foundation models for medical image analysis: An experimental study of dinov2 on radiology benchmarks.arXiv preprint arXiv:2312.02366, 2023

    Mohammed Baharoon, Waseem Qureshi, Jiahong Ouyang, Yanwu Xu, Abdulrhman Aljouie, and Wei Peng. Evaluating general purpose vision foundation models for medical image analysis: An experimental study of dinov2 on radiology benchmarks.arXiv preprint arXiv:2312.02366, 2023

    arXiv 2023

  3. [3]

    Classification, staging and prognosis of lung cancer.European journal of radiology, 45(1):8–17, 2003

    CJ Beadsmoore and NJ Screaton. Classification, staging and prognosis of lung cancer.European journal of radiology, 45(1):8–17, 2003

    work page 2003

  4. [4]

    Applications of artificial intelligence in psma pet/ct for prostate cancer imaging

    Sarah Lindgren Belal, Sophia Frantz, David Minarik, Olof Enqvist, Erik Wikstr ¨om, Lars Edenbrandt, and Elin Tr¨ag˚ardh. Applications of artificial intelligence in psma pet/ct for prostate cancer imaging. InSeminars in nuclear medicine, volume 54, pages 141–149. Elsevier, 2024

    work page 2024

  5. [5]

    Aggressiveness classification of clear cell renal cell carcinoma using registration-independent radiology-pathology correlation learning.Medical physics, 52(1):300–320, 2025

    Indrani Bhattacharya, Karin Stacke, Emily Chan, Jeong Hoon Lee, Justin R Tse, Tie Liang, James D Brooks, Geoffrey A Sonn, and Mirabela Rusu. Aggressiveness classification of clear cell renal cell carcinoma using registration-independent radiology-pathology correlation learning.Medical physics, 52(1):300–320, 2025

    work page 2025

  6. [6]

    Active learning on medical image

    Angona Biswas, Nasim Md Abdullah Al, Md Shahin Ali, Ismail Hossain, Md Azim Ullah, and Sajedul Talukder. Active learning on medical image. InData Driven Approaches on Medical Imaging, pages 51–67. Springer, 2023

    work page 2023

  7. [7]

    Bone metastases.Abeloff’s clinical oncology, pages 809– 830, 2020

    Robert E Coleman, Janet Brown, and Ingunn Holen. Bone metastases.Abeloff’s clinical oncology, pages 809– 830, 2020

    work page 2020

  8. [8]

    Clini- copathological insights and management of liver metastases: Current advances and future perspectives.World Journal of Hepatology, 17(10):110026, 2025

    Noura AA Ebrahim, Thoraya A Farghaly, Ahmed Abdo El-Sherif, Ahmed M Fahmy, Moamen O Othman, Neveen S Tahoun, Omnia Mohamed Korany, Aya Arafat, Reham Oreaba, and Soliman MA Soliman. Clini- copathological insights and management of liver metastases: Current advances and future perspectives.World Journal of Hepatology, 17(10):110026, 2025

    work page 2025

Show all 36 references
  1. [9]

    Results from the autopet challenge on fully automated lesion segmentation in oncologic pet/ct imaging.Nature Machine Intelligence, 6(11):1396–1405, 2024

    Sergios Gatidis, Marcel Fr ¨uh, Matthias P Fabritius, Sijing Gu, Konstantin Nikolaou, Christian La Foug`ere, Jin Ye, Junjun He, Yige Peng, Lei Bi, et al. Results from the autopet challenge on fully automated lesion segmentation in oncologic pet/ct imaging.Nature Machine Intell...

    2024

  2. [10]

    A whole-body fdg-pet/ct dataset with manually annotated tumor lesions.Scientific Data, 9(1):601, 2022

    Sergios Gatidis, Tobias Hepp, Marcel Fr ¨uh, Christian La Foug`ere, Konstantin Nikolaou, Christina Pfannenberg, Bernhard Sch¨olkopf, Thomas K¨ustner, Clemens Cyran, and Daniel Rubin. A whole-body fdg-pet/ct dataset with manually annotated tumor lesions.Scientific Data, 9(1):601, 2022

    2022

  3. [11]

    Erin F Gillespie, Joanna C Yang, Noah J Mathis, Catherine B Marine, Charlie White, Zhigang Zhang, Christo- pher A Barker, Rupesh Kotecha, Alyson McIntosh, Max Vaynrub, et al. Prophylactic radiation therapy versus standard of care for patients with high-risk asymptomatic bone m...

    2024

  4. [12]

    Deep semi-supervised learning for medical image segmentation: A review.Expert Systems with Applications, 245:123052, 2024

    Kai Han, Victor S Sheng, Yuqing Song, Yi Liu, Chengjian Qiu, Siqi Ma, and Zhe Liu. Deep semi-supervised learning for medical image segmentation: A review.Expert Systems with Applications, 245:123052, 2024

    2024

  5. [13]

    How we read oncologic fdg pet/ct.Cancer Imaging, 16(1):35, 2016

    Michael S Hofman and Rodney J Hicks. How we read oncologic fdg pet/ct.Cancer Imaging, 16(1):35, 2016

    2016

  6. [14]

    C2maot: Cross-modal complementary masked autoencoder with optimal transport for cancer segmentation in pet-ct images

    Jiaju Huang, Shaobin Chen, Xinglong Liang, Xiao Yang, Zhuoneng Zhang, Yue Sun, Ying Wang, and Tao Tan. C2maot: Cross-modal complementary masked autoencoder with optimal transport for cancer segmentation in pet-ct images. InInternational Conference on Medical Image Computing an...

    2025

  7. [15]

    Au- tomated lesion segmentation in whole-body pet/ct - multitracer multicenter generalization

    Michael Ingrisch, Jakob Dexl, Katharina Jeblick, Clemens Cyran, Sergios Gatidis, and Thomas Kuestner. Au- tomated lesion segmentation in whole-body pet/ct - multitracer multicenter generalization. Zenodo, Version 1,

  8. [16]

    nnu-net: a self- configuring method for deep learning-based biomedical image segmentation.Nature methods, 18(2):203–211, 2021

    Fabian Isensee, Paul F Jaeger, Simon AA Kohl, Jens Petersen, and Klaus H Maier-Hein. nnu-net: a self- configuring method for deep learning-based biomedical image segmentation.Nature methods, 18(2):203–211, 2021

    2021

  9. [17]

    Jeblick et al

    K. Jeblick et al. A whole-body psma-pet/ct dataset with manually annotated tumor lesions (psma-pet-ct-lesions), 2024

    2024

  10. [18]

    A whole-body psma-pet/ct dataset with manu- ally annotated tumor lesions.Scientific Data, 13(1):1023, 2026

    Katharina Jeblick, Balthasar Schachtner, Andreas Mittermeier, Jakob Dexl, Philipp Wesp, Thomas K¨ustner, Ser- gios Gatidis, Marcel Fr¨uh, Matthias P Fabritius, Felix Herr, et al. A whole-body psma-pet/ct dataset with manu- ally annotated tumor lesions.Scientific Data, 13(1):1023, 2026

    2026

  11. [19]

    Gaohaer Kadeerhan, Bo Xue, Xiao-Lin Wu, Wei-Nan Chen, and Dong-Wen Wang. Incidence trends and survival of metastatic prostate cancer with bone and visceral involvement: 2010-2019 surveillance, epidemiology, and end results.Frontiers in Oncology, 13:1201753, 2023

    2010

  12. [20]

    Deep semisupervised transfer learning for fully automated whole-body tumor quantification and prognosis of cancer on pet/ct.Journal of Nuclear Medicine, 65(4):643–650, 2024

    Kevin H Leung, Steven P Rowe, Moe S Sadaghiani, Jeffrey P Leal, Esther Mena, Peter L Choyke, Yong Du, and Martin G Pomper. Deep semisupervised transfer learning for fully automated whole-body tumor quantification and prognosis of cancer on pet/ct.Journal of Nuclear Medicine, 6...

    2024

  13. [21]

    James Meakin, Paul K. Gerke, Sjoerd Kerkstra, Thomas Koopman, Anne Mickan, Chris van Run, Harm van Zeeland, Francesco Ciompi, Alessa Hering, Colin Jacobs, Nadieh Khalili, Peter Koopmans, Jeroen van der Laak, Geert Litjens, Silvan Quax, Clara I. S´anchez, Jos Tannhauser, Miriam...

    2025

  14. [22]

    Total-body pet/ct: challenges and opportunities

    Clemens Mingels, Kevin J Chung, Austin R Pantel, Axel Rominger, Ian Alberts, Benjamin A Spencer, Lorenzo Nardo, and Thomas Pyka. Total-body pet/ct: challenges and opportunities. InSeminars in nuclear medicine, volume 55, pages 21–30. Elsevier, 2025

    2025

  15. [23]

    Maxime Oquab, Timoth ´ee Darcet, Th ´eo Moutakanni, Huy V . V o, Marc Szafraniec, Vasil Khalidov, Pierre Fer- nandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, Mahmoud Assran, Nicolas Ballas, Wojciech Galuba, Russell Howes, Po-Yao Huang, Shang-Wen Li, Ishan Misra, Mi...

    2023 arXiv

  16. [24]

    Cross attention transformers for multi-modal unsupervised whole-body pet anomaly detection

    Ashay Patel, Petru-Daniel Tudosiu, Walter Hugo Lopez Pinaya, Gary Cook, Vicky Goh, Sebastien Ourselin, and M Jorge Cardoso. Cross attention transformers for multi-modal unsupervised whole-body pet anomaly detection. InMICCAI Workshop on Deep Generative Models, pages 14–23. Spr...

    2022

  17. [25]

    Outcomes of observation vs stereotactic ablative radiation for oligometastatic prostate cancer: the oriole phase 2 randomized clinical trial.JAMA oncology, 6(5):650–659, 2020

    Ryan Phillips, William Yue Shi, Matthew Deek, Noura Radwan, Su Jin Lim, Emmanuel S Antonarakis, Steven P Rowe, Ashley E Ross, Michael A Gorin, Curtiland Deville, et al. Outcomes of observation vs stereotactic ablative radiation for oligometastatic prostate cancer: the oriole p...

    2020

  18. [26]

    From fdg to psma: A hitchhiker’s guide to multitracer, multicenter lesion segmentation in pet/ct imaging.arXiv preprint arXiv:2409.09478, 2024

    Maximilian Rokuss, Balint Kovacs, Yannick Kirchhoff, Shuhan Xiao, Constantin Ulrich, Klaus H Maier-Hein, and Fabian Isensee. From fdg to psma: A hitchhiker’s guide to multitracer, multicenter lesion segmentation in pet/ct imaging.arXiv preprint arXiv:2409.09478, 2024

    2024 arXiv

  19. [27]

    Uncertainty quan- tification for deep learning-based metastatic lesion segmentation on whole body pet/ct.Physics in Medicine & Biology, 70(11):115009, 2025

    Brayden Schott, Victor Santoro-Fernandes, ˇZan Klaneˇcek, Scott Perlman, and Robert Jeraj. Uncertainty quan- tification for deep learning-based metastatic lesion segmentation on whole body pet/ct.Physics in Medicine & Biology, 70(11):115009, 2025

    2025

  20. [28]

    Segmentation of pet/ct lung cancer lesion images via a semi-supervised improved swinunet model.Biomedical Signal Processing and Control, 113:108797, 2026

    Siyuan Tang, Siriguleng Wang, Yue Miao, Naiyu Wang, Liang Liu, Yuhan Qu, Yu Gu, Qingyu Ji, Qiangqiang Bao, and Jinliang Zhao. Segmentation of pet/ct lung cancer lesion images via a semi-supervised improved swinunet model.Biomedical Signal Processing and Control, 113:108797, 20...

    2026

  21. [29]

    Jacob Trotter, Austin R Pantel, Boon-Keng Kevin Teo, Freddy E Escorcia, Taoran Li, Daniel A Pryma, and Neil K Taunk. Positron emission tomography (pet)/computed tomography (ct) imaging in radiation therapy treatment planning: a review of pet imaging tracers and methods to inco...

    2023

  22. [30]

    Liver metastases.Nature reviews Disease primers, 7(1):27, 2021

    Diamantis I Tsilimigras, Pnina Brodt, Pierre-Alain Clavien, Ruth J Muschel, Michael I D’Angelica, Itaru Endo, Rowan W Parks, Majella Doyle, Eduardo de Santibanes, and Timothy M Pawlik. Liver metastases.Nature reviews Disease primers, 7(1):27, 2021

    2021

  23. [31]

    Active learning with uncertainty-guided sample selection for pet/ct tumor segmentation

    Mahsa Vali, Hossein Arabi, and Katarzyna Bozek. Active learning with uncertainty-guided sample selection for pet/ct tumor segmentation. In2025 IEEE International Conference on Imaging Systems and Techniques (IST), pages 1–6. IEEE, 2025

    2025

  24. [32]

    Evaluation of the prognostic role of liver metastases on patient outcomes: systematic review and meta-analysis.The Cancer Journal, 29(5):279–284, 2023

    Jessica J Waninger, Vincent T Ma, Zoey Chopra, Ashley N Pearson, and Michael D Green. Evaluation of the prognostic role of liver metastases on patient outcomes: systematic review and meta-analysis.The Cancer Journal, 29(5):279–284, 2023

    2023

  25. [33]

    Automated segmentation of lesions and organs at risk on [68ga] ga-psma-11 pet/ct images using self-supervised learning with swin unetr

    Elmira Yazdani, Najme Karamzadeh-Ziarati, Seyyed Saeid Cheshmi, Mahdi Sadeghi, Parham Geramifar, Habibeh V osoughi, Mahmood Kazemi Jahromi, and Saeed Reza Kheradpisheh. Automated segmentation of lesions and organs at risk on [68ga] ga-psma-11 pet/ct images using self-supervise...

    2024

  26. [34]

    Fereshteh Yousefirizi, Isaac Shiri, Joo Hyun O, Ingrid Bloise, Patrick Martineau, Don Wilson, Franc ¸ois B´enard, Laurie H Sehn, Kerry J Savage, Habib Zaidi, et al. Semi-supervised learning towards automated segmentation of pet images with limited annotations: application to l...

    2024

  27. [35]

    Seganypet: Universal promptable segmentation from positron emission tomography images

    Yichi Zhang, Le Xue, Wenbo Zhang, Lanlan Li, Yuchen Liu, Chen Jiang, Yuan Cheng, and Yuan Qi. Seganypet: Universal promptable segmentation from positron emission tomography images. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 21107–21116, 2025. 17

    2025

  28. [2024]

    Available athttps://zenodo.org/records/10990932

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.