Uncertainty-Guided Coarse-to-Fine Tumor Segmentation with Anatomy-Aware Post-Processing

Curtis Lisle; Damla Turgut; David Mohaisen; Ilkin Sevgi Isler; Ulas Bagci

REVIEW 4 major objections 6 minor 21 references

Uncertainty-Guided Coarse-to-Fine Tumor Segmentation with Anatomy-Aware Post-Processing

T0 review · 4 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Filtering, not a bigger model, lifts tumor Dice from 0.47 to 0.64

desk verdict Sensible cascade, credible numbers, but the central claim that anatomy-aware filtering drives the gain is unproven because no ablation separates it from ROI cropping. read the letter →

arxiv 2504.12215 v2 pith:T4Y655LP submitted 2025-04-16 cs.CV cs.AI

Ilkin Sevgi Isler , David Mohaisen , Curtis Lisle , Damla Turgut , Ulas Bagci This is my paper · ORCID

classification cs.CVcs.AI

keywords tumorsegmentationcoarse-to-fineuncertainty-guidedanatomy-awarepost-processingcomponentselectionCTimagingMonteCarlodropoutcascade

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 argues that a fully automated two-stage pipeline—coarse whole-volume tumor segmentation followed by region-of-interest refinement—can substantially improve lung tumor delineation in CT, provided the coarse outputs are filtered with anatomical constraints before refinement. On the primary private dataset, the best configuration raises Swin UNETR Dice from 0.4690 to 0.6447 and cuts the 95th-percentile Hausdorff distance from 187.97 mm to 8.89 mm; smaller gains appear on the public NSCLC dataset. The authors attribute most of the gain to component selection: reducing the number of candidate regions forwarded to the second stage correlates strongly with better Dice and Hausdorff scores. Uncertainty-aware losses did not improve Dice or HD95 over a standard Dice-plus-cross-entropy loss, but they improved boundary calibration and interpretability. If the claim holds, existing segmentation models can be upgraded with a modular post-hoc pipeline without architectural changes.

What carries the argument

The load-bearing mechanism is the stage-one post-processing module combined with tight ROI cropping. Connected components in the coarse tumor mask are merged by 3D dilation, then filtered by anatomical criteria: volumetric overlap with a lung mask (components below 80% overlap are rejected unless they lie within 5 voxels of the lung surface and contain more than 50 voxels), followed by optional top-k selection. Surviving components are cropped with zero or sixteen voxel margins and passed to a second-stage model trained with the adaptive loss $L_{\text{adaptive}} = \alpha(x)\,L_{\text{Dice}} + (1-\alpha(x))\,L_{\text{CE}}$ with $\alpha(x)=\exp(-U(x))$, where $U(x)$ is the variance of Monte Carlo dropout forward passes. The filtering acts as what the authors call a spatial attention mechanism: fewer, larger, lung-adjacent candidates make the ROI model's job easier and lower predictive uncertainty.

What would settle it

Run the pipeline on a held-out set rich in chest-wall, pleural, and multifocal tumors and count true positive components that fail both retention criteria (overlap at or below 80% and either distance to lung surface above 5 voxels or volume at or below 50 voxels). If the filter removes any ground-truth tumor component, recall drops and the claim that filtering preserves clinically relevant tumors is falsified; a cleaner quantitative test is comparing filtered versus unfiltered recall on that subset.

Watch

Extended reading notes

Core claim

The paper's central claim is that treating segmentation as a surrogate for localization, then refining only the winning candidate region after anatomy-aware filtering, yields more accurate and clinically plausible tumor masks than full-volume segmentation alone. On the primary dataset, the fully automated Top-1, ROI-0margin pipeline lifts Swin UNETR from Dice 0.4690 and HD95 187.97 mm to Dice 0.6447 and HD95 8.89 mm; on the NSCLC dataset the same pipeline improves Swin UNETR Dice from 0.4731 to 0.4860. The authors further report a strong monotone correlation between component-count reduction and HD95 improvement ($\rho = -0.83$, $p < 0.0001$), which they read as evidence that the post-processing filter, not the second-stage network, drives most of the gain.

Load-bearing premise

The load-bearing premise is that every true tumor component will pass at least one retention rule—more than 80% lung overlap, or adjacency within 5 voxels of the lung surface combined with size above 50 voxels—so that no clinically meaningful tumor is discarded before stage two.

Editorial extensions

If this is right

Top-1 component selection beats Top-2 and All Valid on the primary dataset, so for single-dominant tumors extra components behave as noise; multifocal cases must be handled by relaxing top-k or explicitly detecting satellite lesions.
Tight ROI-0margin crops outperform ROI-16margin crops, meaning the second-stage model benefits from reduced surrounding context once the component is well localized.
The strong negative correlation between component count and Dice/HD95 implies that a stage-one model can be improved mainly by filtering its false-positive components, without retraining.
The framework generalizes to the public NSCLC dataset but with smaller gains, indicating that lesion size and focality modulate the benefit of component filtering.
Because standard Dice-plus-cross-entropy loss matches or beats the uncertainty-aware loss on Dice/HD95, the practical contribution of uncertainty modeling here is calibration and interpretability, not headline metric improvement.

Reading between the lines

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

A testable extension beyond the paper is a pipeline with stage-one filtering but no stage-two ROI model, which would isolate how much of the Dice gain comes from filtering alone versus ROI refinement.
The 80%-overlap, 5-voxel, and 50-voxel thresholds are likely dataset-specific; on cohorts with many chest-wall or mediastinal tumors, the retention rule will discard true extrapulmonary components, so a threshold-sweep calibration curve is the natural next experiment.
The correlation between component count and quality may be confounded by tumor difficulty: small isolated tumors are both easier to segment and naturally produce one component, so component count could partly be a proxy for tumor size or site.
The MC-dropout uncertainty maps could be reused beyond loss reweighting, for example to spatially weight the final mask or to flag uncertain boundaries for clinician review, which the paper only gestures at.

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.

Desk Editor's Note

read the letter

Readable, honest, and the pipeline is a reasonable combination of known pieces: coarse full-CT segmentation, lung-mask-informed filtering (overlap, surface distance, size), top-k component selection, then a second-stage ROI model, with MC-dropout uncertainty reweighting in the loss. The specific recipe is new as a whole, and they provide code and evaluate on two datasets with three backbones. I give them credit for reporting that the uncertainty-aware loss did not improve Dice (Table V) and for a sensitivity check on the size threshold.

The soft spot is the one the stress-test flags, and it is real. Table I compares the single-stage full-CT baseline to the complete two-stage pipeline. There is no condition that keeps the ROI refinement while removing the Section III-F filters, or vice versa. So the improvement from 0.469 to 0.6447 Dice could come mostly from the cropping and higher-resolution refinement; the anatomy-aware filtering might be neutral or even harmful. Table III's correlation between component reduction and performance is confounded—cases with fewer components are probably easier tumors. The paper also lacks patient counts, error bars, and split details, and it claims better 'spatial calibration' from the uncertainty-aware loss without reporting any calibration metric. In fact, Table V shows Dice drops slightly with that loss. The filtering thresholds are partly hand-set and only partially sensitivity-tested; the 'All Valid' condition in Table II badly degrades HD95, which makes me wonder how much of the benefit is top-k selection alone.

The weakest assumption the reader flags is legitimate: components with low lung overlap and no near-surface adjacency are discarded, which could drop true chest-wall or multifocal tumors. The paper acknowledges this in Section V but doesn't quantify the sensitivity cost.

Verdict: the central two-stage claim is plausible but not yet established. It deserves peer review, not a desk reject, because the method is directly usable, the experiments are concrete, and the missing ablation is fixable. The referee should ask for (i) an ablation with and without the post-processing filters, (ii) patient-level statistics, and (iii) either a calibration metric or a softened claim. If the authors deliver those, this becomes a solid niche contribution.

Referee Report

4 major / 6 minor

Summary. The paper presents a two-stage, fully automated tumor segmentation framework for thoracic CT. A first-stage full-volume model produces a coarse prediction; connected components are then filtered with anatomy-aware heuristics (lung-overlap threshold, distance-to-lung-surface rule, minimum component size, and optional top-K selection) to form ROIs. A second-stage ROI model, optionally trained with an uncertainty-aware loss, produces the final segmentation. The authors report that on a private Orlando dataset the fully automated Top-1, ROI-0margin pipeline improves Swin UNETR from Dice 0.469/HD95 187.97 mm to Dice 0.6447/HD95 8.89 mm (Table I), with smaller gains on the public NSCLC-Radiomics dataset (Table VI). They also report a negative correlation between remaining component count and performance (Table III) and state that the uncertainty-aware loss improves spatial calibration.

Significance. If the headline results hold, the framework offers a practical, modular, post-hoc refinement strategy that can be grafted onto existing segmentation models, with publicly released code and a reproducible experimental configuration on a public dataset. The large reported gain on the private dataset is clinically meaningful and the anatomical filtering rules are transparent. However, the current evidence is not yet sufficient: the improvement is not isolated from ROI cropping, no error bars or patient counts support the quantitative claims, and the uncertainty component is not shown to deliver the claimed calibration benefit. With additional ablations and statistical reporting, this could become a useful contribution to cascaded medical-segmentation pipelines.

major comments (4)

[Section IV, Table I; Section III-F] The central claim that anatomy-aware post-processing contributes to the headline gains is not supported because no ablation isolates the Section III-F filtering from the second-stage ROI refinement. The comparison in Table I contrasts the full pipeline (coarse model + III-F filters + fine ROI model) with a full-CT baseline. A condition that keeps the ROI model but removes the lung-overlap, surface-distance, and size filters (e.g., selecting the largest component of the raw coarse prediction) is necessary to attribute the improvement to anatomy-aware filtering rather than to cropping and high-resolution refinement alone. Without such a condition, the Table I gains could be produced entirely by the second stage.
[Section IV, Tables I-VII] No error bars, confidence intervals, or patient counts are reported for any quantitative result, and the size of the private Orlando dataset is not stated anywhere in the manuscript. The headline Dice change from 0.469 to 0.6447 therefore cannot be distinguished from run-to-run variance, especially given the much smaller improvement on NSCLC in Table VI (SwinUNETR 0.4731 to 0.4860). The authors should report the number of test cases, per-case distributions, and paired significance tests (or per-patient bootstrap CIs) for the main comparisons.
[Section III-D, Table V] The claim that the uncertainty-aware loss improves spatial calibration is unsupported. In Table V the uncertainty-aware loss gives slightly worse Dice (0.6321 vs 0.6447) and HD95 (8.92 vs 8.89) than the standard loss, yet the text states it 'improved spatial calibration and boundary interpretability.' No calibration metric (e.g., expected calibration error, reliability diagram, Brier score) is reported anywhere. Either remove this claim or provide a quantitative calibration evaluation on held-out data.
[Section IV, Table III] The correlation evidence in Table III is confounded and is overinterpreted. The statement that 'these findings confirm that anatomical and uncertainty-informed postprocessing ... enhances overall accuracy' does not follow from a correlation between component count and performance, because cases with fewer connected components are plausibly easier tumors (small, isolated, well-defined) regardless of post-processing. To support the causal claim, the authors should compare per-case final Dice with and without the post-processing filters, or otherwise show that the filters specifically remove components that would lower the final segmentation quality.

minor comments (6)

[Section V] The phrase 'followed bb conclusion' appears in the organization paragraph; this is a typo for 'followed by the conclusion.'
[Section III-D] The adaptive loss equation is not numbered, and U(x) is described only as 'the variance of multiple forward passes'; the number of MC dropout samples and the variance computation (per-voxel across samples?) should be specified.
[Table V] The row label 'S4 ROI Model' is unexplained; the architecture and training details of this model should be defined in the caption or text.
[Table VI] The notation 'K=1' in the Orlando Health rows is not defined; the caption should explain that K denotes the number of retained components after post-processing.
[Section IV, Table IV] The text says the comparison includes 'the LungMask library [18], auto-generated masks, and human-annotated ground truth,' but Table IV appears to list only two rows ('Original Lung Mask' and 'New Lung Mask'). Please align the table with the described comparisons.
[Section IV and V] The sensitivity analysis reported in the text covers component size thresholds from 50 to 150 voxels, but the claim in Section V that 'changes in lung overlap and voxel size thresholds had minimal impact' is not fully supported; no lung-overlap threshold sweep is shown. Please report the overlap-threshold sensitivity or qualify the claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the empirical gains are not constructed from their own inputs.

full rationale

The paper makes no formal derivation that reduces to its inputs. The headline result (Table I: Swin UNETR Dice 0.4690 -> 0.6447, HD95 187.97 -> 8.89 mm) is an empirical comparison between a single-stage full-volume baseline and a two-stage pipeline with ROI cropping, component filtering, and optional uncertainty-aware loss; it is not obtained by plugging the target metric back into the method. The adaptive loss alpha(x)=exp(-U(x)) in Section III-D is an explicit design choice, not fitted to reproduce the reported Dice/HD95 values, and Table V shows it does not improve those metrics, so it cannot be a self-fulfilling component. The post-processing thresholds (80% lung overlap, 5-voxel surface distance, 50-voxel size cutoff) are hand-set clinical heuristics accompanied by sensitivity analysis, and the paper itself concedes generalizability concerns in Section V: 'the NSCLC dataset, with larger or multifocal lesions, was more sensitive to ROI and component thresholds' and there are 'potential concerns about the generalizability of heuristic post-processing.' Those are validity/attribution caveats, not circularity. The only self-citation, [17], is cited as background for MC-Swin-U and dropout-based uncertainty; it is not load-bearing because MC dropout is independently introduced via Gal and Ghahramani [14] and because the uncertainty-aware loss is ablated and found not to drive the main gains. The lack of an ablation that isolates the anatomy-aware filtering from ROI refinement is an experimental attribution gap, but it does not make any prediction equivalent to a fitted parameter or to a self-citation chain.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new physical or biological entities are proposed. The central empirical claim rests on hand-set anatomical thresholds, the accuracy of lung masks, and the validity of MC dropout as an uncertainty estimate. These are domain assumptions rather than fitted constants in a derivation.

free parameters (5)

lung overlap retention threshold = 80%
Section III-F.2: components with less than 80% lung overlap are generally discarded; stricter for mediastinum, not quantified. Sensitivity not reported.
distance to lung surface threshold = <=5 voxels
Section III-F.3: low-overlap components are retained if within 5 voxels of the lung boundary and above the size threshold.
minimum component voxel count = 50 voxels
Section IV: components with 50 or fewer voxels are discarded; sensitivity 50 to 150 changes Dice by less than 0.002, above 150 excludes true tumors.
component count selected for second stage = Top-1 (single largest)
Section III-C and Table II: Top-1 gave the best Dice and HD95; Top-2 and All Valid degraded performance.
uncertainty weighting function = alpha(x)=exp(-U(x))
Section III-D: adaptive loss weight defined by MC-dropout variance; chosen ad hoc and not derived, and it did not improve Dice in Table V.

assumptions (4)

domain assumption Lung masks from LungMask library are accurate enough for filtering
Section III-A and F use a pretrained lung segmentation model; Table IV compares masks and shows small differences, but assumes mask errors do not remove true tumors.
domain assumption Most true tumors are either inside the lung or, if extrapulmonary, within 5 voxels of the lung surface and larger than 50 voxels
Section III-F.3 and Conclusion: these clinical priors define which components are kept; if false, sensitivity drops for chest-wall, mediastinal, or multifocal tumors.
domain assumption MC dropout variance approximates predictive uncertainty
Section III-D builds U(x) from variance of MC-dropout forward passes, following Kendall and Gal and prior self-cited work; no calibration check is provided.
domain assumption Connected components of the coarse prediction correspond to candidate tumors
Section III-B and C: component analysis operates on the coarse binary map, so a tumor missed in stage one cannot be recovered by stage two.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Uncertainty-Guided Coarse-to-Fine Tumor Segmentation with Anatomy-Aware Post-Processing." pith.science (2026). https://pith.science/paper/T4Y655LP

@misc{pith2026250412215,
  author       = {Pith},
  title        = {Pith review of: Uncertainty-Guided Coarse-to-Fine Tumor Segmentation with Anatomy-Aware Post-Processing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/T4Y655LP}},
  note         = {Machine review of arXiv:2504.12215}
}

read the original abstract

Reliable tumor segmentation in thoracic computed tomography (CT) remains challenging due to boundary ambiguity, class imbalance, and anatomical variability. We propose an uncertainty-guided, coarse-to-fine segmentation framework that combines full-volume tumor localization with refined region-of-interest (ROI) segmentation, enhanced by anatomically aware post-processing. The first-stage model generates a coarse prediction, followed by anatomically informed filtering based on lung overlap, proximity to lung surfaces, and component size. The resulting ROIs are segmented by a second-stage model trained with uncertainty-aware loss functions to improve accuracy and boundary calibration in ambiguous regions. Experiments on private and public datasets demonstrate improvements in Dice and Hausdorff scores, with fewer false positives and enhanced spatial interpretability. These results highlight the value of combining uncertainty modeling and anatomical priors in cascaded segmentation pipelines for robust and clinically meaningful tumor delineation. On the Orlando dataset, our framework improved Swin UNETR Dice from 0.4690 to 0.6447. Reduction in spurious components was strongly correlated with segmentation gains, underscoring the value of anatomically informed post-processing.

Figures

Figures reproduced from arXiv: 2504.12215 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 19 canonical work pages

[1]

The lung image database con- sortium (lidc) and image database resource initiative (idri): a completed reference database of lung nodules on ct scans,

S. G. Armato III, G. McLennan, L. Bidaut, M. F. McNitt-Gray, C. R. Meyer, A. P. Reeves, B. Zhao et al. , “The lung image database con- sortium (lidc) and image database resource initiative (idri): a completed reference database of lung nodules on ct scans,”Medical physics, vol. 38, no. 2, pp. 915–931, 2011

work page 2011
[2]

Lung tumor segmentation with missing tumor labels,

Z. Zhou, M. M. R. Siddiquee, N. Tajbakhsh, and J. Liang, “Lung tumor segmentation with missing tumor labels,”Medical Image Analysis, vol. 64, p. 101713, 2020

work page 2020
[3]

Fully automatic one-step segmentation of pulmonary tumors from multi-source heterogeneous ct imaging using deep convolutional neural networks,

Q. Tong, G. Song, C. Yang, and et al., “Fully automatic one-step segmentation of pulmonary tumors from multi-source heterogeneous ct imaging using deep convolutional neural networks,” Physics in Medicine & Biology, vol. 65, no. 22, p. 225010, 2020

work page 2020
[4]

Lung tumor segmentation on ct scans using boundary-aware neural networks,

H. Tang, Y . Xu, Y . Zhang, and et al., “Lung tumor segmentation on ct scans using boundary-aware neural networks,” Medical Image Analysis, vol. 75, p. 102307, 2022

work page 2022
[5]

nnu-net: Self-adapting framework for u-net-based medical image segmentation,

F. Isensee, P. F. Jaeger, S. A. Kohl, J. Petersen, and K. H. Maier- Hein, “nnu-net: Self-adapting framework for u-net-based medical image segmentation,” in Nature Methods, vol. 18, no. 2, 2021, pp. 203–211

work page 2021
[6]

Second opinion needed: communicating uncertainty in medical machine learning,

B. Kompa, J. Snoek, and A. L. Beam, “Second opinion needed: communicating uncertainty in medical machine learning,” NPJ Digital Medicine, vol. 4, no. 1, pp. 1–6, 2021

work page 2021
[7]

Deep learning tech- niques for medical image segmentation: achievements and challenges,

M. H. Hesamian, W. Jia, X. He, and P. Kennedy, “Deep learning tech- niques for medical image segmentation: achievements and challenges,” Journal of Digital Imaging , vol. 32, no. 4, pp. 582–596, 2019

work page 2019
[8]

U-net: Convolutional networks for biomedical image segmentation,

O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” MICCAI, pp. 234–241, 2015

work page 2015

Show all 21 references

[9]

Unetr: Transformers for 3d medical image segmentation,

A. Hatamizadeh, V . Nath, Y . Tang, D. Yang, A. Myronenko, B. Land- man, H. R. Roth, and D. Xu, “Unetr: Transformers for 3d medical image segmentation,” IEEE/CVF WACV, pp. 574–584, 2022

2022
[10]

Self-supervised pre-training of swin transformers for 3d medical image analysis,

Y . Tang, D. Yang, W. Li, H. R. Roth, B. Landman, D. Xu, V . Nath, and A. Hatamizadeh, “Self-supervised pre-training of swin transformers for 3d medical image analysis,” in CVPR, 2022, pp. 20 730–20 740

2022
[11]

H-denseunet: Hybrid densely connected unet for liver and tumor segmentation from ct volumes,

X. Li, H. Chen, X. Qi, Q. Dou, C.-W. Fu, and P.-A. Heng, “H-denseunet: Hybrid densely connected unet for liver and tumor segmentation from ct volumes,” in IEEE TMI, vol. 37, no. 12, 2018, pp. 2663–2674

2018
[12]

Cascaded unet for kidney tumor segmentation,

Y . Zhou and et al., “Cascaded unet for kidney tumor segmentation,” in MICCAI Kidney Tumor Segmentation Challenge (KiTS) , 2019

2019
[13]

What uncertainties do we need in bayesian deep learning for computer vision?

A. Kendall and Y . Gal, “What uncertainties do we need in bayesian deep learning for computer vision?” NeurIPS, vol. 30, 2017

2017
[14]

Dropout as a bayesian approximation: Representing model uncertainty in deep learning,

Y . Gal and Z. Ghahramani, “Dropout as a bayesian approximation: Representing model uncertainty in deep learning,” in ICML, 2016, pp. 1050–1059

2016
[15]

Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks,

G. Wang and et al., “Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks,” Neurocomputing, vol. 338, pp. 34–45, 2019

2019
[16]

Confidence calibration and predictive uncertainty estimation for deep medical image segmentation,

A. Mehrtash, W. M. Wells III, C. M. Tempany, P. Abolmaesumi, and T. Kapur, “Confidence calibration and predictive uncertainty estimation for deep medical image segmentation,” in MICCAI, 2020, pp. 676–685

2020
[17]

Self-supervised learning for organs at risk and tumor segmentation with uncertainty quantification,

I. Isler, D. Jha, C. Lisle, J. Rineer, P. Kelly, B. Aydogan, M. Abazeed, D. Turgut, and U. Bagci, “Self-supervised learning for organs at risk and tumor segmentation with uncertainty quantification,” in ICECCME. IEEE, 2023, pp. 1–6

2023
[18]

Automatic lung segmentation in routine imaging is primarily a data diversity problem, not a methodology problem,

J. Hofmanninger, F. Prayer, J. Pan, S. R ¨ohrich, H. Prosch, and G. Langs, “Automatic lung segmentation in routine imaging is primarily a data diversity problem, not a methodology problem,” European radiology experimental, vol. 4, pp. 1–13, 2020

2020
[19]

Automatic liver and tumor segmentation of ct and mri volumes using cascaded fully convolutional neural networks,

X. Yang, S. Sun, D. Liang, S. Zhang, and D. Feng, “Automatic liver and tumor segmentation of ct and mri volumes using cascaded fully convolutional neural networks,” in arXiv preprint arXiv:1704.02703 , 2017

2017 arXiv
[20]

A survey on deep learning in medical image analysis,

G. Litjens, T. Kooi, B. E. Bejnordi, A. A. A. Setio, F. Ciompi, M. Ghafoorian, J. A. van der Laak, B. van Ginneken, and C. I. S ´anchez, “A survey on deep learning in medical image analysis,” Medical Image Analysis, vol. 42, pp. 60–88, 2017

2017
[21]

Common limitations of performance metrics in biomedical image analysis,

A. Reinke and et al., “Common limitations of performance metrics in biomedical image analysis,” Nature Communications, vol. 12, no. 1, p. 6048, 2021

2021

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

The lung image database con- sortium (lidc) and image database resource initiative (idri): a completed reference database of lung nodules on ct scans,

S. G. Armato III, G. McLennan, L. Bidaut, M. F. McNitt-Gray, C. R. Meyer, A. P. Reeves, B. Zhao et al. , “The lung image database con- sortium (lidc) and image database resource initiative (idri): a completed reference database of lung nodules on ct scans,”Medical physics, vol. 38, no. 2, pp. 915–931, 2011

work page 2011

[2] [2]

Lung tumor segmentation with missing tumor labels,

Z. Zhou, M. M. R. Siddiquee, N. Tajbakhsh, and J. Liang, “Lung tumor segmentation with missing tumor labels,”Medical Image Analysis, vol. 64, p. 101713, 2020

work page 2020

[3] [3]

Fully automatic one-step segmentation of pulmonary tumors from multi-source heterogeneous ct imaging using deep convolutional neural networks,

Q. Tong, G. Song, C. Yang, and et al., “Fully automatic one-step segmentation of pulmonary tumors from multi-source heterogeneous ct imaging using deep convolutional neural networks,” Physics in Medicine & Biology, vol. 65, no. 22, p. 225010, 2020

work page 2020

[4] [4]

Lung tumor segmentation on ct scans using boundary-aware neural networks,

H. Tang, Y . Xu, Y . Zhang, and et al., “Lung tumor segmentation on ct scans using boundary-aware neural networks,” Medical Image Analysis, vol. 75, p. 102307, 2022

work page 2022

[5] [5]

nnu-net: Self-adapting framework for u-net-based medical image segmentation,

F. Isensee, P. F. Jaeger, S. A. Kohl, J. Petersen, and K. H. Maier- Hein, “nnu-net: Self-adapting framework for u-net-based medical image segmentation,” in Nature Methods, vol. 18, no. 2, 2021, pp. 203–211

work page 2021

[6] [6]

Second opinion needed: communicating uncertainty in medical machine learning,

B. Kompa, J. Snoek, and A. L. Beam, “Second opinion needed: communicating uncertainty in medical machine learning,” NPJ Digital Medicine, vol. 4, no. 1, pp. 1–6, 2021

work page 2021

[7] [7]

Deep learning tech- niques for medical image segmentation: achievements and challenges,

M. H. Hesamian, W. Jia, X. He, and P. Kennedy, “Deep learning tech- niques for medical image segmentation: achievements and challenges,” Journal of Digital Imaging , vol. 32, no. 4, pp. 582–596, 2019

work page 2019

[8] [8]

U-net: Convolutional networks for biomedical image segmentation,

O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” MICCAI, pp. 234–241, 2015

work page 2015