Pith. sign in

REVIEW 4 major objections 6 minor 56 references

Improving Vessel Segmentation with Multi-Task Learning and Auxiliary Data Available Only During Model Training

T0 review · 4 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read An auxiliary contrast-enhanced MRI sequence available only during training still improves vessel segmentation from non-contrast MRI, and the gain is largest when expert annotations are scarce.

desk verdict Well-executed medical-imaging study with a genuinely new setup, but the central claim that the auxiliary contrast-enhanced modality drives the gains is not isolated from simply adding unlabeled training data. read the letter →

arxiv 2509.03975 v2 pith:PYSDN3JN submitted 2025-09-04 cs.CV

Daniel Sobotka , Alexander Herold , Matthias Perkonigg , Lucian Beer , Nina Bastati , Alina Sablatnig , Ahmed Ba-Ssalamah , Georg Langs This is my paper
classification cs.CV
keywords multi-tasklearningvesselsegmentationliverMRIauxiliarycontrast-enhancedsemi-supervisedY-Netbraintumordeep
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

The paper tries to establish that a richer imaging modality can be used during model training and then discarded at deployment, while still improving predictions made from the routine modality. In this test case, a multi-task network segments liver vessels from non-contrast T1w MRI by simultaneously learning to translate T1w into contrast-enhanced T1wce MRI during training; the T1wce images are never needed at test time. With identical annotations, this raises the Dice overlap score from 0.446 for a U-Net to 0.506 for the multi-task Y-Net, and adding 30 unannotated T1w/T1wce pairs pushes it to 0.539 (p<0.01). The advantage is strongest when only one or a few annotated volumes exist—for a single annotation, Dice rises from 0.242 to 0.372. The same recipe improves brain tumor segmentation from T2w MRI when T1wce is the training-only auxiliary, suggesting the strategy is not specific to one organ or sequence.

What carries the argument

The Y-Net architecture: a 3D U-Net with one shared encoder and two task-specific decoders, one for vessel labels (T_S) and one for translating the input T1w image into a contrast-enhanced T1wce image (T_T). The two decoders exchange features at matching spatial resolutions through Neural Discriminative Dimensionality Reduction (NDDR) layers—1x1x1 convolutions that learn how to fuse concatenated decoder features. A homoscedastic-uncertainty loss with learned weights balances segmentation cross-entropy and translation mean-squared error. During training, these components let the model use both annotated triplets and unannotated image pairs; at inference, only the non-contrast image enters and

What would settle it

Train the same Y-Net with identical unannotated pairs but replace the auxiliary T1wce target with a task known to be irrelevant to vessels—for example, translating the image to a scrambled or unrelated output while keeping the same loss and data size. If the Dice gain over the U-Net persists, the benefit comes from the extra training signal rather than from the informative contrast-enhanced modality; if the gain disappears, task compatibility is necessary.

Watch

Extended reading notes

Core claim

The central claim is that auxiliary contrast-enhanced MRI, available only during training, improves segmentation of vessels in non-contrast MRI, and does so most when annotated examples are scarce. A Y-Net with a shared encoder and two decoders—one predicting vessel labels, one predicting the contrast-enhanced volume—learns both tasks together. Features flow between the two decoders through layerwise fusion layers, so the translation task acts as a training-time teacher for the segmentation task. Unannotated pairs of non-contrast and contrast-enhanced images can be added to the training set, and they make the segmentation better even though no vessel labels come with them. On the liver datas

Load-bearing premise

The load-bearing premise is that translating a non-contrast image into its contrast-enhanced version teaches features that also help find vessels; if the translation and segmentation tasks are not actually compatible, the gains rest on extra data alone and the paper's explanation for them collapses.

Editorial extensions

If this is right

  • If the claim is right, contrast-enhanced sequences can be treated as a training-only resource: hospitals can keep routine non-contrast protocols at deployment and still get better vessel segmentation than a model trained on non-contrast data alone.
  • The method directly addresses the low-annotation regime: with a single annotated liver volume, the multi-task model with 30 unannotated pairs reaches Dice 0.372 versus 0.242 for the U-Net, and the gap shrinks as annotations increase.
  • Unannotated paired images without labels provide most of the benefit: Y-Net-mix adding 30 such pairs outperforms the plain Y-Net and matches or beats pretraining-plus-fine-tuning (Y-Net-TL) in cross-validation, while on the hold-out set only Y-Net-mix beats the U-Net.
  • The benefit transfers across domains: the same architecture and training recipe improve brain tumor segmentation from T2w MRI using T1wce as the auxiliary modality, indicating the approach is not tuned to liver vessels.
  • Improvements are not uniform across vessel sizes: the largest Dice gains occur for vessels above 10 mm in diameter, while small vessels (0–5 mm) remain difficult and still need more annotated examples.

Reading between the lines

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

  • The underlying principle should generalize beyond contrast/non-contrast MRI pairs: any auxiliary imaging modality that reveals structure relevant to the target label—such as a later CT phase, a higher-resolution sequence, or an alternative contrast mechanism—could be used the same way, provided the two tasks are compatible.
  • A direct testable extension would be to compare this multi-task approach against self-training or pseudo-labeling on the same unannotated pairs; if the Y-Net wins, the shared-task structure, not merely the extra images, is carrying the gain.
  • The paper does not isolate how much of the benefit comes from the NDDR fusion layers versus simply pretraining the encoder on translation; an ablation that removes the cross-decoder connections would separate those two contributions.
  • Clinical deployment of the idea would require checking that the model's improved recall does not come with silently worse precision for the structures a radiologist cares about; the reported precision trade-off makes this worth measuring per vessel diameter.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes a multi-task learning framework, Y-Net, for liver vessel segmentation from non-contrast T1w MRI. A shared encoder feeds two decoders: one for vessel segmentation and one for predicting contrast-enhanced T1wce images, with NDDR feature fusion and an uncertainty-weighted loss. Unannotated T1w/T1wce pairs are exploited either by joint training (Y-Net-mix) or by pretraining/fine-tuning (Y-Net-TL). Experiments on liver MRI (12 annotated patients in 3-fold CV plus 3 hold-out cases, with up to 30 unannotated pairs) and on a public brain tumor dataset (8 annotated, 30 auxiliary cases) report Dice improvements, especially with few annotations. The central claim is that an auxiliary contrast-enhanced modality available only during training improves segmentation through shared task structure.

Significance. If the claimed effect is real, the practical value is high: it would allow models to benefit from richer imaging modalities during training while deploying on cheaper, more widely acquired non-contrast sequences, exactly the setting the authors target. The paper has clear strengths: the code is released, the brain-tumor validation uses a public dataset, the vessel-diameter analysis is informative, and the authors explicitly acknowledge several limitations (small cohort, unstable uncertainty weighting). However, as detailed below, the experimental design does not currently isolate the contribution of the auxiliary contrast-enhanced modality from the contribution of simply adding unlabeled data, and the primary non-confounded MTL comparison is not statistically significant. The central attribution to 'shared task structure' therefore needs an additional control experiment.

major comments (4)
  1. [§3.4/Table 3 and §2.2] The headline quantitative claim (Dice 0.506→0.539, p<0.01) is confounded: Y-Net-mix uses 30 additional unlabeled T1w/T1wce volumes, while both U-Net and Y-Net use zero unlabeled volumes. In the unlabeled phase only the shared encoder and translation decoder are trained (§2.2), so the gain may simply reflect encoder pretraining on extra T1w data. There is no control using the same 30 unlabeled T1w volumes with an auxiliary task that does not require T1wce (e.g., T1w reconstruction/inpainting or a self-supervised T1w-only objective). Without such a control, the abstract's statement that 'the feature representation benefits from the shared task structure' is unsupported. The same confound applies to the brain experiment (§3.9). This is the central missing experiment.
  2. [§3.3/Table 2] The only experiment that isolates multi-task learning from additional unlabeled data is the Y-Net vs U-Net comparison on the same 12 annotated patients, which gives Dice 0.506 vs 0.446 (p=0.09) and MSD 4.615 vs 3.929 (p=0.26). This is not statistically significant. The significant p<0.01 gains in Tables 3 and 4 all come from models that also use 5–30 unlabeled pairs—exactly the confounded comparison above. Please report paired per-fold/per-patient statistics, confidence intervals, and specify the test used; as written, the central claim rests on a non-significant primary comparison plus confounded auxiliary-data comparisons.
  3. [§3.4 and Eq. (1)] In the Y-Net-mix experiment the uncertainty weights σ_S and σ_T are not learned; they are fixed to values 'learned in the first experiment' because the automated search is unstable under class imbalance. This means the reported Y-Net-mix results do not actually exercise the proposed adaptive loss balancing. The fixed values are selected from a model trained on the same small 12-patient cohort, not an independent validation set, so there is a risk of subtle selection bias. Please disclose the fixed numeric values, justify their transfer to different annotation/pair counts, and report sensitivity to these values (e.g., ±20%) or retrain in a way that makes the uncertainty weights stable.
  4. [§3.8/Table 6] On the hold-out test set, Y-Net-TL (30 auxiliary pairs) is worse than the U-Net baseline (Dice 0.456 vs 0.474), and Y-Net-mix improves only modestly (0.504). The paper attributes the Y-Net-TL failure to one outlier in the >15 mm group, but Table 7 shows Y-Net-TL is also lower than U-Net in the 10–15 mm group (0.52 vs 0.53) and only marginally better in the small-vessel groups. This contradicts the broad statement in §4 that 'an auxiliary training modality improves segmentation accuracy' when framed as a general cross-domain claim. Please temper the generalization claim or provide a detailed analysis of why Y-Net-TL fails on hold-out data; as written, the evidence does not support the conclusion that auxiliary-data transfer is universally beneficial.
minor comments (6)
  1. [Table 1] 'Adams optimizer' should be 'Adam optimizer' (also used inconsistently with Section 3.2).
  2. [§3.7/Table 5] The mean surface distance for U-Net in the 10–15 mm group is reported as 'inf'. Please state how undefined values (e.g., empty predictions) are handled in the aggregate statistics and whether this affects the reported means.
  3. [§3.2–§3.3] No statistical test is described for any p-value. Specify whether paired t-test, Wilcoxon signed-rank, or another test is used, and whether the unit of analysis is per-fold, per-subject, or per-volume. Also clarify whether the same folds are used across all methods and unlabeled-pair counts.
  4. [§3.2] The split into 12 patients for 3-fold cross-validation and 3 hold-out patients is stated only later in the text. Please state this split clearly before the experiments and explain how the 15 annotated triplets are partitioned.
  5. [§3.6/Fig. 4] In Fig. 4(a), the x-axis is labeled 'Annotated volumes used for training', but the figure caption says 'cases'. Please be explicit that in this experiment the unlabeled-pair count is fixed at 30 for Y-Net-mix and Y-Net-TL, so the comparison across annotation counts is not confounded by pair count.
  6. [§4] The final paragraph on portal hypertension, HVPG, and clinical outlook is interesting but reads as disconnected from the presented experiments. Consider moving it to a separate 'clinical implications' section or shortening it.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the reported segmentation improvements are empirical comparisons, not derived from the loss function or from self-citations.

full rationale

I walked the paper's derivation chain. The central claim is that multi-task training with an auxiliary contrast-enhanced modality improves segmentation on non-contrast input. This claim is supported by direct experimental comparisons (Tables 2-6, 8) on held-out and external data, not by an analytic derivation. Equation (1) is a standard homoscedastic-uncertainty loss weighting; it does not by construction guarantee Y-Net > U-Net, and the observed Dice differences are measured, not implied by the equation. The learned task weights sigma_S and sigma_T are calibrated in Experiment 1 and then fixed in Experiment 2; this is parameter transfer, not a fitted input renamed as a prediction, and the reported Dice values are independent of the sigma values except through training dynamics. The task-compatibility assumption in Section 1.3 is cited from external work [30] and is explicitly an assumption, not a self-citation or a theorem used to force the result. No uniqueness theorem is invoked, and no result is justified solely by a citation to the authors' own prior work. Flagged limitations: Section 3.4 admits 'The multi-task loss function and automated search for optimal sigma_S and sigma_T is not stable when the number of training samples for each task is substantially different. Therefore, we fixed sigma_S and sigma_T to values learned in the first experiment,' and Section 4 admits 'the number of annotated images from different patients is small.' These are validity concerns, not circularity. The skeptic's concern that Y-Net-mix also receives 30 additional unlabeled volumes, so the gain may reflect extra data rather than the auxiliary modality, is an experimental confound or alternative explanation, not a reduction of the prediction to its input by construction. No equation is equivalent to another by definition, and no fitted parameter is presented as a prediction. The paper is self-contained against external benchmarks (brain tumor data) and reports hold-out results. Therefore, no significant circularity is present; score 0.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The paper introduces no new physical or conceptual entities. Its central claim rests on a standard deep-learning architecture, a task-compatibility assumption for multi-task learning, and a set of data-processing assumptions about image registration and label transfer. The main fitted parameters are the task loss weights, which are learned and then fixed in one experiment.

free parameters (3)
  • Task loss weights sigma_S and sigma_T = Learned during training; fixed to values from experiment 1 for Y-Net-mix
    Equation 1 weights segmentation and translation losses. Section 3.4 states the automated search is unstable under sample imbalance and sigma values are fixed, so part of the reported gain is conditional on this calibration.
  • Network architecture hyperparameters = Learning rate 0.001, batch size 1, patch size 128x96x64, 100 epochs, 40 fine-tune epochs
    These choices come from prior U-Net and Y-Net practice; they are not fitted to the data but do affect all reported numbers.
  • Vessel thickness bin thresholds = 0-5 mm, 5-10 mm, 10-15 mm, >15 mm
    Hand-chosen cutoffs in Section 3.2 used for stratified Dice evaluation; they influence how improvements are attributed to small versus large vessels.
assumptions (6)
  • domain assumption Paired T1w and T1wce volumes are co-registered so voxel-wise pairing is valid.
    Section 3.1 says the segmentation and T1wce were registered to the T1w image. If registration is imperfect, the translation target and the training labels are misaligned.
  • domain assumption Radiologist vessel annotations made on contrast-enhanced images transfer to non-contrast images after registration.
    The ground truth L_i is drawn on T1wce but is used to supervise segmentation of T1w. This assumes the registered label map is a valid target for the non-contrast image.
  • ad hoc to paper Image translation from T1w to T1wce shares a feature representation predictive of vessel location.
    The whole NDDR fusion and multi-task loss assumes the auxiliary synthesis task and the segmentation task are compatible. The paper cites general MTL compatibility results but provides no task-specific evidence beyond the experimental outcome.
  • standard math Homoscedastic uncertainty weighting is a valid way to balance segmentation and translation losses.
    Equation 1 follows Kendall et al. The paper itself notes in Section 3.4 that this weighting is not stable when task sample sizes differ substantially.
  • domain assumption Voxel overlap metrics (Dice, Jaccard, mean surface distance) capture clinically meaningful segmentation accuracy.
    All conclusions are based on these metrics; no downstream clinical task is evaluated.
  • domain assumption 3-fold cross-validation on 12 patients plus a 3-case hold-out generalizes to the chronic liver disease population.
    The paper lists the small number of annotated patients as a limitation in the Discussion, but the central claim depends on this small sample.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Improving Vessel Segmentation with Multi-Task Learning and Auxiliary Data Available Only During Model Training." pith.science (2026). https://pith.science/paper/PYSDN3JN

@misc{pith2026250903975,
  author       = {Pith},
  title        = {Pith review of: Improving Vessel Segmentation with Multi-Task Learning and Auxiliary Data Available Only During Model Training},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PYSDN3JN}},
  note         = {Machine review of arXiv:2509.03975}
}
read the original abstract

Liver vessel segmentation in magnetic resonance imaging data is important for the computational analysis of vascular remodelling, associated with a wide spectrum of diffuse liver diseases. Existing approaches rely on contrast enhanced imaging data, but the necessary dedicated imaging sequences are not uniformly acquired. Images without contrast enhancement are acquired more frequently, but vessel segmentation is challenging, and requires large-scale annotated data. We propose a multi-task learning framework to segment vessels in liver MRI without contrast. It exploits auxiliary contrast enhanced MRI data available only during training to reduce the need for annotated training examples. Our approach draws on paired native and contrast enhanced data with and without vessel annotations for model training. Results show that auxiliary data improves the accuracy of vessel segmentation, even if they are not available during inference. The advantage is most pronounced if only few annotations are available for training, since the feature representation benefits from the shared task structure. A validation of this approach to augment a model for brain tumor segmentation confirms its benefits across different domains. An auxiliary informative imaging modality can augment expert annotations even if it is only available during training.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

56 extracted references · 53 canonical work pages

  1. [1]

    INTRODUCTION Contrast enhanced magnetic resonance imaging (MRI) is used to assess fine grained soft tissue properties when study- ing diffuse liver diseases and their progress to liver cirrhosis. The difficulty of standardization, the risk and time constraints * Equal contribution associated with contrast enhanced imaging [1] limit the range of sequences ...

    work page Pith review arXiv 2025

  2. [2]

    METHODOLOGY We propose a shared encoder multi-task learning approach for a 3D U-Net architecture [27] for semantic segmentation. The network consists of a Y-Net [41] with a shared encoder 2 copy and concatenate 2 32 64 128 256 128 64 32 2x(Conv+ReLU+GN) MaxPooling+2x(Conv+ReLU+GN) Upsampling+Concatenate+2x(Conv+ReLU+GN) Conv Softmax 3D non-contrast enhanc...

    work page

  3. [3]

    EXPERIMENTS We evaluated if multi-task learning can improve the accuracy of segmentation models by comparing the performance of dif- ferent training strategies to exploit privileged data available only during training to models trained only on data also avail- able during test time. 3.1. Data and Model We studied MR image volumes of45chronic liver disease...

    work page

  4. [4]

    An auxiliary modality available only during training improves the segmentation accuracy of a Y- Net model when applied to new data, even if it is not avail- able during test time

    DISCUSSION We propose a 3D multi-task learning approach for medical image segmentation. An auxiliary modality available only during training improves the segmentation accuracy of a Y- Net model when applied to new data, even if it is not avail- able during test time. We show that providing such privileged informative data during training has benefits espe...

    work page

  5. [5]

    If less than8annotations are 6 Table 5

    showed decreased vessel to volume ratio mean difference with increasing vessel size between ground truth and the Y- Net + 30 T1w/T1wce pairs. If less than8annotations are 6 Table 5. Dice, mean surface distance, recall and precision related to the vessel thickness between baseline U-Net, proposed Y-Net, Y-Net-mix trained with additional 30 T1w/T1wce traini...

    work page

  6. [6]

    Does the functional liver imaging score derived from gadoxetic acid–enhanced MRI predict outcomes in chronic liver disease?,

    Nina Bastati, Lucian Beer, Mattias Mandorfer, Sarah Poetter-Lang, Dietmar Tamandl, Yesim Bican, Michael Christoph Elmer, Henrik Einspieler, Georg Semmler, Benedikt Simbrunner, et al., “Does the functional liver imaging score derived from gadoxetic acid–enhanced MRI predict outcomes in chronic liver disease?,”Radiology, vol. 294, no. 1, pp. 98–107, 2020

    work page 2020

  7. [7]

    Revisiting the risks of MRI with gadolinium based contrast agents—review of literature and guidelines,

    Aurang Z Khawaja, Deirdre B Cassidy, Julien Al Shakarchi, Damian G McGrogan, Nicholas G Inston, and Robert G Jones, “Revisiting the risks of MRI with gadolinium based contrast agents—review of literature and guidelines,”Insights into imaging, vol. 6, no. 5, pp. 553–558, 2015

    work page 2015

  8. [8]

    Ash and nash,

    F Scaglioni, S Ciccia, M Marino, G Bedogni, and S Bel- lentani, “Ash and nash,”Digestive Diseases, vol. 29, no. 2, pp. 202–210, 2011

    work page 2011

Show all 56 references
  1. [9]

    Synergy between nafld and afld and potential biomarkers,

    Raj Lakshman, Ruchi Shah, Karina Reyes-Gordillo, and Ravi Varatharajalu, “Synergy between nafld and afld and potential biomarkers,”Clinics and research in hep- atology and gastroenterology, vol. 39, pp. S29–S34, 2015

    2015

  2. [10]

    Hepatic vessels segmentation using deep learning and preprocessing enhancement,

    Omar Ibrahim Alirr and Ashrani Aizzuddin Abd Rahni, “Hepatic vessels segmentation using deep learning and preprocessing enhancement,”Journal of applied clini- cal medical physics, vol. 24, no. 5, pp. e13966, 2023

    2023

  3. [11]

    An automated liver tumour segmenta- tion from abdominal ct scans for hepatic surgical plan- ning,

    Omar Ibrahim Alirr, Ashrani Aizzuddin Abd Rahni, and Ehsan Golkar, “An automated liver tumour segmenta- tion from abdominal ct scans for hepatic surgical plan- ning,”International journal of computer assisted radi- ology and surgery, vol. 13, pp. 1169–1176, 2018

    2018

  4. [12]

    Hepatic vessel segmentation using vari- ational level set combined with non-local robust statis- tics,

    Siyu Lu, Hui Huang, Ping Liang, Gang Chen, and Liang Xiao, “Hepatic vessel segmentation using vari- ational level set combined with non-local robust statis- tics,”Magnetic resonance imaging, vol. 36, pp. 180– 186, 2017

    2017

  5. [13]

    Compu- tational methods for liver vessel segmentation in med- ical imaging: A review,

    Marcin Ciecholewski and Michał Kassja ´nski, “Compu- tational methods for liver vessel segmentation in med- ical imaging: A review,”Sensors, vol. 21, no. 6, pp. 2027, 2021

    2027

  6. [14]

    Multiscale vessel enhance- ment filtering,

    Alejandro F Frangi, Wiro J Niessen, Koen L Vincken, and Max A Viergever, “Multiscale vessel enhance- ment filtering,” inInternational conference on med- 11 ical image computing and computer-assisted interven- tion. Springer, 1998, pp. 130–137

    1998

  7. [15]

    Three-dimensional multi-scale line filter for segmentation and visualization of curvilinear structures in medical images,

    Yoshinobu Sato, Shin Nakajima, Nobuyuki Shiraga, Hideki Atsumi, Shigeyuki Yoshida, Thomas Koller, Guido Gerig, and Ron Kikinis, “Three-dimensional multi-scale line filter for segmentation and visualization of curvilinear structures in medical images,”Medical image analysis, vo...

    1998

  8. [16]

    Design and validation of a tool for neurite tracing and analysis in fluorescence mi- croscopy images,

    Erik Meijering, M Jacob, J-CF Sarria, Pl Steiner, H Hirling, and M Unser, “Design and validation of a tool for neurite tracing and analysis in fluorescence mi- croscopy images,”Cytometry Part A: the journal of the International Society for Analytical Cytology, vol. 58, no. 2, ...

    2004

  9. [17]

    Retinal ves- sel segmentation using the 2-d gabor wavelet and su- pervised classification,

    Jo ˜ao VB Soares, Jorge JG Leandro, Roberto M Cesar, Herbert F Jelinek, and Michael J Cree, “Retinal ves- sel segmentation using the 2-d gabor wavelet and su- pervised classification,”IEEE Transactions on medical Imaging, vol. 25, no. 9, pp. 1214–1222, 2006

    2006

  10. [18]

    Robust liver vessel extraction us- ing 3d u-net with variant dice loss function,

    Qing Huang, Jinfeng Sun, Hui Ding, Xiaodong Wang, and Guangzhi Wang, “Robust liver vessel extraction us- ing 3d u-net with variant dice loss function,”Computers in biology and medicine, vol. 101, pp. 153–162, 2018

    2018

  11. [19]

    Au- tomatic liver vessel segmentation using 3d region grow- ing and hybrid active contour model,

    Ye-zhan Zeng, Sheng-hui Liao, Ping Tang, Yu-qian Zhao, Miao Liao, Yan Chen, and Yi-xiong Liang, “Au- tomatic liver vessel segmentation using 3d region grow- ing and hybrid active contour model,”Computers in bi- ology and medicine, vol. 97, pp. 63–73, 2018

    2018

  12. [20]

    Accurate liver vessel segmentation via active contour model with dense vessel candidates,

    Minyoung Chung, Jeongjin Lee, Jin Wook Chung, and Yeong-Gil Shin, “Accurate liver vessel segmentation via active contour model with dense vessel candidates,” Computer methods and programs in biomedicine, vol. 166, pp. 61–75, 2018

    2018

  13. [21]

    Automatic segmentation methods for liver and hepatic vessels from ct and mri volumes, applied to the couinaud scheme,

    Marie-Ange Lebre, Antoine Vacavant, Manuel Grand- Brochier, Hugo Rositi, Armand Abergel, Pascal Chabrot, and Benoit Magnin, “Automatic segmentation methods for liver and hepatic vessels from ct and mri volumes, applied to the couinaud scheme,”Computers in biology and medicine,...

    2019

  14. [22]

    A novel method to model hepatic vascular network using vessel segmenta- tion, thinning, and completion,

    Xiaoyu Guo, Ruoxiu Xiao, Tao Zhang, Cheng Chen, Jiayu Wang, and Zhiliang Wang, “A novel method to model hepatic vascular network using vessel segmenta- tion, thinning, and completion,”Medical & biological engineering & computing, pp. 1–16, 2020

    2020

  15. [23]

    Combining deep learning with anatomical analysis for segmentation of the portal vein for liver sbrt planning,

    Bulat Ibragimov, Diego Toesca, Daniel Chang, Albert Koong, and Lei Xing, “Combining deep learning with anatomical analysis for segmentation of the portal vein for liver sbrt planning,”Physics in Medicine & Biology, vol. 62, no. 23, pp. 8943, 2017

    2017

  16. [24]

    Training liver vessel segmentation deep neural net- works on noisy labels from contrast ct imaging,

    Minfeng Xu, Yu Wang, Ying Chi, and Xiansheng Hua, “Training liver vessel segmentation deep neural net- works on noisy labels from contrast ct imaging,” in 2020 IEEE 17th International Symposium on Biomed- ical Imaging (ISBI). IEEE, 2020, pp. 1552–1555

    2020

  17. [25]

    Segmen- tation of vascular regions in ultrasound images: A deep learning approach,

    Deepak Mishra, Santanu Chaudhury, Mukul Sarkar, Sidharth Manohar, and Arvinder Singh Soin, “Segmen- tation of vascular regions in ultrasound images: A deep learning approach,” in2018 IEEE International Sympo- sium on Circuits and Systems (ISCAS). IEEE, 2018, pp. 1–5

    2018

  18. [26]

    Vesselnet: A deep convo- lutional neural network with multi pathways for robust hepatic vessel segmentation,

    Titinunt Kitrungrotsakul, Xian-Hua Han, Yutaro Iwamoto, Lanfen Lin, Amir Hossein Foruzan, Wei Xiong, and Yen-Wei Chen, “Vesselnet: A deep convo- lutional neural network with multi pathways for robust hepatic vessel segmentation,”Computerized Medical Imaging and Graphics, vol. ...

    2019

  19. [27]

    for liver vessel segmentation in contrast enhanced MRI and CT images. 1.3. State of the art in multi-task learning (MTL) MTL aims to improve generalization performance of multiple related tasks by leveraging useful shared information of these tasks [28, 29]. In contrast to tra...

  20. [28]

    Topnet: Topol- ogy preserving metric learning for vessel tree recon- struction and labelling,

    Deepak Keshwani, Yoshiro Kitamura, Satoshi Ihara, Satoshi Iizuka, and Edgar Simo-Serra, “Topnet: Topol- ogy preserving metric learning for vessel tree recon- struction and labelling,” inInternational Conference on Medical Image Computing and Computer-Assisted In- tervention. S...

    2020

  21. [29]

    Mr-to-us reg- istration using multiclass segmentation of hepatic vas- culature with a reduced 3d u-net,

    Bart R Thomson, Jasper N Smit, Oleksandra V Ivashchenko, Niels FM Kok, Koert FD Kuhlmann, Theo JM Ruers, and Matteo Fusaglia, “Mr-to-us reg- istration using multiclass segmentation of hepatic vas- culature with a reduced 3d u-net,” inInternational Con- ference on Medical Image...

    2020

  22. [30]

    An attention-guided deep neu- ral network with multi-scale feature fusion for liver ves- sel segmentation,

    Qingsen Yan, Bo Wang, Wei Zhang, Chuan Luo, Wei Xu, Zhengqing Xu, Yanning Zhang, Qinfeng Shi, Liang Zhang, and Zheng You, “An attention-guided deep neu- ral network with multi-scale feature fusion for liver ves- sel segmentation,”IEEE Journal of Biomedical and Health Informatics, 2020

    2020

  23. [31]

    Segmentation of hepatic vessels from MRI images for planning of electroporation-based treatments in the liver,

    Marija Marcan, Denis Pavliha, Maja Marolt Music, Igor Fuckan, Ratko Magjarevic, and Damijan Miklav- cic, “Segmentation of hepatic vessels from MRI images for planning of electroporation-based treatments in the liver,”Radiology and oncology, vol. 48, no. 3, pp. 267– 281, 2014. 12

    2014

  24. [32]

    Vessel segmentation from abdominal magnetic reso- nance images: adaptive and reconstructive approach,

    Evgin Goceri, Zarine K Shah, and Metin N Gurcan, “Vessel segmentation from abdominal magnetic reso- nance images: adaptive and reconstructive approach,” International journal for numerical methods in biomed- ical engineering, vol. 33, no. 4, pp. e2811, 2017

    2017

  25. [33]

    3d u-net: learning dense volumetric segmentation from sparse annotation,

    ¨Ozg¨un C ¸ ic ¸ek, Ahmed Abdulkadir, Soeren S Lienkamp, Thomas Brox, and Olaf Ronneberger, “3d u-net: learning dense volumetric segmentation from sparse annotation,” inInternational conference on medical image computing and computer-assisted intervention. Springer, 2016, pp. 424–432

    2016

  26. [34]

    Multitask learning,

    Rich Caruana, “Multitask learning,”Machine learning, vol. 28, no. 1, pp. 41–75, 1997

    1997

  27. [35]

    A survey on multi-task learning,

    Yu Zhang and Qiang Yang, “A survey on multi-task learning,”IEEE Transactions on Knowledge and Data Engineering, 2021

    2021

  28. [36]

    Which tasks should be learned together in multi-task learn- ing?,

    Trevor Standley, Amir R Zamir, Dawn Chen, Leonidas Guibas, Jitendra Malik, and Silvio Savarese, “Which tasks should be learned together in multi-task learn- ing?,”arXiv preprint arXiv:1905.07553, 2019

    1905 arXiv

  29. [37]

    Multi- task learning for brain tumor segmentation,

    Leon Weninger, Qianyu Liu, and Dorit Merhof, “Multi- task learning for brain tumor segmentation,” inInterna- tional MICCAI Brainlesion Workshop. Springer, 2019, pp. 327–337

    2019

  30. [38]

    Multi-task deep learning based CT imag- ing analysis for covid-19 pneumonia: Classification and segmentation,

    Amine Amyar, Romain Modzelewski, Hua Li, and Su Ruan, “Multi-task deep learning based CT imag- ing analysis for covid-19 pneumonia: Classification and segmentation,”Computers in Biology and Medicine, p. 104037, 2020

    2020

  31. [39]

    A survey on multi-task learning,

    Yu Zhang and Qiang Yang, “A survey on multi-task learning,”arXiv preprint arXiv:1707.08114, 2017

    2017 arXiv

  32. [40]

    Cross-stitch networks for multi-task learning,

    Ishan Misra, Abhinav Shrivastava, Abhinav Gupta, and Martial Hebert, “Cross-stitch networks for multi-task learning,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 3994–4003

    2016

  33. [41]

    Revis- iting multi-task learning in the deep learning era,

    Simon Vandenhende, Stamatios Georgoulis, Marc Proesmans, Dengxin Dai, and Luc Van Gool, “Revis- iting multi-task learning in the deep learning era,”arXiv preprint arXiv:2004.13379, 2020

    2004 arXiv

  34. [42]

    Multi- task learning using uncertainty to weigh losses for scene geometry and semantics,

    Alex Kendall, Yarin Gal, and Roberto Cipolla, “Multi- task learning using uncertainty to weigh losses for scene geometry and semantics,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 7482–7491

    2018

  35. [43]

    Gradnorm: Gradient normaliza- tion for adaptive loss balancing in deep multitask net- works,

    Zhao Chen, Vijay Badrinarayanan, Chen-Yu Lee, and Andrew Rabinovich, “Gradnorm: Gradient normaliza- tion for adaptive loss balancing in deep multitask net- works,” inInternational Conference on Machine Learn- ing. PMLR, 2018, pp. 794–803

    2018

  36. [44]

    Dynamic task prioritization for multitask learning,

    Michelle Guo, Albert Haque, De-An Huang, Serena Yeung, and Li Fei-Fei, “Dynamic task prioritization for multitask learning,” inProceedings of the Euro- pean Conference on Computer Vision (ECCV), 2018, pp. 270–287

    2018

  37. [45]

    End-to-end multi-task learning with attention,

    Shikun Liu, Edward Johns, and Andrew J Davison, “End-to-end multi-task learning with attention,” inPro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2019, pp. 1871–1880

    2019

  38. [46]

    Joint left atrial segmentation and scar quantification based on a dnn with spatial encoding and shape at- tention,

    Lei Li, Xin Weng, Julia A Schnabel, and Xiahai Zhuang, “Joint left atrial segmentation and scar quantification based on a dnn with spatial encoding and shape at- tention,” inInternational Conference on Medical Im- age Computing and Computer-Assisted Intervention. Springer, 202...

    2020

  39. [47]

    Y-net: a one-to-two deep learning framework for digital holographic reconstruction,

    Kaiqiang Wang, Jiazhen Dou, Qian Kemao, Jianglei Di, and Jianlin Zhao, “Y-net: a one-to-two deep learning framework for digital holographic reconstruction,”Op- tics Letters, vol. 44, no. 19, pp. 4765–4768, 2019

    2019

  40. [48]

    Nddr-cnn: Layerwise feature fusing in multi- task cnns by neural discriminative dimensionality re- duction,

    Yuan Gao, Jiayi Ma, Mingbo Zhao, Wei Liu, and Alan L Yuille, “Nddr-cnn: Layerwise feature fusing in multi- task cnns by neural discriminative dimensionality re- duction,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 3205–3214

    2019

  41. [49]

    Group normalization,

    Yuxin Wu and Kaiming He, “Group normalization,” in Proceedings of the European conference on computer vision (ECCV), 2018, pp. 3–19

    2018

  42. [50]

    Building skeleton models via 3-d medial sur- face axis thinning algorithms,

    Ta-Chih Lee, Rangasami L Kashyap, and Chong-Nam Chu, “Building skeleton models via 3-d medial sur- face axis thinning algorithms,”CVGIP: Graphical Mod- els and Image Processing, vol. 56, no. 6, pp. 462–478, 1994

    1994

  43. [51]

    Adam: A method for stochastic optimization,

    Diederik P Kingma and Jimmy Ba, “Adam: A method for stochastic optimization,”arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  44. [52]

    A threshold selection method from gray-level histograms,

    Nobuyuki Otsu et al., “A threshold selection method from gray-level histograms,”Automatica, vol. 11, no. 285-296, pp. 23–27, 1975

    1975

  45. [53]

    A large annotated 13 medical image dataset for the development and eval- uation of segmentation algorithms,

    Amber L Simpson, Michela Antonelli, Spyridon Bakas, Michel Bilello, Keyvan Farahani, Bram Van Ginneken, Annette Kopp-Schneider, Bennett A Landman, Geert Litjens, Bjoern Menze, et al., “A large annotated 13 medical image dataset for the development and eval- uation of segmentat...

    1902 arXiv

  46. [54]

    Liver segment approximation in ct data for surgical resection planning,

    Reinhard Beichel, Thomas Pock, Christian Janko, Ro- man B Zotter, Bernhard Reitinger, Alexander Bornik, Kalman Palagyi, Erich Sorantin, Georg Werkgartner, Horst Bischof, et al., “Liver segment approximation in ct data for surgical resection planning,” inMedi- cal Imaging 2004:...

    2004

  47. [55]

    Vascular branching ge- ometry relating to portal hypertension: a study of liver microvasculature in cirrhotic rats by x-ray phase- contrast computed tomography,

    Mengyu Sun, Wenjuan Lv, Xinyan Zhao, Lili Qin, Yuqing Zhao, Xiaohong Xin, Jianbo Jian, Xiaodong Chen, and Chunhong Hu, “Vascular branching ge- ometry relating to portal hypertension: a study of liver microvasculature in cirrhotic rats by x-ray phase- contrast computed tomograp...

    2020

  48. [56]

    Accurate and ver- satile 3d segmentation of plant tissues at cellular resolu- tion,

    Adrian Wolny, Lorenzo Cerrone, Athul Vijayan, Rachele Tofanelli, Amaya Vilches Barro, Marion Lou- veaux, Christian Wenzl, S ¨oren Strauss, David Wilson- S´anchez, Rena Lymbouridou, et al., “Accurate and ver- satile 3d segmentation of plant tissues at cellular resolu- tion,”Eli...

    2020

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.