pith. sign in

arxiv: 2607.01963 · v1 · pith:Z2M55EHRnew · submitted 2026-07-02 · ⚛️ physics.med-ph

Accelerating MRI Colon Volume Measurements and Reducing Inter-Observer Variation through Automatic Segmentation and Human-in-the-Loop Correction

Abi Spicer , Stephen Lloyd-Brown , Neele Dellschaft , Luca Marciani , Robin Spiller , Maura Corsetti , Penny Gowland , Xin Chen
show 1 more author
Caroline Hoad
This is my paper

Pith reviewed 2026-07-03 01:57 UTC · model grok-4.3

classification ⚛️ physics.med-ph
keywords MRIcolon segmentationautomatic segmentationcolonic volumennU-Nethuman-in-the-loop correctioninter-observer repeatability
0
0 comments X

The pith

Automatic ML segmentation with human correction reduces MRI colon analysis time from 56 to 11 minutes while preserving measurement accuracy.

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

The paper shows that an automatic segmentation model can be combined with quick human corrections to measure colon volumes from MRI much faster than full manual work. This addresses the problem that manual methods take too long for routine clinical use. The approach cuts analysis time by more than 80 percent and keeps high agreement with manual results for total colon volume. It also makes measurements more consistent between different people doing the analysis. If this holds, colon content measurements could become part of regular patient assessments and research on gut function.

Core claim

The central claim is that nnU-Net based automatic segmentation of the colon on mDIXON MRI, followed by manual correction of the generated masks, reduces the time for analysis from an average of 56 minutes to 11 minutes. The corrected masks agree closely with fully manual segmentations for whole colon volume with ICC of 0.96 and for regional volumes with ICC 0.80-0.95, while also improving inter-observer repeatability over manual methods.

What carries the argument

The nnU-Net model for automatic segmentation of colonic regions including ascending, transverse, descending and sigmoid-rectal, with subsequent human correction of the masks.

If this is right

  • Whole colonic volumes from the ML method are suitable for use with only minimal checks.
  • Regional colonic volumes show good to excellent agreement after correction.
  • The corrected method improves repeatability between observers compared to full manual segmentation.
  • The time reduction makes detailed colon content analysis feasible in clinical settings.

Where Pith is reading between the lines

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

  • This human-in-the-loop method could be tested on scans from different MRI scanners to check if the time savings hold.
  • Similar automatic approaches might speed up volume measurements in other parts of the body or with different imaging types.
  • Larger studies could use this to track changes in colon function over time with more patients.

Load-bearing premise

The automatic model, trained on the authors' specific dataset and scanner, will produce masks that need only short corrections on new scans from the same setup.

What would settle it

Finding that new scans require more than 20 minutes of correction on average or show ICC below 0.85 for whole volume would show the time savings and accuracy claims do not hold.

read the original abstract

The movement distribution, and volume of both chyme and gas in the colon, are important metrics to understand colonic function in health, disease, and the effects of treatments and different foodstuffs. Current methods available for assessment of these colonic contents using MRI consist mainly of manual segmentation or semi-automatic segmentation. However, these methods of segmentation are very labour intensive and too slow for clinical applications, require expert knowledge and some semi-automatic methods require use of bowel preparation. MRI scans were acquired in 2 breath holds using mDIXON sequences. We used the 'No New U-Net' (nnU-Net) ML model to automatically segment the colon, including colonic regions (ascending, transverse, descending and sigmoid-rectal). The ML-generated masks were corrected manually and the time taken for correction was recorded. ML segmentations were compared to both manual segmentations and observer corrected ML (CorrML) segmentations. Observer repeatability was also evaluated for both manual and CorrML methods to create a benchmark for the allowable error in the automatic segmentations. Analysis time was significantly reduced (p<0.0001) from 56 mins (+-11 mins (SD)) for manual masks to 11 mins (+-5 mins (SD)) for CorrML masks. Both DICE and ICC values showed excellent agreement between manual, ML and CorrML segmentations for whole colonic volume (ICC = 0.96) whilst regional volumes were good-excellent (ICC = 0.80-0.95). Inter-observer repeatability was improved when using CorrML methods over manual segmentation (ICC manual > 0.89, CorrML > 0.93). Analysis time was reduced by over 80% when using CorrML methods and whole colonic volumes measured by ML would be suitable for use with minimal checks. Hence the methods proposed here would be clinically useful.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 1 minor

Summary. The manuscript applies the nnU-Net model to automatically segment the whole colon and its regions (ascending, transverse, descending, sigmoid-rectal) from 2-breath-hold mDIXON MRI scans. It compares fully manual segmentation, raw ML output, and human-corrected ML masks (CorrML), reporting a statistically significant reduction in analysis time from 56±11 min to 11±5 min (p<0.0001), excellent whole-colon agreement (ICC=0.96) and good-to-excellent regional agreement (ICC 0.80-0.95) between methods, plus improved inter-observer repeatability with CorrML (ICC>0.93) versus manual (ICC>0.89). The authors conclude that CorrML reduces effort by >80% and is clinically useful.

Significance. If the reported time savings and agreement metrics generalize, the approach would meaningfully lower the barrier to routine colonic volume and content quantification in clinical research and practice. The work directly addresses a known bottleneck (labor-intensive manual or semi-automatic segmentation) with a concrete, measurable improvement and supplies repeatability benchmarks that could serve as reference values for future studies.

major comments (2)
  1. [Abstract / Methods] Abstract and Methods: The central claim that CorrML reduces analysis time by >80% while preserving clinical utility rests on the untested assumption that nnU-Net masks require only brief corrections on future scans acquired under the same mDIXON protocol. No training-set size, cross-validation procedure, or external test set (different scanner, field strength, or minor protocol variation) is reported, so the generalizability of the time-reduction and ICC results cannot be assessed from the given data.
  2. [Results] Results: The reported inter-observer ICC improvement (manual >0.89 to CorrML >0.93) and the claim that ML volumes are 'suitable for use with minimal checks' are derived from the same internal cohort used to train and evaluate the model; without an independent test cohort these figures may overestimate performance on new subjects.
minor comments (1)
  1. [Abstract] Abstract: The phrase 'MRI scans were acquired in 2 breath holds using mDIXON sequences' should specify the number of subjects or scans used for training versus testing to allow immediate evaluation of sample size.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments highlighting the need for clearer reporting on model training and evaluation cohorts. We address each major comment below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract / Methods] Abstract and Methods: The central claim that CorrML reduces analysis time by >80% while preserving clinical utility rests on the untested assumption that nnU-Net masks require only brief corrections on future scans acquired under the same mDIXON protocol. No training-set size, cross-validation procedure, or external test set (different scanner, field strength, or minor protocol variation) is reported, so the generalizability of the time-reduction and ICC results cannot be assessed from the given data.

    Authors: We agree that the manuscript should explicitly report the training details to allow assessment of generalizability. In the revised Methods section we will add: (i) the size of the training dataset (number of subjects and scans), (ii) the cross-validation procedure used by nnU-Net (default 5-fold), and (iii) confirmation that the time and ICC metrics were measured on a held-out test subset. We will also insert a limitations paragraph in the Discussion noting the absence of external validation on different scanners or protocols and that the reported time savings apply to the tested mDIXON protocol. revision: yes

  2. Referee: [Results] Results: The reported inter-observer ICC improvement (manual >0.89 to CorrML >0.93) and the claim that ML volumes are 'suitable for use with minimal checks' are derived from the same internal cohort used to train and evaluate the model; without an independent test cohort these figures may overestimate performance on new subjects.

    Authors: The inter-observer repeatability experiments were performed on scans held out from model training to prevent leakage, yet they remain within the same single-site cohort. We will revise the Results and Discussion to (a) clarify the separation between training and repeatability sets, (b) replace the phrase 'suitable for use with minimal checks' with a more qualified statement, and (c) add an explicit caveat that the ICC improvements and time savings are internal benchmarks and may not fully generalize to new subjects or sites. This constitutes a partial revision because the underlying data remain internal. revision: partial

Circularity Check

0 steps flagged

No circularity; direct empirical validation of segmentation pipelines

full rationale

The paper conducts an empirical study comparing manual segmentation, nnU-Net ML masks, and human-corrected ML (CorrML) masks on the authors' mDIXON MRI dataset. It reports measured quantities (analysis time, Dice scores, ICC agreement, inter-observer repeatability) obtained by applying standard metrics to the outputs of each pipeline. No derivation chain, equations, or first-principles predictions exist that could reduce to fitted parameters or self-referential definitions. No self-citations are invoked to establish uniqueness or to smuggle in ansatzes. The central claims rest on observed performance differences within the study cohort rather than any tautological reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on standard assumptions of MRI contrast and nnU-Net generalization; no new physical entities or ad-hoc constants are introduced.

axioms (1)
  • domain assumption mDIXON MRI sequences provide adequate contrast for reliable colon boundary detection across subjects
    Invoked by the choice of acquisition protocol and the decision to apply nnU-Net without additional preprocessing steps.

pith-pipeline@v0.9.1-grok · 5907 in / 1329 out tokens · 33264 ms · 2026-07-03T01:57:30.104792+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

43 extracted references · 38 canonical work pages · 1 internal anchor

  1. [1]

    Whole Colon

    and varying levels of gas, adding a level of complexity to the dataset. MRI scans were performed at baseline (BL), T = 0 (immediately after consuming the meal), and then hourly over a 6-hour period and finally a single scan at T = 24 hours on a 3T Philips wide bore Ingenia scanner (Best, The Netherlands). At each time point a dual echo mDIXON (field of vi...

    2023

  2. [2]

    Abdominal distension after eating lettuce: The role of intestinal gas evaluated in vitro and by abdominal CT imaging

    Barba E, Sánchez B, Burri E, Accarino A, Monclus E, Navazo I, et al. Abdominal distension after eating lettuce: The role of intestinal gas evaluated in vitro and by abdominal CT imaging. Neurogastroenterology and motility. 2019 Dec 1;31(12). doi:10.1111/NMO.13703 PubMed PMID: 31402544

    work page doi:10.1111/nmo.13703 2019

  3. [3]

    Mechanisms of abdominal distension in severe intestinal dysmotility: abdomino-thoracic response to gut retention

    Barba E, Quiroga S, Accarino A, Lahoya EM, Malagelada C, Burri E, et al. Mechanisms of abdominal distension in severe intestinal dysmotility: abdomino-thoracic response to gut retention. Neurogastroenterology and motility. 2013 Jun;25(6). doi:10.1111/NMO.12128 PubMed PMID: 23607758

    work page doi:10.1111/nmo.12128 2013

  4. [4]

    Intestinal gas and bloating: effect of prokinetic stimulation

    Accarino A, Perez F, Azpiroz F, Quiroga S, Malagelada JR. Intestinal gas and bloating: effect of prokinetic stimulation. Am J Gastroenterol. 2008 Aug;103(8):2036–42. doi:10.1111/J.1572-0241.2008.01866.X PubMed PMID: 18802999

    work page doi:10.1111/j.1572-0241.2008.01866.x 2008

  5. [6]

    Colon Hypersensitivity to Distension, Rather Than Excessive Gas Production, Produces Carbohydrate-Related Symptoms in Individuals With Irritable Bowel Syndrome

    Major G, Pritchard S, Murray K, Alappadan JP , Hoad CL, Marciani L, et al. Colon Hypersensitivity to Distension, Rather Than Excessive Gas Production, Produces Carbohydrate-Related Symptoms in Individuals With Irritable Bowel Syndrome. Gastroenterology. 2017 Jan 1;152(1):124-133.e2. doi:10.1053/j.gastro.2016.09.062 PubMed PMID: 27746233

    work page doi:10.1053/j.gastro.2016.09.062 2017

  6. [7]

    Wilkinson-Smith V, Scott M, Menys A, Wiklendt L, Marciani L, Atkinson D, et al. Combined magnetic resonance imaging, high resolution manometry and a randomised trial of bisacodyl versus hyoscine shows the significance of an enlarged colon in constipation: the RECLAIM study. Gut (2024) (In press) [Internet]. 2024 Sep 23 [cited 2026 Feb 9]. Available from: ...

    2024

  7. [8]

    Efficacy and Mechanism Evaluation Ondansetron for irritable bowel syndrome with diarrhoea: randomised controlled trial

    Gunn D, Topan R, Fried R, Holloway I, Brindle R, Hartley S, et al. Efficacy and Mechanism Evaluation Ondansetron for irritable bowel syndrome with diarrhoea: randomised controlled trial. 2023. doi:10.3310/YTFW7874

    work page doi:10.3310/ytfw7874 2023

  8. [9]

    Thinking outside the box: a review of gastrointestinal symptoms and complications in cystic fibrosis

    Yule A, Sills D, Smith S, Spiller R, Smyth AR. Thinking outside the box: a review of gastrointestinal symptoms and complications in cystic fibrosis. Expert Rev Respir Med. 2023 Jul 3;17(7):547–61. doi:10.1080/17476348.2023.2228194 PubMed PMID: 37345513

    work page doi:10.1080/17476348.2023.2228194 2023

  9. [10]

    Postprandial changes in gastrointestinal function and transit in cystic fibrosis assessed by Magnetic Resonance Imaging

    Ng C, Dellschaft NS, Hoad CL, Marciani L, Ban L, Prayle AP , et al. Postprandial changes in gastrointestinal function and transit in cystic fibrosis assessed by Magnetic Resonance Imaging. Journal of Cystic Fibrosis. 2021 Jul 1;20(4):591–7. doi:10.1016/J.JCF.2020.06.004 PubMed PMID: 32561324

    work page doi:10.1016/j.jcf.2020.06.004 2021

  10. [11]

    Mechanisms underlying effects of kiwifruit on intestinal function shown by MRI in healthy volunteers

    Wilkinson-Smith V, Dellschaft N, Ansell J, Hoad C, Marciani L, Gowland P , et al. Mechanisms underlying effects of kiwifruit on intestinal function shown by MRI in healthy volunteers. Aliment Pharmacol Ther. 2019 Mar 1;49(6):759–68. doi:10.1111/apt.15127 PubMed PMID: 30706488

    work page doi:10.1111/apt.15127 2019

  11. [12]

    Contrasting effects of viscous and particulate fibers on colonic fermentation in vitro and in vivo, and their impact on intestinal water studied by MRI in a randomized trial

    Gunn D, Murthy R, Major G, Wilkinson-Smith V, Hoad C, Marciani L, et al. Contrasting effects of viscous and particulate fibers on colonic fermentation in vitro and in vivo, and their impact on intestinal water studied by MRI in a randomized trial. Am J Clin Nutr. 2020 Sep 1;112(3):595–602. doi:10.1093/AJCN/NQAA173 PubMed PMID: 32619212

    work page doi:10.1093/ajcn/nqaa173 2020

  12. [13]

    Colonic content: effect of diet, meals, and defecation

    Bendezú RA, Mego M, Monclus E, Merino X, Accarino A, Malagelada JR, et al. Colonic content: effect of diet, meals, and defecation. Neurogastroenterology and Motility. 2017 Feb 1;29(2):e12930. doi:10.1111/NMO.12930;WEBSITE:WEBSITE:PERICLES;WGROUP:STRING:PUBLICATION PubMed PMID: 27545449

    work page doi:10.1111/nmo.12930;website:website:pericles;wgroup:string:publication 2017

  13. [14]

    Magnetic Resonance Imaging Reveals Novel Insights into the Dual Mode of Action of Bisacodyl: A Randomized, Placebo-controlled Trial in Constipation

    Aliyu A, Dellschaft N, Hoad C, Williams H, Gaudoin E, Sulaiman S, et al. Magnetic Resonance Imaging Reveals Novel Insights into the Dual Mode of Action of Bisacodyl: A Randomized, Placebo-controlled Trial in Constipation. Clin Pharmacol Ther. 2025 May 1;117(5):1284–91. doi:10.1002/CPT.3532;PAGEGROUP:STRING:PUBLICATION PubMed PMID: 39679695

    work page doi:10.1002/cpt.3532;pagegroup:string:publication 2025

  14. [15]

    Aalborg Universitet Effects of opium tincture on gastrointestinal function and motility in healthy volunteers: A magnetic resonance imaging study [Internet]

    Bolvig E, Gerdt D, Elise Møller L, Brøndum J, Mohr A. Aalborg Universitet Effects of opium tincture on gastrointestinal function and motility in healthy volunteers: A magnetic resonance imaging study [Internet]. 2024. doi:10.1111/nmo.14941

    work page doi:10.1111/nmo.14941 2024

  15. [16]

    Fasting and postprandial volumes of the undisturbed colon: normal values and changes in diarrhea-predominant irritable bowel syndrome measured using serial MRI

    Pritchard SE, Marciani L, Garsed KC, Hoad CL, Thongborisute W, Roberts E, et al. Fasting and postprandial volumes of the undisturbed colon: normal values and changes in diarrhea-predominant irritable bowel syndrome measured using serial MRI. Neurogastroenterology & Motility. 2014 Jan 1;26(1):124–30. doi:10.1111/NMO.12243 PubMed PMID: 24131490

    work page doi:10.1111/nmo.12243 2014

  16. [17]

    A novel semi-automatic segmentation method for volumetric assessment of the colon based on magnetic resonance imaging

    Sandberg TH, Nilsson M, Poulsen JL, Gram M, Frøkjær JB, Østergaard LR, et al. A novel semi-automatic segmentation method for volumetric assessment of the colon based on magnetic resonance imaging. Abdominal Imaging 2015 40:7. 2015 Jun 9;40(7):2232–41. doi:10.1007/s00261-015-0475-z PubMed PMID: 26054979

    work page doi:10.1007/s00261-015-0475-z 2015

  17. [18]

    Learning active contour models for medical image segmentation

    Chen X, Williams BM, Vallabhaneni SR, Czanner G, Williams R, Zheng Y . Learning active contour models for medical image segmentation. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. 2019 Jun 1;2019- June:11624–32. doi:10.1109/CVPR.2019.01190

    work page doi:10.1109/cvpr.2019.01190 2019

  18. [20]

    Automatic segmentation of liver structures in multi-phase MRI using variants of nnU-Net and Swin UNETR

    Raab F, Strotzer Q, Stroszczynski C, Fellner C, Einspieler I, Haimerl M, et al. Automatic segmentation of liver structures in multi-phase MRI using variants of nnU-Net and Swin UNETR. Scientific Reports 2025 15:1. 2025 Jul 16;15(1):25740-. doi:10.1038/s41598-025- 07084-5 PubMed PMID: 40670420

    work page doi:10.1038/s41598-025- 2025

  19. [21]

    nnUNet for Automatic Kidney and Cyst Segmentation in Autosomal Dominant Polycystic Kidney Disease

    Krishnan C, Schmidt E, Onuoha E, Mrug M, Cardenas CE, Kim H. nnUNet for Automatic Kidney and Cyst Segmentation in Autosomal Dominant Polycystic Kidney Disease. Curr Med Imaging. 2024 Feb 7;20(1):1–9. doi:10.2174/0115734056272767231130110017 PubMed PMID: 38389364

    work page doi:10.2174/0115734056272767231130110017 2024

  20. [22]

    Automated liver and spleen segmentation for MR elastography maps using U-Nets

    Jaitner N, Ludwig J, Meyer T, Boehm O, Anders M, Huang B, et al. Automated liver and spleen segmentation for MR elastography maps using U-Nets. Sci Rep. 2025 Dec 1;15(1):10762-. doi:10.1038/S41598-025-95157-W;SUBJMETA PubMed PMID: 40155744

    work page doi:10.1038/s41598-025-95157-w;subjmeta 2025

  21. [23]

    TotalSegmentator MRI: Robust Sequence-independent Segmentation of Multiple Anatomic Structures in MRI [Internet]

    Akinci D’antonoli T, Lucas •, Berger K, Indrakanti AK, Vishwanathan N, Weiss J, et al. TotalSegmentator MRI: Robust Sequence-independent Segmentation of Multiple Anatomic Structures in MRI [Internet]. 2025. doi:10.1148/radiol.241613

    work page doi:10.1148/radiol.241613 2025

  22. [24]

    The effect of depth context in the segmentation of the colon in MRI volumes [Internet]

    Benson E, Rier L, Millican I, Pritchard S, Costigan C, Pound M, et al. The effect of depth context in the segmentation of the colon in MRI volumes [Internet]. doi:10.1101/2020.03.06.20027722

    work page doi:10.1101/2020.03.06.20027722 2020

  23. [25]

    Advanced gastrointestinal tract organ differentiation using an integrated swin transformer U-Net model for cancer care

    Sharma N, Gupta S, Almogren A, Bharany S, Altameem A, Rehman AU. Advanced gastrointestinal tract organ differentiation using an integrated swin transformer U-Net model for cancer care. Front Phys. 2024 Dec 12;12:1478750. doi:10.3389/fphy.2024.1478750

    work page doi:10.3389/fphy.2024.1478750 2024

  24. [26]

    Encoder–Decoder Variant Analysis for Semantic Segmentation of Gastrointestinal Tract Using UW-Madison Dataset

    Sharma N, Gupta S, Elkamchouchi DH, Bharany S. Encoder–Decoder Variant Analysis for Semantic Segmentation of Gastrointestinal Tract Using UW-Madison Dataset. Bioengineering 2025, Vol 12,. 2025 Mar 17;12(3). doi:10.3390/bioengineering12030309

    work page doi:10.3390/bioengineering12030309 2025

  25. [27]

    ResECA- Unet: Enhancement of GI Tract Segmentation Using an Improved U-Net Framework [Internet]

    Nuruzzaman Nobel S, Faruque Sifat O, Islam MR, Sayeed MS, Amiruzzaman M. ResECA- Unet: Enhancement of GI Tract Segmentation Using an Improved U-Net Framework [Internet]. 2024 Mar 29. doi:10.20944/preprints202403.1833.v1

    work page doi:10.20944/preprints202403.1833.v1 2024

  26. [29]

    A comprehensive dataset of magnetic resonance enterography images with intestinal segment annotations

    Zhong Z, Huang L, Feng ST , Lin H, Wang X, Lu B, et al. A comprehensive dataset of magnetic resonance enterography images with intestinal segment annotations. Scientific Data 2025 12:1. 2025 Mar 11;12(1):425-. doi:10.1038/s41597-025-04760-z PubMed PMID: 40069172

    work page doi:10.1038/s41597-025-04760-z 2025

  27. [30]

    Automatic colon segmentation on T1-FS MR images

    Orellana B, Navazo I, Brunet P , Monclús E, Bendezú Á, Azpiroz F. Automatic colon segmentation on T1-FS MR images. Computerized Medical Imaging and Graphics. 2025 Jul 1;123(5):102528. doi:10.1016/j.compmedimag.2025.102528 PubMed PMID: 40112651

    work page doi:10.1016/j.compmedimag.2025.102528 2025

  28. [31]

    nnU-Net–based Segmentation of Tumor Subcompartments in Pediatric Medulloblastoma Using Multiparametric MRI: A Multi-institutional Study

    Bareja R, Ismail M, Martin D, Nayate A, Yadav I, Labbad M, et al. nnU-Net–based Segmentation of Tumor Subcompartments in Pediatric Medulloblastoma Using Multiparametric MRI: A Multi-institutional Study. https://doi.org/101148/ryai230115. 2024 Aug 21;6(5). doi:10.1148/RYAI.230115

    work page doi:10.1148/ryai.230115 2024

  29. [32]

    A nnU-Net-based automatic segmentation of FCD type II lesions in 3D FLAIR MRI images

    Joshi S, Pant M, Malhotra A, Deep K, Snasel V. A nnU-Net-based automatic segmentation of FCD type II lesions in 3D FLAIR MRI images. Front Artif Intell. 2025 Jun 27;8:1601815. doi:10.3389/FRAI.2025.1601815/TEXT

    work page doi:10.3389/frai.2025.1601815/text 2025

  30. [33]

    Deep learning application for abdominal organs segmentation on 0.35 T MR-Linac images

    Zhou Y , Lalande A, Chevalier C, Baude J, Aubignac L, Boudet J, et al. Deep learning application for abdominal organs segmentation on 0.35 T MR-Linac images. Front Oncol. 2023 Jan 8;13:1285924. doi:10.3389/FONC.2023.1285924/TEXT

    work page doi:10.3389/fonc.2023.1285924/text 2023

  31. [34]

    U-Net: Convolutional Networks for Biomedical Image Segmentation

    Ronneberger O, Fischer P , Brox T. 2015-U-Net. ArXiv [Internet]. 2015 [cited 2026 Jun 3];1–8. Available from: http://lmb.informatik.uni-freiburg.de/%0Aarxiv:1505.04597v1

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  32. [35]

    On-the-Fly Data Augmentation for Brain Tumor Segmentation [Internet]

    Jain I, Willems S, Latre S, De Schepper T. On-the-Fly Data Augmentation for Brain Tumor Segmentation [Internet]. 2025 Sep 29 [cited 2026 Jun 3]. Available from: https://arxiv.org/pdf/2509.24973

    work page arXiv 2025

  33. [36]

    Towards Reliable Pediatric Brain Tumor Segmentation: Task-Specific nnU-Net Enhancements [Internet]

    Li X, Xu ZQJ, Ren Y , Qiu T, Wang X. Towards Reliable Pediatric Brain Tumor Segmentation: Task-Specific nnU-Net Enhancements [Internet]. 2025 Nov 1 [cited 2026 Jun 3]. Available from: https://arxiv.org/pdf/2511.00449

    work page arXiv 2025

  34. [37]

    Frequency-Aware Ensemble Learning for BraTS 2025 Pediatric Brain Tumor Segmentation [Internet]

    Yi Y , Zhuang Q, Xu ZQJ, Wang X, Ren Y , Qiu T. Frequency-Aware Ensemble Learning for BraTS 2025 Pediatric Brain Tumor Segmentation [Internet]. 2025 Sep 18 [cited 2026 Jun 3]. Available from: https://arxiv.org/pdf/2509.19353

    work page arXiv 2025

  35. [38]

    Ingesting methylcellulose fibre gels are as effective as psyllium in reducing colonic fermentation of inulin - a first step to IBS symptom relief

    Neele Dellschaft, Alaa Alhasani, Joshua Reid, Abi Spicer, Caroline Hoad, Luca Marciani, et al. Ingesting methylcellulose fibre gels are as effective as psyllium in reducing colonic fermentation of inulin - a first step to IBS symptom relief. In: International Society for Magnetic Resonance in Medicine 2025. Hawaii: ISMRM; 2025. p. 4927

    2025

  36. [39]

    Dual-echo Dixon imaging with flexible choice of echo times

    Eggers H, Brendel B, Duijndam A, Herigault G. Dual-echo Dixon imaging with flexible choice of echo times. Magn Reson Med. 2011;65(1):96–107. doi:10.1002/MRM.22578 PubMed PMID: 20860006

    work page doi:10.1002/mrm.22578 2011

  37. [40]

    Medical Image Processing, Analysis & Visualization In Clinical Research

    McAuliffe MJ, Lalonde FM, McGarry D, Gandler W, Csaky K, Trus BL. Medical Image Processing, Analysis & Visualization In Clinical Research. IEEE COMPUTER-BASED MEDICAL SYSTEMS (CBMS). 2001;381–6

    2001

  38. [41]

    User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliability

    Yushkevich PA, Piven J, Hazlett HC, Smith RG, Ho S, Gee JC, et al. User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliability. Neuroimage. 2006 Jul 1;31(3):1116–28. doi:10.1016/J.NEUROIMAGE.2006.01.015 PubMed PMID: 16545965

    work page doi:10.1016/j.neuroimage.2006.01.015 2006

  39. [42]

    Advanced gastrointestinal tract organ differentiation using an integrated swin transformer U-Net model for cancer care

    Sharma N, Gupta S, Almogren A, Bharany S, Altameem A, Rehman AU. Advanced gastrointestinal tract organ differentiation using an integrated swin transformer U-Net model for cancer care. Front Phys. 2024 Dec 12;12:1478750. doi:10.3389/FPHY .2024.1478750/TEXT

    work page doi:10.3389/fphy 2024

  40. [43]

    DeepGI: An Automated Approach for Gastrointestinal Tract Segmentation in MRI Scans

    Zhang Y , Gong Y , Cui D, Li X, Shen X. DeepGI: An Automated Approach for Gastrointestinal Tract Segmentation in MRI Scans

  41. [44]

    Fully Automated Colon Delineation and Volume Estimation in T2-Weighted MRI with a 2D U- Net

    Plocharski M, Hvaas GB, Møller MG, Thostrup NS, Samuelsen F, Mark EB. Fully Automated Colon Delineation and Volume Estimation in T2-Weighted MRI with a 2D U- Net. Stud Health Technol Inform. 2026 May 21;336:37–41. doi:10.3233/SHTI260104 PubMed PMID: 42174781

    work page doi:10.3233/shti260104 2026

  42. [45]

    Large-scale multi-center CT and MRI segmentation of pancreas with deep learning

    Zhang Z, Keles E, Durak G, Taktak Y , Susladkar O, Gorade V, et al. Large-scale multi-center CT and MRI segmentation of pancreas with deep learning. Med Image Anal. 2025 Jan 1;99:103382. doi:10.1016/J.MEDIA.2024.103382 PubMed PMID: 39541706

    work page doi:10.1016/j.media.2024.103382 2025

  43. [46]

    Magnetic Resonance Imaging Reveals Novel Insights into the Dual Mode of Action of Bisacodyl: A Randomized, Placebo-controlled Trial in Constipation

    Aliyu A, Dellschaft N, Hoad C, Williams H, Gaudoin E, Sulaiman S, et al. Magnetic Resonance Imaging Reveals Novel Insights into the Dual Mode of Action of Bisacodyl: A Randomized, Placebo-controlled Trial in Constipation. Clin Pharmacol Ther. 2024 May 1;117(5):1284. doi:10.1002/CPT.3532 PubMed PMID: 39679695

    work page doi:10.1002/cpt.3532 2024