Deep learning-based segmentation of kidneys and renal cysts on T2-weighted MRI from patients with autosomal dominant polycystic kidney disease.
Artificial intelligence
Cysts
Image processing (computer-assisted)
Magnetic resonance imaging
Polycystic kidney diseases
Journal
European radiology experimental
ISSN: 2509-9280
Titre abrégé: Eur Radiol Exp
Pays: England
ID NLM: 101721752
Informations de publication
Date de publication:
30 Oct 2024
30 Oct 2024
Historique:
received:
24
05
2024
accepted:
19
09
2024
medline:
31
10
2024
pubmed:
31
10
2024
entrez:
31
10
2024
Statut:
epublish
Résumé
Our aim was to train and test a deep learning-based algorithm for automatically segmenting kidneys and renal cysts in patients with autosomal dominant polycystic kidney disease (ADPKD). We retrospectively selected all ADPKD patients who underwent renal MRI with coronal T2-weighted imaging at our institution from 2008 to 2022. The 20 most recent examinations constituted the test dataset, to mimic pseudoprospective enrolment. The remaining ones constituted the training dataset to which eight normal renal MRIs were added. Kidneys and cysts ground truth segmentations were performed on coronal T2-weighted images by a junior radiologist supervised by an experienced radiologist. Kidneys and cysts of the 20 test MRIs were segmented by the algorithm and three independent human raters. Segmentations were compared using overlap metrics. The total kidney volume (TKV), total cystic volume (TCV), and cystic index (TCV divided by TKV) were compared using Bland-Altman analysis. We included 164 ADPKD patients. Dice similarity coefficients ranged from 85.9% to 87.4% between the algorithms and the raters' segmentations and from 84.2% to 86.2% across raters' segmentations. For TCV assessment, the biases ± standard deviations (SD) were 3-19 ± 137-151 mL between the algorithm and the raters, and 22-45 ± 49-57 mL across raters. The algorithm underestimated TKV and TCV in two outliers with TCV > 2800 mL. For cystic index assessment, the biases ± SD were 2.5-6.9% ± 6.7-8.3% between the algorithm and the raters, and 2.1-9.4 ± 7.4-11.6% across raters. The algorithm's performance fell within the range of inter-rater variability, but large TKV and TCV were underestimated. Accurate automated segmentation of the renal cysts will enable the large-scale evaluation of the prognostic value of TCV and cystic index in ADPKD patients. If these biomarkers are prognostic, then automated segmentation will facilitate their use in daily routine. Cystic volume is an emerging biomarker in ADPKD. The algorithm's performance in segmenting kidneys and cysts fell within interrater variability. The segmentation of very large cysts, under-represented in the training dataset, needs improvement.
Sections du résumé
BACKGROUND
BACKGROUND
Our aim was to train and test a deep learning-based algorithm for automatically segmenting kidneys and renal cysts in patients with autosomal dominant polycystic kidney disease (ADPKD).
METHODS
METHODS
We retrospectively selected all ADPKD patients who underwent renal MRI with coronal T2-weighted imaging at our institution from 2008 to 2022. The 20 most recent examinations constituted the test dataset, to mimic pseudoprospective enrolment. The remaining ones constituted the training dataset to which eight normal renal MRIs were added. Kidneys and cysts ground truth segmentations were performed on coronal T2-weighted images by a junior radiologist supervised by an experienced radiologist. Kidneys and cysts of the 20 test MRIs were segmented by the algorithm and three independent human raters. Segmentations were compared using overlap metrics. The total kidney volume (TKV), total cystic volume (TCV), and cystic index (TCV divided by TKV) were compared using Bland-Altman analysis.
RESULTS
RESULTS
We included 164 ADPKD patients. Dice similarity coefficients ranged from 85.9% to 87.4% between the algorithms and the raters' segmentations and from 84.2% to 86.2% across raters' segmentations. For TCV assessment, the biases ± standard deviations (SD) were 3-19 ± 137-151 mL between the algorithm and the raters, and 22-45 ± 49-57 mL across raters. The algorithm underestimated TKV and TCV in two outliers with TCV > 2800 mL. For cystic index assessment, the biases ± SD were 2.5-6.9% ± 6.7-8.3% between the algorithm and the raters, and 2.1-9.4 ± 7.4-11.6% across raters.
CONCLUSION
CONCLUSIONS
The algorithm's performance fell within the range of inter-rater variability, but large TKV and TCV were underestimated.
RELEVANCE STATEMENT
CONCLUSIONS
Accurate automated segmentation of the renal cysts will enable the large-scale evaluation of the prognostic value of TCV and cystic index in ADPKD patients. If these biomarkers are prognostic, then automated segmentation will facilitate their use in daily routine.
KEY POINTS
CONCLUSIONS
Cystic volume is an emerging biomarker in ADPKD. The algorithm's performance in segmenting kidneys and cysts fell within interrater variability. The segmentation of very large cysts, under-represented in the training dataset, needs improvement.
Identifiants
pubmed: 39477840
doi: 10.1186/s41747-024-00520-7
pii: 10.1186/s41747-024-00520-7
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
122Informations de copyright
© 2024. The Author(s).
Références
Lanktree MB, Haghighi A, Guiard E et al (2018) Prevalence estimates of polycystic kidney and liver disease by population sequencing. J Am Soc Nephrol 29:2593–2600. https://doi.org/10.1681/ASN.2018050493
doi: 10.1681/ASN.2018050493
pubmed: 30135240
pmcid: 6171271
Willey CJ, Blais JD, Hall AK, Krasa HB, Makin AJ, Czerwiec FS (2017) Prevalence of autosomal dominant polycystic kidney disease in the European Union. Nephrol Dial Transplant 32:1356–1363. https://doi.org/10.1093/ndt/gfw240
doi: 10.1093/ndt/gfw240
pubmed: 27325254
Parfrey PS, Bear JC, Morgan J et al (1990) The diagnosis and prognosis of autosomal dominant polycystic kidney disease. N Engl J Med 323:1085–1090. https://doi.org/10.1056/NEJM199010183231601
doi: 10.1056/NEJM199010183231601
pubmed: 2215575
Simon P (1995) Prognosis of autosomal dominant polycystic kidney disease. Nephron 71:247–248. https://doi.org/10.1159/000188730
doi: 10.1159/000188730
pubmed: 8569973
Grantham JJ, Chapman AB, Torres VE (2006) Volume progression in autosomal dominant polycystic kidney disease: the major factor determining clinical outcomes. Clin J Am Soc Nephrol 1:148–157. https://doi.org/10.2215/CJN.00330705
doi: 10.2215/CJN.00330705
pubmed: 17699202
Grantham JJ, Mulamalla S, Swenson-Fields KI (2011) Why kidneys fail in autosomal dominant polycystic kidney disease. Nat Rev Nephrol 7:556–566. https://doi.org/10.1038/nrneph.2011.109
doi: 10.1038/nrneph.2011.109
pubmed: 21862990
Chapman AB, Guay-Woodford LM, Grantham JJ et al (2003) Renal structure in early autosomal-dominant polycystic kidney disease (ADPKD): the consortium for radiologic imaging studies of polycystic kidney disease (CRISP) cohort. Kidney Int 64:1035–1045. https://doi.org/10.1046/j.1523-1755.2003.00185.x
doi: 10.1046/j.1523-1755.2003.00185.x
pubmed: 12911554
Alam A, Dahl NK, Lipschutz JH et al (2015) Total kidney volume in autosomal dominant polycystic kidney disease: a biomarker of disease progression and therapeutic efficacy. Am J Kidney Dis 66:564–576. https://doi.org/10.1053/j.ajkd.2015.01.030
doi: 10.1053/j.ajkd.2015.01.030
pubmed: 25960302
Grantham JJ, Torres VE, Chapman AB et al (2006) Volume progression in polycystic kidney disease. N Engl J Med 354:2122–2130. https://doi.org/10.1056/NEJMoa054341
doi: 10.1056/NEJMoa054341
pubmed: 16707749
Irazabal MV, Rangel LJ, Bergstralh EJ et al (2015) Imaging classification of autosomal dominant polycystic kidney disease: a simple model for selecting patients for clinical trials. J Am Soc Nephrol 26:160–172. https://doi.org/10.1681/ASN.2013101138
doi: 10.1681/ASN.2013101138
pubmed: 24904092
Caroli A, Kline TL (2023) Abdominal imaging in ADPKD: beyond total kidney volume. J Clin Med 12:5133–5158. https://doi.org/10.3390/jcm1215533
doi: 10.3390/jcm1215533
pubmed: 37568535
pmcid: 10420262
Torres VE, Higashihara E, Devuyst O et al (2016) Effect of tolvaptan in autosomal dominant polycystic kidney disease by CKD stage: results from the tempo 3:4 trial. Clin J Am Soc Nephrol 11:803–811. https://doi.org/10.2215/CJN.06300615
doi: 10.2215/CJN.06300615
pubmed: 26912543
pmcid: 4858477
Gul S, Khan MS, Bibi A, Khandakar A, Ayari MA, Chowdhury MEH (2022) Deep learning techniques for liver and liver tumor segmentation: a review. Comput Biol Med 147:105620. https://doi.org/10.1016/j.compbiomed.2022.105620
doi: 10.1016/j.compbiomed.2022.105620
pubmed: 35667155
Wesdorp NJ, Zeeuw JM, Postma SCJ et al (2023) Deep learning models for automatic tumor segmentation and total tumor volume assessment in patients with colorectal liver metastases. Eur Radiol Exp 7:75–87. https://doi.org/10.1186/s41747-023-00383-4
doi: 10.1186/s41747-023-00383-4
pubmed: 38038829
pmcid: 10692044
Cayot B, Milot L, Nempont O et al (2022) Polycystic liver: automatic segmentation using deep learning on CT is faster and as accurate compared to manual segmentation. Eur Radiol 32:4780–4790. https://doi.org/10.1007/s00330-022-08549-1
doi: 10.1007/s00330-022-08549-1
pubmed: 35142898
Zöllner FG, Kocinski M, Hansen L et al (2021) Kidney segmentation in renal magnetic resonance imaging—current status and prospects. IEEE Access 9:71577–71605. https://doi.org/10.1109/ACCESS.2021.3078430
doi: 10.1109/ACCESS.2021.3078430
Kline TL, Korfiatis P, Edwards ME et al (2017) Performance of an artificial multi-observer deep neural network for fully automated segmentation of polycystic kidneys. J Digit Imaging 30:442–448. https://doi.org/10.1007/s10278-017-9978-1
doi: 10.1007/s10278-017-9978-1
pubmed: 28550374
pmcid: 5537093
Raj A, Tollens F, Hansen L et al (2022) Deep learning-based total kidney volume segmentation in autosomal dominant polycystic kidney disease using attention, cosine loss, and sharpness aware minimization. Diagnostics 12:1159–1175. https://doi.org/10.13390/diagnostics12051159
doi: 10.13390/diagnostics12051159
pubmed: 35626314
pmcid: 9139731
Cui H, Ma Y, Yang M et al (2023) Automatic segmentation of kidney volume using multi-module hybrid based u-shape in polycystic kidney disease. IEEE Access 11:58113–58124. https://doi.org/10.1109/ACCESS.2023.3284029
doi: 10.1109/ACCESS.2023.3284029
Goel A, Shih G, Riyahi S et al (2022) Deployed deep learning kidney segmentation for polycystic kidney disease MRI. Radiol Artif Intell 4:e210205–e210511. https://doi.org/10.1148/ryai.210205
doi: 10.1148/ryai.210205
pubmed: 35391774
Woznicki P, Siedek F, van Gastel MDA et al (2022) Automated kidney and liver segmentation in MR images in patients with autosomal dominant polycystic kidney disease: a multicenter study. Kidney360 3:2048–2058. https://doi.org/10.34067/KID.0003192022
doi: 10.34067/KID.0003192022
pubmed: 36591351
pmcid: 9802567
van Gastel MDA, Edwards ME, Torres VE, Erickson BJ, Gansevoort RT, Kline TL (2019) Automatic measurement of kidney and liver volumes from mr images of patients affected by autosomal dominant polycystic kidney disease. J Am Soc Nephrol 30:1514–1522. https://doi.org/10.1681/ASN.2018090902
doi: 10.1681/ASN.2018090902
pubmed: 31270136
pmcid: 6683702
Harris PC, Bae KT, Rossetti S et al (2006) Cyst number but not the rate of cystic growth is associated with the mutated gene in autosomal dominant polycystic kidney disease. J Am Soc Nephrol 17:3013–3019. https://doi.org/10.1681/ASN.2006080835
doi: 10.1681/ASN.2006080835
pubmed: 17035604
King BF, Reed JE, Bergstralh EJ, Sheedy 2nd PF, Torres VE (2000) Quantification and longitudinal trends of kidney, renal cyst, and renal parenchyma volumes in autosomal dominant polycystic kidney disease. J Am Soc Nephrol 11:1505–1511. https://doi.org/10.1681/ASN.V1181505
doi: 10.1681/ASN.V1181505
pubmed: 10906164
Bae KT, Zhou W, Shen C et al (2019) Growth pattern of kidney cyst number and volume in autosomal dominant polycystic kidney disease. Clin J Am Soc Nephrol 14:823–833. https://doi.org/10.2215/CJN.10360818
doi: 10.2215/CJN.10360818
pubmed: 31088850
pmcid: 6556721
Bae KT, Tao C, Wang J et al (2013) Novel approach to estimate kidney and cyst volumes using mid-slice magnetic resonance images in polycystic kidney disease. Am J Nephrol 38:333–341. https://doi.org/10.1159/000355375
doi: 10.1159/000355375
pubmed: 24107679
Bae KT, Sun H, Lee JG et al (2014) Novel methodology to evaluate renal cysts in polycystic kidney disease. Am J Nephrol 39:210–217. https://doi.org/10.1159/000358604
doi: 10.1159/000358604
pubmed: 24576800
Kline TL, Edwards ME, Fetzer J et al (2021) Automatic semantic segmentation of kidney cysts in MR images of patients affected by autosomal-dominant polycystic kidney disease. Abdom Radiol (NY) 46:1053–1061. https://doi.org/10.1007/s00261-020-02748-4
doi: 10.1007/s00261-020-02748-4
pubmed: 32940759
Gregory AV, Anaam DA, Vercnocke AJ et al (2021) Semantic instance segmentation of kidney cysts in MR images: a fully automated 3D approach developed through active learning. J Digit Imaging 34:773–787. https://doi.org/10.1007/s10278-021-00452-3
doi: 10.1007/s10278-021-00452-3
pubmed: 33821360
pmcid: 8455788
Schmidt EK, Krishnan C, Onuoha E et al (2024) Deep learning-based automated kidney and cyst segmentation of autosomal dominant polycystic kidney disease using single versus multi-institutional data. Clin Imaging 106:110068–110074. https://doi.org/10.1016/j.clinimag.2023.110068
doi: 10.1016/j.clinimag.2023.110068
pubmed: 38101228
Yeghiazaryan V, Voiculescu I (2018) Family of boundary overlap metrics for the evaluation of medical image segmentation. J Med Imaging 5:015006–015024. https://doi.org/10.1117/1.JMI.5.1.015006
doi: 10.1117/1.JMI.5.1.015006
Bland JM, Altman DG (1996) Measurement error. BMJ 313:744. https://doi.org/10.1136/bmj.313.7059.744
doi: 10.1136/bmj.313.7059.744
pubmed: 8819450
pmcid: 2352101
Benchoufi M, Matzner-Lober E, Molinari N, Jannot AS, Soyer P (2020) Interobserver agreement issues in radiology. Diagn Interv Imaging 101:639–641. https://doi.org/10.1016/j.diii.2020.09.001
doi: 10.1016/j.diii.2020.09.001
pubmed: 32958434
Rouviere O, Moldovan PC, Vlachomitrou A et al (2022) Combined model-based and deep learning-based automated 3D zonal segmentation of the prostate on T2-weighted MR images: clinical evaluation. Eur Radiol 32:3248–3259. https://doi.org/10.1007/s00330-021-08408-5
doi: 10.1007/s00330-021-08408-5
pubmed: 35001157
Moliere S, Hamzaoui D, Granger B et al (2024) Reference standard for the evaluation of automatic segmentation algorithms: quantification of inter observer variability of manual delineation of prostate contour on MRI. Diagn Interv Imaging 105:65–73. https://doi.org/10.1016/j.diii.2023.08.001
doi: 10.1016/j.diii.2023.08.001
pubmed: 37822196
Rouviere O (2024) Evaluation of automated prostate segmentation: the complex issue of the optimal number of expert segmentations. Diagn Interv Imaging 105:45–46. https://doi.org/10.1016/j.diii.2023.10.002
doi: 10.1016/j.diii.2023.10.002
pubmed: 37863708
Warfield SK, Zou KH, Wells WM (2004) Simultaneous truth and performance level estimation (staple): an algorithm for the validation of image segmentation. IEEE Trans Med Imaging 23:903–921. https://doi.org/10.1109/TMI.2004.828354
doi: 10.1109/TMI.2004.828354
pubmed: 15250643
pmcid: 1283110