Predicting clinically significant prostate cancer from quantitative image features including compressed sensing radial MRI of prostate perfusion using machine learning: comparison with PI-RADS v2 assessment scores.
Prostatic neoplasms
magnetic resonance imaging (MRI)
supervised machine learning
Journal
Quantitative imaging in medicine and surgery
ISSN: 2223-4292
Titre abrégé: Quant Imaging Med Surg
Pays: China
ID NLM: 101577942
Informations de publication
Date de publication:
Apr 2020
Apr 2020
Historique:
entrez:
2
5
2020
pubmed:
2
5
2020
medline:
2
5
2020
Statut:
ppublish
Résumé
To investigate if supervised machine learning (ML) classifiers would be able to predict clinically significant cancer (sPC) from a set of quantitative image-features and to compare these results with established PI-RADS v2 assessment scores. We retrospectively included 201, histopathologically-proven, peripheral zone (PZ) prostate cancer lesions. Gleason scores ≤3+3 were considered as clinically insignificant (inPC) and Gleason scores ≥3+4 as sPC and were encoded in a binary fashion, serving as ground-truth. MRI was performed at 3T with high spatiotemporal resolution DCE using Golden-angle RAdial SParse (GRASP) MRI. Perfusion maps (Ktrans, Kep, Ve), apparent diffusion coefficient (ADC), and absolute T2-signal intensities (SI) were determined in all lesions and served as input parameters for four supervised ML models: Gradient Boosting Machines (GBM), Neural Networks (NNet), Random Forest (RF) and Support Vector Machines (SVM). ML results and PI-RADS scores were compared with the ground-truth. Next ROC-curves and AUC values were calculated. All ML models outperformed PI-RADS v2 assessment scores in the prediction of sPC (RF, GBM, NNet and SVM Using quantitative imaging parameters as input, supervised ML models outperformed PI-RADS v2 assessment scores in the prediction of sPC. These results indicate that quantitative imagining parameters contain relevant information for the prediction of sPC from image features.
Sections du résumé
BACKGROUND
BACKGROUND
To investigate if supervised machine learning (ML) classifiers would be able to predict clinically significant cancer (sPC) from a set of quantitative image-features and to compare these results with established PI-RADS v2 assessment scores.
METHODS
METHODS
We retrospectively included 201, histopathologically-proven, peripheral zone (PZ) prostate cancer lesions. Gleason scores ≤3+3 were considered as clinically insignificant (inPC) and Gleason scores ≥3+4 as sPC and were encoded in a binary fashion, serving as ground-truth. MRI was performed at 3T with high spatiotemporal resolution DCE using Golden-angle RAdial SParse (GRASP) MRI. Perfusion maps (Ktrans, Kep, Ve), apparent diffusion coefficient (ADC), and absolute T2-signal intensities (SI) were determined in all lesions and served as input parameters for four supervised ML models: Gradient Boosting Machines (GBM), Neural Networks (NNet), Random Forest (RF) and Support Vector Machines (SVM). ML results and PI-RADS scores were compared with the ground-truth. Next ROC-curves and AUC values were calculated.
RESULTS
RESULTS
All ML models outperformed PI-RADS v2 assessment scores in the prediction of sPC (RF, GBM, NNet and SVM
CONCLUSIONS
CONCLUSIONS
Using quantitative imaging parameters as input, supervised ML models outperformed PI-RADS v2 assessment scores in the prediction of sPC. These results indicate that quantitative imagining parameters contain relevant information for the prediction of sPC from image features.
Identifiants
pubmed: 32355645
doi: 10.21037/qims.2020.03.08
pii: qims-10-04-808
pmc: PMC7188610
doi:
Types de publication
Journal Article
Langues
eng
Pagination
808-823Informations de copyright
2020 Quantitative Imaging in Medicine and Surgery. All rights reserved.
Déclaration de conflit d'intérêts
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at http://dx.doi.org/10.21037/qims.2020.03.08). The authors have no conflicts of interest to declare.
Références
Magn Reson Med. 2014 Sep;72(3):707-17
pubmed: 24142845
IEEE Trans Med Imaging. 2007 Jan;26(1):68-76
pubmed: 17243585
Comput Med Imaging Graph. 2017 Mar;56:1-10
pubmed: 28192761
Magn Reson Imaging. 2013 Jul;31(6):947-52
pubmed: 23602725
Radiology. 2013 Nov;269(2):493-503
pubmed: 23878284
J Clin Oncol. 2009 Nov 20;27(33):5627-33
pubmed: 19858401
Ann Biomed Eng. 2009 Apr;37(4):749-62
pubmed: 19169821
Eur Urol. 2016 Jan;69(1):16-40
pubmed: 26427566
J Digit Imaging. 2018 Jun;31(3):290-303
pubmed: 29181613
Radiology. 2017 Dec;285(3):859-869
pubmed: 28727501
Eur Radiol. 2019 Aug;29(8):4150-4159
pubmed: 30456585
Magn Reson Med. 2007 Dec;58(6):1182-95
pubmed: 17969013
J Magn Reson Imaging. 2015 May;41(5):1365-73
pubmed: 24833417
Radiographics. 2017 Mar-Apr;37(2):505-515
pubmed: 28212054
Radiology. 2019 Mar;290(3):702-708
pubmed: 30599102
J Magn Reson Imaging. 2009 Aug;30(2):327-34
pubmed: 19629981
Eur Radiol. 2012 Apr;22(4):746-57
pubmed: 22322308
J Urol. 2016 Sep;196(3):690-6
pubmed: 27101772
Magn Reson Med. 2009 Jan;61(1):103-16
pubmed: 19097216
Lancet Oncol. 2019 Jan;20(1):100-109
pubmed: 30470502
Br J Radiol. 2018 Feb;91(1083):20170645
pubmed: 29189042
Radiology. 2014 Apr;271(1):143-52
pubmed: 24475824
Eur Urol. 2019 Mar;75(3):385-396
pubmed: 29908876
Radiology. 2010 May;255(2):485-94
pubmed: 20413761
J Urol. 2017 Sep;198(3):583-590
pubmed: 28373133
Radiology. 2018 Oct;289(1):128-137
pubmed: 30063191
Invest Radiol. 2013 Jan;48(1):10-6
pubmed: 23192165
J Urol. 2018 Oct;200(4):767-773
pubmed: 29733838
Eur Radiol. 2019 Sep;29(9):4754-4764
pubmed: 31187216
World J Urol. 2016 Apr;34(4):525-32
pubmed: 26293117
Eur Radiol. 2012 Feb;22(2):468-75
pubmed: 21913058
Eur Radiol. 2017 Oct;27(10):4082-4090
pubmed: 28374077
IEEE Trans Med Imaging. 2014 Aug;33(8):1689-701
pubmed: 24802294
IEEE Trans Image Process. 2014 Mar;23(3):979-91
pubmed: 24464613
AJR Am J Roentgenol. 2013 Sep;201(3):W471-8
pubmed: 23971479
Eur J Radiol. 2019 Nov;120:108662
pubmed: 31539790
J Magn Reson. 2016 Aug;269:104-112
pubmed: 27288764
Sci Rep. 2019 Feb 7;9(1):1570
pubmed: 30733585
N Engl J Med. 2018 May 10;378(19):1767-1777
pubmed: 29552975
Abdom Radiol (NY). 2017 Jan;42(1):290-297
pubmed: 27576605
Biometrics. 1988 Sep;44(3):837-45
pubmed: 3203132
Abdom Radiol (NY). 2019 Jun;44(6):2233-2243
pubmed: 30955071
J Magn Reson Imaging. 2012 Oct;36(4):912-9
pubmed: 22711415
Eur Radiol. 2014 Oct;24(10):2597-605
pubmed: 25033819
J Magn Reson Imaging. 1999 Sep;10(3):223-32
pubmed: 10508281
Front Neuroinform. 2013 Dec 30;7:45
pubmed: 24416015
Acta Radiol. 2016 Jan;57(1):107-14
pubmed: 25505225