Advertisement

Machine learning-based analysis of MR radiomics can help to improve the diagnostic performance of PI-RADS v2 in clinically relevant prostate cancer

Abstract

Objective

To investigate whether machine learning-based analysis of MR radiomics can help improve the performance PI-RADS v2 in clinically relevant prostate cancer (PCa).

Methods

This IRB-approved study included 54 patients with PCa undergoing multi-parametric (mp) MRI before prostatectomy. Imaging analysis was performed on 54 tumours, 47 normal peripheral (PZ) and 48 normal transitional (TZ) zone based on histological-radiological correlation. Mp-MRI was scored via PI-RADS, and quantified by measuring radiomic features. Predictive model was developed using a novel support vector machine trained with: (i) radiomics, (ii) PI-RADS scores, (iii) radiomics and PI-RADS scores. Paired comparison was made via ROC analysis.

Results

For PCa versus normal TZ, the model trained with radiomics had a significantly higher area under the ROC curve (Az) (0.955 [95% CI 0.923–0.976]) than PI-RADS (Az: 0.878 [0.834–0.914], p < 0.001). The Az between them was insignificant for PCa versus PZ (0.972 [0.945–0.988] vs. 0.940 [0.905–0.965], p = 0.097). When radiomics was added, performance of PI-RADS was significantly improved for PCa versus PZ (Az: 0.983 [0.960–0.995]) and PCa versus TZ (Az: 0.968 [0.940–0.985]).

Conclusion

Machine learning analysis of MR radiomics can help improve the performance of PI-RADS in clinically relevant PCa.

Key Points

Machine-based analysis of MR radiomics outperformed in TZ cancer against PI-RADS.

Adding MR radiomics significantly improved the performance of PI-RADS.

DKI-derived Dapp and Kapp were two strong markers for the diagnosis of PCa.

This is a preview of subscription content, log in to check access.

Access options

Buy single article

Instant unlimited access to the full article PDF.

US$ 39.95

Price includes VAT for USA

Subscribe to journal

Immediate online access to all issues from 2019. Subscription will auto renew annually.

US$ 199

This is the net price. Taxes to be calculated in checkout.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5

Abbreviations

Az:

Area under the ROC curve

DCE:

Dynamic contrast-enhanced

DKI:

Diffusional kurtosis imaging

DWI:

Diffusion-weighted imaging

Mp-MRI:

Multi-parametric magnetic resonance imaging

PCa:

Prostate cancer

PI-RADS:

Prostate Imaging Reporting and Data System

PZ:

Peripheral zone

RESOLVE:

Readout segmentation of long variable echo-trains

ROC:

Receiver operating characteristic curve

SVM:

Support vector machine

TZ:

Transitional zone

VOI:

Volume of interest

References

  1. 1.

    Siegel RL, Miller KD, Jemal A (2016) Cancer statistics, 2016. CA Cancer J Clin 66:7–30

  2. 2.

    Yakar D, Debats OA, Bomers JG et al (2012) Predictive value of MRI in the localization, staging, volume estimation, assessment of aggressiveness, and guidance of radiotherapy and biopsies in prostate cancer. J Magn Reson Imaging 35:20–31

  3. 3.

    Vargas HA, Akin O, Franiel T et al (2011) Diffusion-weighted endorectal MR imaging at 3 T for prostate cancer: tumor detection and assessment of aggressiveness. Radiology 259:775–784

  4. 4.

    Soylu FN, Peng Y, Jiang Y et al (2013) Seminal vesicle invasion in prostate cancer: evaluation by using multiparametric endorectal MR imaging. Radiology 267:797–806

  5. 5.

    Wang Q, Li H, Yan X et al (2015) Histogram analysis of diffusion kurtosis magnetic resonance imaging in differentiation of pathologic Gleason grade of prostate cancer. Urol Oncol 33(337):e315–e324

  6. 6.

    Hambrock T, Somford DM, Huisman HJ et al (2011) Relationship between apparent diffusion coefficients at 3.0-T MR imaging and Gleason grade in peripheral zone prostate cancer. Radiology 259:453–461

  7. 7.

    Park SY, Oh YT, Jung DC et al (2016) Prediction of biochemical recurrence after radical prostatectomy with PI-RADS version 2 in prostate cancers: initial results. Eur Radiol 26:2502–2509

  8. 8.

    Weinreb JC, Barentsz JO, Choyke PL et al (2016) PI-RADS prostate imaging - Reporting and data system: 2015, Version 2. Eur Urol 69:16–40

  9. 9.

    Peng Y, Jiang Y, Yang C et al (2013) Quantitative analysis of multiparametric prostate MR images: differentiation between prostate cancer and normal tissue and correlation with Gleason score--a computer-aided diagnosis development study. Radiology 267:787–796

  10. 10.

    Kickingereder P, Bonekamp D, Nowosielski M et al (2016) Radiogenomics of glioblastoma: Machine learning-based classification of molecular characteristics by using multiparametric and multiregional MR imaging features. Radiology 281:907–918

  11. 11.

    Zhang YD, Wang J, Wu CJ et al (2016) An imaging-based approach predicts clinical outcomes in prostate cancer through a novel support vector machine classification. Oncotarget. doi:10.18632/oncotarget.11293

  12. 12.

    Zhang YD, Shen CM, Meng HT et al (2016) Allele and haplotype diversity of new multiplex of 19 ChrX-STR loci in Han population from Guanzhong region (China). Electrophoresis 37:1669–1675

  13. 13.

    Zhang YD, Wu CJ, Bao ML et al (2016) New RESOLVE-based diffusional kurtosis imaging in MRI-visible prostate cancer: effect of reduced b value on image quality and diagnostic effectiveness. AJR Am J Roentgenol 207:330–338

  14. 14.

    Oto A, Kayhan A, Jiang Y et al (2010) Prostate cancer: differentiation of central gland cancer from benign prostatic hyperplasia by using diffusion-weighted and dynamic contrast-enhanced MR imaging. Radiology 257:715–723

  15. 15.

    Cho BH, Yu H, Lee J, Chee YJ, Kim IY, Kim SI (2008) Nonlinear support vector machine visualization for risk factor analysis using nomograms and localized radial basis function kernels. IEEE Trans Inf Technol Biomed 12:247–256

  16. 16.

    Majumder SK, Ghosh N, Gupta PK (2005) Support vector machine for optical diagnosis of cancer. J Biomed Opt 10:024034

  17. 17.

    Fehr D, Veeraraghavan H, Wibmer A et al (2015) Automatic classification of prostate cancer Gleason scores from multiparametric magnetic resonance images. Proc Natl Acad Sci U S A 112:E6265–E6273

  18. 18.

    Larroza A, Moratal D, Paredes-Sanchez A et al (2015) Support vector machine classification of brain metastasis and radiation necrosis based on texture analysis in MRI. J Magn Reson Imaging 42:1362–1368

  19. 19.

    Cawley GC, Talbot NL (2004) Fast exact leave-one-out cross-validation of sparse least-squares support vector machines. Neural Netw 17:1467–1475

  20. 20.

    Vickers AJ, Elkin EB (2006) Decision curve analysis: a novel method for evaluating prediction models. Med Decis Making 26:565–574

  21. 21.

    Baldisserotto M, Neto EJ, Carvalhal G et al (2016) Validation of PI-RADS v.2 for prostate cancer diagnosis with MRI at 3T using an external phased-array coil. J Magn Reson Imaging. doi:10.1002/jmri.25284

  22. 22.

    Kasel-Seibert M, Lehmann T, Aschenbach R et al (2016) Assessment of PI-RADS v2 for the Detection of Prostate Cancer. Eur J Radiol 85:726–731

  23. 23.

    Rosenkrantz AB, Ginocchio LA, Cornfeld D et al (2016) Interobserver Reproducibility of the PI-RADS Version 2 Lexicon: A Multicenter Study of Six Experienced Prostate Radiologists. Radiology. doi:10.1148/radiol.2016152542:152542

  24. 24.

    Sung YS, Kwon HJ, Park BW et al (2011) Prostate cancer detection on dynamic contrast-enhanced MRI: computer-aided diagnosis versus single perfusion parameter maps. AJR Am J Roentgenol 197:1122–1129

  25. 25.

    Niaf E, Rouviere O, Mege-Lechevallier F, Bratan F, Lartizien C (2012) Computer-aided diagnosis of prostate cancer in the peripheral zone using multiparametric MRI. Phys Med Biol 57:3833–3851

  26. 26.

    Koyasu S, Iima M, Umeoka S et al (2014) The clinical utility of reduced-distortion readout-segmented echo-planar imaging in the head and neck region: initial experience. Eur Radiol 24:3088–3096

  27. 27.

    Tokoro H, Fujinaga Y, Ohya A et al (2014) Usefulness of free-breathing readout-segmented echo-planar imaging (RESOLVE) for detection of malignant liver tumors: comparison with single-shot echo-planar imaging (SS-EPI). Eur J Radiol 83:1728–1733

  28. 28.

    Bogner W, Pinker K, Zaric O et al (2015) Bilateral diffusion-weighted MR imaging of breast tumors with submillimeter resolution using readout-segmented echo-planar imaging at 7 T. Radiology 274:74–84

  29. 29.

    Rosenkrantz AB, Prabhu V, Sigmund EE, Babb JS, Deng FM, Taneja SS (2013) Utility of diffusional kurtosis imaging as a marker of adverse pathologic outcomes among prostate cancer active surveillance candidates undergoing radical prostatectomy. AJR Am J Roentgenol 201:840–846

  30. 30.

    Rosenkrantz AB, Sigmund EE, Johnson G et al (2012) Prostate cancer: feasibility and preliminary experience of a diffusional kurtosis model for detection and assessment of aggressiveness of peripheral zone cancer. Radiology 264:126–135

  31. 31.

    Suo S, Chen X, Wu L et al (2014) Non-Gaussian water diffusion kurtosis imaging of prostate cancer. Magn Reson Imaging 32:421–427

  32. 32.

    Quentin M, Pentang G, Schimmoller L et al (2014) Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: preliminary results. Magn Reson Imaging 32:880–885

  33. 33.

    Tamura C, Shinmoto H, Soga S et al (2014) Diffusion kurtosis imaging study of prostate cancer: preliminary findings. J Magn Reson Imaging 40:723–729

  34. 34.

    Lin X, Yang F, Zhou L et al (2012) A support vector machine-recursive feature elimination feature selection method based on artificial contrast variables and mutual information. J Chromatogr B Analyt Technol Biomed Life Sci 910:149–155

  35. 35.

    Roethke MC, Kuru TH, Schultze S et al (2014) Evaluation of the ESUR PI-RADS scoring system for multiparametric MRI of the prostate with targeted MR/TRUS fusion-guided biopsy at 3.0 Tesla. Eur Radiol 24:344–352

  36. 36.

    Junker D, Quentin M, Nagele U et al (2015) Evaluation of the PI-RADS scoring system for mpMRI of the prostate: a whole-mount step-section analysis. World J Urol 33:1023–1030

  37. 37.

    Wang R, Wang H, Zhao C et al (2015) Evaluation of Multiparametric Magnetic Resonance Imaging in Detection and Prediction of Prostate Cancer. PLoS One 10, e0130207

  38. 38.

    Zhang L, Li Y, Jin Z, Yu JC, Chan KM (2015) An NIR-triggered and thermally responsive drug delivery platform through DNA/copper sulfide gates. Nanoscale 7:12614–12624

Download references

Author information

Correspondence to Yu-Dong Zhang.

Ethics declarations

Guarantor

The scientific guarantor of this publication is Y. Zhang.

Conflict of interest

The authors of this manuscript declare no relationships with any companies whose products or services may be related to the subject matter of the article.

Funding

This research was supported by A Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD; JX10231801; Z.Y.D.) and China Postdoctoral Fund (2015 M580453 to Y. Zhang).

Statistics and biometry

One of the authors (Y. Zhang) has significant statistical expertise.

No complex statistical methods were necessary for this paper.

Ethical approval

Institutional Review Board approval was obtained.

Informed consent

Written informed consent was waived by the Institutional Review Board.

Methodology

• retrospective

• diagnostic or prognostic study

• performed at one institution

Electronic supplementary material

Below is the link to the electronic supplementary material.

ESM 1

(DOCX 23 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Wang, J., Wu, C., Bao, M. et al. Machine learning-based analysis of MR radiomics can help to improve the diagnostic performance of PI-RADS v2 in clinically relevant prostate cancer. Eur Radiol 27, 4082–4090 (2017) doi:10.1007/s00330-017-4800-5

Download citation

Keywords

  • Prostate cancer
  • Prostate Imaging Reporting and Data System v2
  • Machine learning
  • Support vector machine
  • Multi-parametric MRI