Advertisement

European Radiology

, Volume 28, Issue 4, pp 1625–1633 | Cite as

Machine learning-based quantitative texture analysis of CT images of small renal masses: Differentiation of angiomyolipoma without visible fat from renal cell carcinoma

  • Zhichao Feng
  • Pengfei Rong
  • Peng Cao
  • Qingyu Zhou
  • Wenwei Zhu
  • Zhimin Yan
  • Qianyun Liu
  • Wei Wang
Urogenital

Abstract

Objective

To evaluate the diagnostic performance of machine-learning based quantitative texture analysis of CT images to differentiate small (≤ 4 cm) angiomyolipoma without visible fat (AMLwvf) from renal cell carcinoma (RCC).

Methods

This single-institutional retrospective study included 58 patients with pathologically proven small renal mass (17 in AMLwvf and 41 in RCC groups). Texture features were extracted from the largest possible tumorous regions of interest (ROIs) by manual segmentation in preoperative three-phase CT images. Interobserver reliability and the Mann-Whitney U test were applied to select features preliminarily. Then support vector machine with recursive feature elimination (SVM-RFE) and synthetic minority oversampling technique (SMOTE) were adopted to establish discriminative classifiers, and the performance of classifiers was assessed.

Results

Of the 42 extracted features, 16 candidate features showed significant intergroup differences (P < 0.05) and had good interobserver agreement. An optimal feature subset including 11 features was further selected by the SVM-RFE method. The SVM-RFE+SMOTE classifier achieved the best performance in discriminating between small AMLwvf and RCC, with the highest accuracy, sensitivity, specificity and AUC of 93.9 %, 87.8 %, 100 % and 0.955, respectively.

Conclusion

Machine learning analysis of CT texture features can facilitate the accurate differentiation of small AMLwvf from RCC.

Key Points

Although conventional CT is useful for diagnosis of SRMs, it has limitations.

Machine-learning based CT texture analysis facilitate differentiation of small AMLwvf from RCC.

The highest accuracy of SVM-RFE+SMOTE classifier reached 93.9 %.

Texture analysis combined with machine-learning methods might spare unnecessary surgery for AMLwvf.

Keywords

Angiomyolipoma Renal cell carcinoma Computed tomography Texture analysis Machine learning 

Abbreviations

ACC

Accuracy

AMLwvf

Angiomyolipoma without visible fat

AUC

Area under the curve

CMP

Corticomedullary phase

FOV

Field of view

GLCM

Grey-level co-occurrence matrix

ICC

Interobserver agreement

NP

Nephrographic phase

PACS

Picture archiving and communication system

RBF

Radial basis function

RCC

Renal cell carcinoma

RFE

Recursive feature elimination

ROC

Receiver operating characteristic

ROI

Region of interest

SMOTE

Synthetic minority oversampling technique

SRM

Small renal mass

SVM

Support vector machine

UP

Unenhanced phase

Notes

Funding

The authors state that this work has not received any funding.

Compliance with ethical standards

Guarantor

The scientific guarantor of this publication is Zhichao Feng, M.D.

Conflict of interest

The authors of this manuscript declare a relationship with the following company: GE Healthcare.

Peng Cao is a senior scientist for GE Healthcare (Shanghai, China) and provided the software and necessary training for this study. He has no intention to apply for a patent based on this paper or invent any product, and did not provide any financial support. GE Healthcare did not play any additional role in the study design, data collection and analysis, or preparation of the manuscript. There are no other author disclosures, and the other authors (Zhichao Feng, Pengfei Rong, Qingyu Zhou, Wenwei Zhu, Zhimin Yan, Qianyun Liu and Wei Wang) have no conflicts of interest.

Statistics and biometry

Pengfei Rong and Wei Wang kindly provided statistical advice for this manuscript.

Informed consent

Written informed consent was waived by the Institutional Review Board.

Ethical approval

Institutional Review Board approval was obtained.

Methodology

• retrospective

• diagnostic or prognostic study

• performed at one institution

References

  1. 1.
    Kutikov A, Fossett LK, Ramchandani P et al (2006) Incidence of benign pathologic findings at partial nephrectomy for solitary renal mass presumed to be renal cell carcinoma on preoperative imaging. Urology 68:737–740CrossRefPubMedGoogle Scholar
  2. 2.
    Fujii Y, Komai Y, Saito K et al (2008) Incidence of benign pathologic lesions at partial nephrectomy for presumed RCC renal masses: Japanese dual-center experience with 176 consecutive patients. Urology 72:598–602CrossRefPubMedGoogle Scholar
  3. 3.
    Flum AS, Hamoui N, Said MA et al (2016) Update on the Diagnosis and Management of Renal Angiomyolipoma. J Urol 195:834–846CrossRefPubMedGoogle Scholar
  4. 4.
    Umbreit EC, Shimko MS, Childs MA et al (2012) Metastatic potential of a renal mass according to original tumour size at presentation. Bju Int 109:190–194 discussion 194CrossRefPubMedGoogle Scholar
  5. 5.
    Kim JK, Park SY, Shon JH et al (2004) Angiomyolipoma with minimal fat: differentiation from renal cell carcinoma at biphasic helical CT. Radiology 230:677–684CrossRefPubMedGoogle Scholar
  6. 6.
    Zhang YY, Luo S, Liu Y et al (2013) Angiomyolipoma with minimal fat: differentiation from papillary renal cell carcinoma by helical CT. Clin Radiol 68:365–370CrossRefPubMedGoogle Scholar
  7. 7.
    Hakim SW, Schieda N, Hodgdon T et al (2016) Angiomyolipoma (AML) without visible fat: Ultrasound, CT and MR imaging features with pathological correlation. Eur Radiol 26:592–600CrossRefPubMedGoogle Scholar
  8. 8.
    Yang CW, Shen SH, Chang YH et al (2013) Are there useful CT features to differentiate renal cell carcinoma from lipid-poor renal angiomyolipoma? AJR Am J Roentgenol 201:1017–1028CrossRefPubMedGoogle Scholar
  9. 9.
    Xie P, Yang Z, Yuan Z (2016) Lipid-poor renal angiomyolipoma: Differentiation from clear cell renal cell carcinoma using wash-in and washout characteristics on contrast-enhanced computed tomography. Oncol Lett 11:2327–2331CrossRefPubMedPubMedCentralGoogle Scholar
  10. 10.
    Davnall F, Yip CS, Ljungqvist G et al (2012) Assessment of tumor heterogeneity: an emerging imaging tool for clinical practice? Insights Imaging 3:573–589CrossRefPubMedPubMedCentralGoogle Scholar
  11. 11.
    Hodgdon T, Mcinnes MD, Schieda N et al (2015) Can Quantitative CT Texture Analysis be Used to Differentiate Fat-poor Renal Angiomyolipoma from Renal Cell Carcinoma on Unenhanced CT Images? Radiology 276:787–796CrossRefPubMedGoogle Scholar
  12. 12.
    Yan L, Liu Z, Wang G et al (2015) Angiomyolipoma with minimal fat: differentiation from clear cell renal cell carcinoma and papillary renal cell carcinoma by texture analysis on CT images. Acad Radiol 22:1115–1121CrossRefPubMedGoogle Scholar
  13. 13.
    Kumar V, Gu Y, Basu S et al (2012) Radiomics: the process and the challenges. Magn Reson Imaging 30:1234–1248CrossRefPubMedPubMedCentralGoogle Scholar
  14. 14.
    Parmar C, Grossmann P, Bussink J et al (2015) Machine Learning methods for Quantitative Radiomic Biomarkers. Sci Rep 5:13087CrossRefPubMedPubMedCentralGoogle Scholar
  15. 15.
    Zhang X, Yan LF, Hu YC et al (2017) Optimizing a machine learning based glioma grading system using multi-parametric MRI histogram and texture features. Oncotarget 8:47816–47830PubMedPubMedCentralGoogle Scholar
  16. 16.
    Nketiah G, Elschot M, Kim E et al (2017) T2-weighted MRI-derived textural features reflect prostate cancer aggressiveness: preliminary results. Eur Radiol 27:3050–3059CrossRefPubMedGoogle Scholar
  17. 17.
    Wibmer A, Hricak H, Gondo T et al (2015) Haralick texture analysis of prostate MRI: utility for differentiating non-cancerous prostate from prostate cancer and differentiating prostate cancers with different Gleason scores. Eur Radiol 25:2840–2850CrossRefPubMedPubMedCentralGoogle Scholar
  18. 18.
    Aerts HJ, Velazquez ER, Leijenaar RT et al (2014) Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun 5:4006PubMedPubMedCentralGoogle Scholar
  19. 19.
    Guyon I, Weston J, Barnhill S et al (2002) Gene selection for cancer classification using support vector machines. Mach Learn 46:389–422CrossRefGoogle Scholar
  20. 20.
    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 112:E6265–E6273CrossRefPubMedPubMedCentralGoogle Scholar
  21. 21.
    Wang J, Wu C J, Bao M L et al (2017) 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.  https:/doi.org/10.1007/s00330-017-4800-5
  22. 22.
    Zhang X, Xu X, Tian Q, et al (2017) Radiomics assessment of bladder cancer grade using texture features from diffusion-weighted imaging. J Magn Reson Imaging​ 46: 1281-1288Google Scholar
  23. 23.
    Chawla NV, Bowyer KW, Hall LO et al (2011) SMOTE: synthetic minority over-sampling technique. J Artif Intell Res 16:321–357Google Scholar
  24. 24.
    Zhang Y, Oikonomou A, Wong A et al (2017) Radiomics-based Prognosis Analysis for Non-Small Cell Lung Cancer. Sci Rep 7:46349CrossRefPubMedPubMedCentralGoogle Scholar
  25. 25.
    Yu H, Scalera J, Khalid M et al (2017) Texture analysis as a radiomic marker for differentiating renal tumors. Abdom Radiol (NY).  https:/doi.org/10.1007/s00261-017-1144-1
  26. 26.
    Lee HS, Hong H, Jung DC et al (2017) Differentiation of fat-poor angiomyolipoma from clear cell renal cell carcinoma in contrast-enhanced MDCT images using quantitative feature classification. Med Phys 44:3604–3614CrossRefPubMedGoogle Scholar
  27. 27.
    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 910:149–155CrossRefPubMedGoogle Scholar
  28. 28.
    Jinzaki M, Tanimoto A, Narimatsu Y et al (1997) Angiomyolipoma: imaging findings in lesions with minimal fat. Radiology 205:497–502CrossRefPubMedGoogle Scholar
  29. 29.
    Tsukada J, Jinzaki M, Yao M et al (2013) Epithelioid angiomyolipoma of the kidney: radiological imaging. Int J Urol 20:1105–1111CrossRefPubMedGoogle Scholar
  30. 30.
    Ishigami K, Pakalniskis MG, Leite LV et al (2015) Characterization of renal cell carcinoma, oncocytoma, and lipid-poor angiomyolipoma by unenhanced, nephrographic, and delayed phase contrast-enhanced computed tomography. Clin Imaging 39:76–84CrossRefPubMedGoogle Scholar
  31. 31.
    Silverman SG, Mortele KJ, Tuncali K et al (2007) Hyperattenuating renal masses: etiologies, pathogenesis, and imaging evaluation. Radiographics 27:1131–1143CrossRefPubMedGoogle Scholar
  32. 32.
    Liu S, Zheng H, Zhang Y et al (2017) Whole-volume apparent diffusion coefficient-based entropy parameters for assessment of gastric cancer aggressiveness. J Magn Reson Imaging.  https:/doi.org/10.1002/jmri.25752
  33. 33.
    Choi ER, Lee HY, Jeong JY et al (2016) Quantitative image variables reflect the intratumoral pathologic heterogeneity of lung adenocarcinoma. Oncotarget 7:67302–67313PubMedPubMedCentralGoogle Scholar
  34. 34.
    Sidhu HS, Benigno S, Ganeshan B et al (2017) Textural analysis of multiparametric MRI detects transition zone prostate cancer. Eur Radiol 27:2348–2358CrossRefPubMedGoogle Scholar
  35. 35.
    Hanahan D, Weinberg RA (2011) Hallmarks of cancer: the next generation. Cell 144:646–674CrossRefPubMedGoogle Scholar
  36. 36.
    Johnson PT, Horton KM, Fishman EK (2010) How not to miss or mischaracterize a renal cell carcinoma: protocols, pearls, and pitfalls. AJR Am J Roentgenol 194:W307–W315CrossRefPubMedGoogle Scholar
  37. 37.
    Takahashi N, Leng S, Kitajima K et al (2015) Small (< 4 cm) Renal Masses: Differentiation of Angiomyolipoma Without Visible Fat From Renal Cell Carcinoma Using Unenhanced and Contrast-Enhanced CT. AJR Am J Roentgenol 205:1194–1202CrossRefPubMedGoogle Scholar
  38. 38.
    Xu X, Liu Y, Zhang X et al (2017) Preoperative prediction of muscular invasiveness of bladder cancer with radiomic features on conventional MRI and its high-order derivative maps. Abdom Radiol (NY) 42:1896–1905CrossRefGoogle Scholar
  39. 39.
    Yip SSF, Parmar C, Blezek D et al (2017) Application of the 3D slicer chest imaging platform segmentation algorithm for large lung nodule delineation. Plos One 12:e0178944CrossRefPubMedPubMedCentralGoogle Scholar
  40. 40.
    Ng F, Kozarski R, Ganeshan B et al (2013) Assessment of tumor heterogeneity by CT texture analysis: can the largest cross-sectional area be used as an alternative to whole tumor analysis? Eur J Radiol 82:342–348CrossRefPubMedGoogle Scholar

Copyright information

© European Society of Radiology 2017

Authors and Affiliations

  • Zhichao Feng
    • 1
  • Pengfei Rong
    • 1
  • Peng Cao
    • 2
  • Qingyu Zhou
    • 1
  • Wenwei Zhu
    • 1
  • Zhimin Yan
    • 1
  • Qianyun Liu
    • 1
  • Wei Wang
    • 1
  1. 1.Department of Radiology, The Third Xiangya HospitalCentral South UniversityChangshaChina
  2. 2.GE HealthcareShanghaiChina

Personalised recommendations