Skip to main content

A quantitative model based on clinically relevant MRI features differentiates lower grade gliomas and glioblastoma

European Radiology volume 30, pages 3073–3082 (2020)Cite this article

We’re sorry, something doesn't seem to be working properly.

Please try refreshing the page. If that doesn't work, please contact support so we can address the problem.

A Correction to this article was published on 09 March 2020

This article has been updated

Abstract

Objectives

To establish a quantitative MR model that uses clinically relevant features of tumor location and tumor volume to differentiate lower grade glioma (LRGG, grades II and III) and glioblastoma (GBM, grade IV).

Methods

We extracted tumor location and tumor volume (enhancing tumor, non-enhancing tumor, peritumor edema) features from 229 The Cancer Genome Atlas (TCGA)-LGG and TCGA-GBM cases. Through two sampling strategies, i.e., institution-based sampling and repeat random sampling (10 times, 70% training set vs 30% validation set), LASSO (least absolute shrinkage and selection operator) regression and nine–machine learning method–based models were established and evaluated.

Results

Principal component analysis of 229 TCGA-LGG and TCGA-GBM cases suggested that the LRGG and GBM cases could be differentiated by extracted features. For nine machine learning methods, stack modeling and support vector machine achieved the highest performance (institution-based sampling validation set, AUC > 0.900, classifier accuracy > 0.790; repeat random sampling, average validation set AUC > 0.930, classifier accuracy > 0.850). For the LASSO method, regression model based on tumor frontal lobe percentage and enhancing and non-enhancing tumor volume achieved the highest performance (institution-based sampling validation set, AUC 0.909, classifier accuracy 0.830). The formula for the best performance of the LASSO model was established.

Conclusions

Computer-generated, clinically meaningful MRI features of tumor location and component volumes resulted in models with high performance (validation set AUC > 0.900, classifier accuracy > 0.790) to differentiate lower grade glioma and glioblastoma.

Key Points

• Lower grade glioma and glioblastoma have significant different location and component volume distributions.

• We built machine learning prediction models that could help accurately differentiate lower grade gliomas and GBM cases. We introduced a fast evaluation model for possible clinical differentiation and further analysis.

This is a preview of subscription content, access via your institution.

Access options

Buy single article

Instant access to the full article PDF.

USD 39.95

Price includes VAT (USA)
Tax calculation will be finalised during checkout.

Fig. 1
Fig. 2
Fig. 3

Change history

  • 09 March 2020

    The original version of this article, published on 05 February 2020, unfortunately contained a mistake.

Abbreviations

AUC:

Area under ROC

BraTS:

Multimodal Brain Tumor Segmentation Challenge

CA:

Classification accuracy

FLAIR:

Fluid-attenuated inversion recovery

FMRIB:

Functional magnetic resonance imaging of the brain

FSL:

FMRIB Software Library

GBM:

Glioblastoma

IG:

Information gain

KNN:

k-nearest neighbors algorithm

LASSO:

Least absolute shrinkage and selection operator

LGG:

Low-grade glioma

LRGG:

Lower grade glioma

MNI:

Montreal Neurological Institute

NIfTI:

Neuroimaging Informatics Technology Initiative

PCs:

Principal components

PCA:

Principal component analysis

RF:

Random forest

SVM:

Support vector machine

TCGA:

The Cancer Genome Atlas

TCIA:

The Cancer Imaging Archive

Volume_ED:

Peritumoral edema tumor volume

Volume_ET:

Enhancing tumor volume

Volume_NET:

Non-enhancing tumor volume

Volume_TC:

Tumor core volume

Volume_WT:

Whole tumor volume

References

  1. Schiff D, Van den Bent M, Vogelbaum MA et al (2019) Recent developments and future directions in adult lower-grade gliomas: Society for Neuro-Oncology (SNO) and European Association of Neuro-Oncology (EANO) consensus. Neuro Oncol 21:837–853

    Article  Google Scholar 

  2. Asari S, Makabe T, Katayama S, Itoh T, Tsuchida S, Ohmoto T (1994) Assessment of the pathological grade of astrocytic gliomas using an MRI score. Neuroradiology 36:308–310

    CAS  Article  Google Scholar 

  3. Pierallini A, Bonamini M, Bozzao A et al (1997) Supratentorial diffuse astrocytic tumours: proposal of an MRI classification. Eur Radiol 7:395–399

    CAS  Article  Google Scholar 

  4. Cho H-H, Park H (2017) Classification of low-grade and highgrade glioma using multi-modal image radiomics features. Conf Proc IEEE Eng Med Biol Soc 2017:3081–3084

  5. Togao O, Hiwatashi A, Yamashita K et al (2015) Differentiation of high-grade and low-grade diffuse gliomas by intravoxel incoherent motion MR imaging. Neuro Oncol 18:132–141

    Article  Google Scholar 

  6. Rizzo S, Botta F, Raimondi S et al (2018) Radiomics: the facts and the challenges of image analysis. Eur Radiol Exp 2:36

    Article  Google Scholar 

  7. Larjavaara S, Mäntylä R, Salminen T et al (2007) Incidence of gliomas by anatomic location. Neuro Oncol 9:319–325

    Article  Google Scholar 

  8. Carlson MR, Pope WB, Horvath S et al (2007) Relationship between survival and edema in malignant gliomas: role of vascular endothelial growth factor and neuronal pentraxin 2. Clin Cancer Res 13:2592–2598

    CAS  Article  Google Scholar 

  9. Carrillo J, Lai A, Nghiemphu P et al (2012) Relationship between tumor enhancement, edema, IDH1 mutational status, MGMT promoter methylation, and survival in glioblastoma. AJNR Am J Neuroradiol 33:1349–1355

    CAS  Article  Google Scholar 

  10. Clark K, Vendt B, Smith K et al (2013) The Cancer Imaging Archive (TCIA): maintaining and operating a public information repository. J Digit Imaging 26:1045–1057

    Article  Google Scholar 

  11. Koçak B, Durmaz ES, Ates E, Kilickesmez O (2019) Radiomics with artificial intelligence: a practical guide for beginners. Diagn Interv Radiol. https://doi.org/10.5152/dir.2019.19321

  12. Bakas S, Zeng K, Sotiras A et al (2015) GLISTRboost: combining multimodal MRI segmentation, registration, and biophysical tumor growth modeling with gradient boosting machines for glioma segmentation. Brainlesion 9556:144–155

  13. Grabner G, Janke AL, Budge MM, Smith D, Pruessner J, Collins DL (2006) Symmetric atlasing and model based segmentation: an application to the hippocampus in older adults. In: Larsen R, Nielsen M, Sporring J (eds) Medical Image Computing and Computer-Assisted Intervention – MICCAI 2006.Lecture Notes in Computer Science, vol 4191. Springer, Berlin, Heidelberg

  14. Woolrich MW, Jbabdi S, Patenaude B et al (2009) Bayesian analysis of neuroimaging data in FSL. Neuroimage 45:S173–S186

    Article  Google Scholar 

  15. Mazziotta J, Toga A, Evans A et al (2001) A probabilistic atlas and reference system for the human brain: International Consortium for Brain Mapping (ICBM). Philos Trans R Soc Lond B Biol Sci 356:1293–1322

    CAS  Article  Google Scholar 

  16. Bakas S, Akbari H, Sotiras A et al (2017) Advancing the cancer genome atlas glioma MRI collections with expert segmentation labels and radiomic features. Sci Data 4:170117

    Article  Google Scholar 

  17. Demšar J, Curk T, Erjavec A et al (2013) Orange: data mining toolbox in Python. J Mach Learn Res 14:2349–2353

    Google Scholar 

  18. Piotr Romanski LK (2018) Package FSelector - CRAN. CRAN. Available via https://cran.r-project.org/web/packages/FSelector/index.html. Accessed Oct 22 2019

  19. Friedman J, Hastie T, Tibshirani R (2010) Regularization paths for generalized linear models via coordinate descent. J Stat Softw 33:1

    Article  Google Scholar 

  20. Kuhn M (2008) Building predictive models in R using the caret package. J Stat Softw 28:1–26

    Article  Google Scholar 

  21. Robin X, Turck N, Hainard A et al (2011) pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics 12:77

    Article  Google Scholar 

  22. Hsieh KL-C, Lo C-M, Hsiao C-J (2017) Computer-aided grading of gliomas based on local and global MRI features. Comput Methods Programs Biomed 139:31–38

    Article  Google Scholar 

  23. Stroebe H (1895) Uber entstehung und bau der gehirngliome. Beitr Pathol Anat Allg Pathol 18:405–485

    Google Scholar 

  24. Burkhard C, Di Patre P-L, Schüler D et al (2003) A population-based study of the incidence and survival rates in patients with pilocytic astrocytoma. J Neurosurg 98:1170–1174

    Article  Google Scholar 

  25. Hargrave D, Bartels U, Bouffet E (2006) Diffuse brainstem glioma in children: critical review of clinical trials. Lancet Oncol 7:241–248

    Article  Google Scholar 

  26. Pallud J, Capelle L, Taillandier L et al (2009) Prognostic significance of imaging contrast enhancement for WHO grade II gliomas. Neuro Oncol 11:176–182

    Article  Google Scholar 

  27. Suchorska B, Schüller U, Biczok A et al (2019) Contrast enhancement is a prognostic factor in IDH1/2 mutant, but not in wild-type WHO grade II/III glioma as confirmed by machine learning. Eur J Cancer 107:15–27

  28. Wang Y, Wang K, Li S et al (2015) Patterns of tumor contrast enhancement predict the prognosis of anaplastic gliomas with IDH1 mutation. AJNR Am J Neuroradiol 36:2023–2029

    CAS  Article  Google Scholar 

  29. Scott J, Brasher P, Sevick R, Rewcastle N, Forsyth P (2002) How often are nonenhancing supratentorial gliomas malignant? A population study. Neurology 59:947–949

    CAS  Article  Google Scholar 

  30. Reyes-Botero G, Dehais C, Idbaih A et al (2013) Contrast enhancement in 1p/19q-codeleted anaplastic oligodendrogliomas is associated with 9p loss, genomic instability, and angiogenic gene expression. Neuro Oncol 16:662–670

    Article  Google Scholar 

Download references

Acknowledgments

The results here are in whole or part based upon data generated by the TCGA Research Network: https://www.cancer.gov/tcga.

Funding

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

Author information

Authors and Affiliations

  1. Department of Neurosurgery, Xiangya Hospital, Central South University, Changsha, 410008, China

    Hang Cao & Xuejun Li

  2. Department of Neurosurgery, Yale School of Medicine, New Haven, CT, 06520, USA

    E. Zeynep Erson-Omay, Murat Günel & Jennifer Moliterno

  3. Department of Radiology and Biomedical Imaging, MRRC, Yale School of Medicine, The Anlyan Center N137, PO Box 208043, New Haven, CT, 06520-8043, USA

    Robert K. Fulbright

Authors
  1. Hang Cao
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. E. Zeynep Erson-Omay
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Xuejun Li
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Murat Günel
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Jennifer Moliterno
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Robert K. Fulbright
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Robert K. Fulbright.

Ethics declarations

Guarantor

The scientific guarantor of this publication is Dr. Robert Fulbright.

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.

Statistics and biometry

No complex statistical methods were necessary for this paper.

Informed consent

Written informed consent was waived by the Institutional Review Board.

Ethical approval

Institutional Review Board approval was not required because only deidentified data is used in this study.

Study subjects or cohorts overlap

Some study subjects or cohorts have been previously reported. The TCGA-LGG and TCGA-GBM cohorts were involved in this study. Due to the enormous publishing scale, here we are not able to list or include every associated study. However, the cohort resource and include/exclude criteria have been introduced and cited in detail.

The major differences between the cohort used in our study and other study are as follows:

1. In our study, only TCGA-LGG and TCGA-GBM cases with MRI images available on TCIA were included.

2. The abovementioned cases were subsequently selected by segmentation data availability; only cases that have GLISTRboost-generated and manually checked segmentation data were selected.

Methodology

• retrospective

• diagnostic study

• multicenter study

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

The original version of this article was revised: The name of E. Zeynep Erson-Omay was presented incorrectly.

Electronic supplementary material

ESM 1

(DOCX 663 kb)

ESM 2

(XLSX 175 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Cao, H., Erson-Omay, E.Z., Li, X. et al. A quantitative model based on clinically relevant MRI features differentiates lower grade gliomas and glioblastoma. Eur Radiol 30, 3073–3082 (2020). https://doi.org/10.1007/s00330-019-06632-8

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00330-019-06632-8

Keywords

  • Glioma
  • Neoplasm grading
  • Tumor burden
  • Machine learning