A simple model for glioma grading based on texture analysis applied to conventional brain MRI - PubMed Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020 May 15;15(5):e0228972.
doi: 10.1371/journal.pone.0228972. eCollection 2020.

A simple model for glioma grading based on texture analysis applied to conventional brain MRI

José Gerardo Suárez-García  1 , Javier Miguel Hernández-López  1 , Eduardo Moreno-Barbosa  1 , Benito de Celis-Alonso  1
Affiliations

A simple model for glioma grading based on texture analysis applied to conventional brain MRI

José Gerardo Suárez-García et al. PLoS One. .

Abstract

Accuracy of glioma grading is fundamental for the diagnosis, treatment planning and prognosis of patients. The purpose of this work was to develop a low-cost and easy-to-implement classification model which distinguishes low-grade gliomas (LGGs) from high-grade gliomas (HGGs), through texture analysis applied to conventional brain MRI. Different combinations of MRI contrasts (T1Gd and T2) and one segmented glioma region (necrotic and non-enhancing tumor core, NCR/NET) were studied. Texture features obtained from the gray level size zone matrix (GLSZM) were calculated. An under-sampling method was proposed to divide the data into different training subsets and subsequently extract complementary information for the creation of distinct classification models. The sensitivity, specificity and accuracy of the models were calculated, and the best model explicitly reported. The best model included only three texture features and reached a sensitivity, specificity and accuracy of 94.12%, 88.24% and 91.18%, respectively. According to the features of the model, when the NCR/NET region was studied, HGGs had a more heterogeneous texture than LGGs in the T1Gd images, and LGGs had a more heterogeneous texture than HGGs in the T2 images. These novel results partially contrast with results from the literature. The best model proved to be useful for the classification of gliomas. Complementary results showed that the heterogeneity of gliomas depended on the MRI contrast studied. The chosen model stands out as a simple, low-cost, easy-to-implement, reproducible and highly accurate glioma classifier. Importantly, it should be accessible to populations with reduced economic and scientific resources.

PubMed Disclaimer

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Fig 1
Fig 1. Division of data and under-sampling.
From the available 64 LGGS and 191 HGGs, one testing subset and 100 training subsets with balanced classes were created by randomly choosing gliomas.
Fig 2
Fig 2. Obtaining the ordered highest frequency features.
Considering the d first ordered features (according to their p-value) of each training subset, histograms of the features located in the same place were created. Then, from the histograms, the highest frequency features were obtained.
Fig 3
Fig 3. Flow diagram.
Complete process proposed for the classification of low-grade and high-grade gliomas.
Fig 4
Fig 4. Results of models.
Three graphs are shown with the results of combinations 1 (a), 3 (b) and 6 (c), using different numbers of variables (horizontal axis). These results consist of the percentages (vertical axis) of sensitivity, specificity and accuracy obtained after applying the models to the testing subset. Three further graphs (d, e and f) indicate the values (in arbitrary units, au) of the mean absolute errors (mae) obtained in each model. The best classification results were obtained in combination 6 by the model with five variables (▼, ▲).
Fig 5
Fig 5. Results from reduced models.
a. Percentages of sensitivity, specificity and accuracy (vertical axis), obtained by the 30 reduced models, in addition to the combination of variables utilized in each one (horizontal axis), using the following numbering: 1, Fszm.z.perc; 2, Fszm.zs.var; 3, Fszm.lzlge; 4, Fszm.lze; and 5, Fszm.zsnu. The first four were measured in T2 contrasts and the fifth in T1Gd contrasts. All features were measured in the NCR/NET region. b. Values (in arbitrary units, au) of the mean absolute errors (mae) obtained in each reduced model. The reduced model that obtained the best results with the lowest number of variables and the smallest error corresponded to the one that combined variables 1-2-5 (▼, ▲).
Fig 6
Fig 6. Predictions made by the best reduced model, when applied to the testing subset.
Testing gliomas (34 LGGs and 34 HGGs; vertical axis) and their predictions (in arbitrary units, au; horizontal axis) are presented. A solid vertical line at zero indicates the chosen threshold. Dotted vertical lines at -10 and 10 indicate the ideal prediction of the LGGs and HGGs, respectively. The filled circles and squares correspond to the true HGGs and true LGGs, respectively, and the empty circles and squares correspond to the false LGGs (or HGGs misclassified) and false HGGs (or LGGs misclassified), respectively.
Fig 7
Fig 7. Boxplots of the texture features or variables 1-2-5, calculated from the testing gliomas.
The grades of the testing gliomas (horizontal axis) and their texture values (in arbitrary units, au; vertical axis) are presented. a. Boxplot of feature number 1, Fszm.z.perc (measured in the T21 contrast). b. Boxplot of feature number 2, Fszm.zs.var (measured in the T21 contrast). c. Boxplot of feature number 5, Fszm.zsnu (measured in the T1Gd1 contrast).
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Bogdańska MU, Bodnar M, Piotrowska MJ, Murek M, Schucht P, Beck J, et al. A mathematical model describes the malignant transformation of low grade gliomas: Prognostic implications. PloS One. 2017;12(8):e0179999 10.1371/journal.pone.0179999 - DOI - PMC - PubMed
    1. Louis DN, Perry A, Reifenberger G, Von Deimling A, Figarella-Branger D, Cavenee WK, et al. The 2016 World Health Organization classification of tumors of the central nervous system: a summary. Acta Neuropathol. 2016;131(6):803–820. 10.1007/s00401-016-1545-1 - DOI - PubMed
    1. Rachinger W, Grau S, Holtmannspötter M, Herms J, Tonn JC, Kreth FW. Serial stereotactic biopsy of brainstem lesions in adults improves diagnostic accuracy compared with MRI only. J Neurol Neurosurg Psychiatry. 2009;80(10):1134–1139. 10.1136/jnnp.2009.174250 - DOI - PubMed
    1. Ho VK, Reijneveld JC, Enting RH, Bienfait HP, Robe P, Baumert BG, et al. Changing incidence and improved survival of gliomas. Eur J Cancer. 2014;50(13):2309–2318. 10.1016/j.ejca.2014.05.019 - DOI - PubMed
    1. Wang Q, Li Q, Mi R, Ye H, Zhang H, Chen B, et al. Radiomics Nomogram Building From Multiparametric MRI to Predict Grade in Patients With Glioma: A Cohort Study. J Magn Reson Imaging. 2018. - PubMed

Publication types

Grants and funding

The author JGSG was supported by the National Council of Science and Technology (https://www.conacyt.gob.mx/), through a scholarship for postgraduate studies (grant number 461568). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.