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

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 (T 1Gd and T 2) 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 T 1Gd images, and LGGs had a more heterogeneous texture than HGGs in the T 2 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.

Bias and Class Imbalance in Oncologic Data—Towards Inclusive and Transferrable AI in Large Scale Oncology Data Sets

“…Their method achieved 37.83% AUC (Area under the Curve) improvement for brain magnetic resonance imaging (MRI) images. In another study, Suarez-Garcia et al also developed a basic model for glioma grading based on texture analysis using the under-sampling method to handle class imbalance data [ 38 ].…”

A Novel System for Precise Grading of Glioma

“…Despite the fact that the biopsy is the reference standard for identifying the grade of gliomas, it is not favorable because of high invasiveness, expense, and its adverse effects such as bleeding and infection. Therefore, many researchers were motivated to investigate imaging techniques for a non-invasive, early, and precise grading of gliomas for a timely management plan [ 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 ]. In particular, magnetic resonance imaging (MRI) is the most common imaging modality for the diagnosis and assessment of cerebral neoplasms, including gliomas.…”

“…They reported an overall accuracy of 89.8%. Suarez-Garcia et al [ 18 ] investigated the role of multimodal-MRI along with texture features to identify HGG from LGG. A total of 285 subjects were obtained (HGG = 210 and LGG = 75) from the BRATs 2018 challenge.…”

“…Despite the fact that these papers represent the state-of-the-art research studies for grading gliomas using MRIs along with computer-aided diagnostic (CAD) systems, they had some limitations that need to be addressed. Most of these studies [ 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 ] only focused on differentiating between LGG and HGG instead of grading these tumors into Grades I, II, III, or IV for a proper management plan. In addition, none of the aforementioned studies incorporated any functional features for a better diagnostic performance.…”

“…In addition, none of the aforementioned studies incorporated any functional features for a better diagnostic performance. Some of the authors [ 16 , 18 , 20 , 21 ] incorporated texture features in their work, but they did not investigate any shape or appearance features which could have led to higher performance as well.…”