Tumor Grade and Overall Survival Prediction of Gliomas Using Radiomics

Abstract

Glioma is one of the most common and deadly malignant brain tumors originating from glial cells. For personalized treatment, an accurate preoperative prognosis for glioma patients is highly desired. Recently, various machine learning-based approaches have been developed to predict the prognosis based on preoperative magnetic resonance imaging (MRI) radiomics, which extract quantitative features from radiographic images. However, major challenges remain for methodologic developments to optimize feature extraction and provide rapid information flow in clinical settings. This study investigates two machine learning-based prognosis prediction tasks using radiomic features extracted from preoperative multimodal MRI brain data: (i) prediction of tumor grade (higher-grade vs. lower-grade gliomas) from preoperative MRI scans and (ii) prediction of patient overall survival (OS) in higher-grade gliomas (<12 months vs. > 12 months) from preoperative MRI scans. Specifically, these two tasks utilize the conventional machine learning-based models built with various classifiers. Moreover, feature selection methods are applied to increase model performance and decrease computational costs. In the experiments, models are evaluated in terms of their predictive performance and stability using a bootstrap approach. Experimental results show that classifier choice and feature selection technique plays a significant role in model performance and stability for both tasks; a variability analysis indicates that classification method choice is the most dominant source of performance variation for both tasks.

Application of feed-forward neural network approaches to ApplicationApplication of feed-forward neural network approaches to radiomics-based survival analysis in glioma patients radiomics-based survival analysis in glioma patients

“…2.2, the shape features were only calculated for the whole glioma tumor regions and are independent of the pulse sequence used [27]. Patient age was also found to be correlated with the OS, which agrees with has been established in other studies [2,13,15,18].…”

“…The purpose of the Pearson correlation analysis in this study was twofold: (a) identify imaging biomarkers that may improve OS prediction in glioma patients, and (b) select relevant covariates for the subsequent neural networkbased survival analysis. Feature selection was employed in order to maximize prediction performance in some ML models while minimizing computational cost [15,32]. Fig.…”

“…ML takes advantage of statistical analysis and data-driven optimization methods to incorporate pattern recognition in mapping a set of input variables to the output [6,14]. An example of such ML applications in glioma management is the prediction of tumor grade and OS based on genomics, radiomics and clinical data [15]. Another widely used ML application is the automatic segmentation of tumor subregions in multimodality images [16].…”

Comprehensive Benchmarking of CNN-Based Tumor Segmentation Methods Using Multimodal MRI Data

“…The evolution of radiomics has significantly contributed to transforming radiological images into analysable data, enabling the extraction of diagnostic and prognostic information (Habib et al, 2021; Singh et al, 2021). In the current research, investigators employ multimodal MRI images to depict glioblastoma’s intrinsic heterogeneity and phenotype (Cui et al, n.d.; Ye et al, 2021). Regions exhibiting hypo/hyper-intensity in various MRI modalities are crucial in providing complementary profiles of glioblastoma subregions (Ellingson, 2015; Liu et al, 2021).…”