Radiomics in glioma: emerging trends and challenges

Abstract

Radiomics is a promising neuroimaging technique for extracting and analyzing quantitative glioma features. This review discusses the application, emerging trends, and challenges associated with using radiomics in glioma. Integrating deep learning algorithms enhances various radiomics components, including image normalization, region of interest segmentation, feature extraction, feature selection, and model construction and can potentially improve model accuracy and performance. Moreover, investigating specific tumor habitats of glioblastomas aids in a better understanding of glioblastoma aggressiveness and the development of effective treatment strategies. Additionally, advanced imaging techniques, such as diffusion-weighted imaging, perfusion-weighted imaging, magnetic resonance spectroscopy, magnetic resonance fingerprinting, functional MRI, and positron emission tomography, can provide supplementary information for tumor characterization and classification. Furthermore, radiomics analysis helps understand the glioma immune microenvironment by predicting immune-related biomarkers and characterizing immune responses within tumors. Integrating multi-omics data, such as genomics, transcriptomics, proteomics, and pathomics, with radiomics, aids the understanding of the biological significance of the underlying radiomics features and improves the prediction of genetic mutations, prognosis, and treatment response in patients with glioma. Addressing challenges, such as model reproducibility, model generalizability, model interpretability, and multi-omics data integration, is crucial for the clinical translation of radiomics in glioma.

Figure 1

Radiomic workflow and emerging trends. The radiomic workflow begins with image acquisition and reconstruction. These images undergo preprocessing through coregistration and normalization (* other preprocessing step include motor correction, bias field correction, resampling, and intensity standardization). 2D ROI and 3D VOI are then segmented for focus analysis. Quantitative features, such as shape, texture, and transformed features are extracted and filtered to construction predictive models using machine learning techniques. Emerging trends in radiomic research include incorporation of advanced MRI sequence or advanced imaging modality. Moreover, application of deep learning technique enable auto‐segmentation and the direct use of image for model training (deep radiomics). In addition, radiomic studies have increasingly focused on investigating specific habitat rather than the whole tumor volume.

Figure 2

Visualization of infiltration margin of glioma utilizing radiogenomics technique. Co‐registration of cellularity features obtained from localized biopsies following pathological examination with preoperative MR images to yield a predictive map of the glioma infiltration on preoperative MRI.

Figure 3

Visualization of intratumoral heterogeneity. By co‐registering molecular features obtained from localized biopsies after immunohistochemistry staining or by pairing sequencing data with preoperative MR images, a heatmap can be generated to represent the variation in molecular or pathway alteration across the glioma.

Figure 4

Proposed workflow for integrating multi‐omics data into radiomics. Various types of multi‐omics data, such as MRI, genomic sequencing, transcriptomic sequencing, proteomics, and pathological data, are collected using specific methods. Features are extracted and selected, and signatures are constructed using machine and deep learning algorithms. Multi‐omics data can be integrated by either early fusion or late fusion to create a fused model. Before validation in a prospective cohort, the final model can be visualized using explainable artificial intelligence techniques.