Radiomics as an emerging tool in the management of brain metastases

Abstract

Brain metastases (BM) are associated with significant morbidity and mortality in patients with advanced cancer. Despite significant advances in surgical, radiation, and systemic therapy in recent years, the median overall survival of patients with BM is less than 1 year. The acquisition of medical images, such as computed tomography (CT) and magnetic resonance imaging (MRI), is critical for the diagnosis and stratification of patients to appropriate treatments. Radiomic analyses have the potential to improve the standard of care for patients with BM by applying artificial intelligence (AI) with already acquired medical images to predict clinical outcomes and direct the personalized care of BM patients. Herein, we outline the existing literature applying radiomics for the clinical management of BM. This includes predicting patient response to radiotherapy and identifying radiation necrosis, performing virtual biopsies to predict tumor mutation status, and determining the cancer of origin in brain tumors identified via imaging. With further development, radiomics has the potential to aid in BM patient stratification while circumventing the need for invasive tissue sampling, particularly for patients not eligible for surgical resection.

Keywords: artificial intelligence; brain metastases; radiology; radiomics.

Figure 1.

Radiomics pipelines utilizing a T2-weighted MRI image of a patient with BM, alongside the segmentation for the image. (A) A conventional handcrafted radiomic workflow in which images acquired from patients are manually, automatically, or semi-automatically segmented/contoured to delineate regions of interest. After applying perprocessing steps, such as normalization and denoising, the outlined tumors will undergo feature extraction using mathematical features such as first-order statistics, shape-based, and gray level co-occurrence matrix. After feature extraction, a classification model could be developed to predict the outcome of interest. (B) In a DL-based radiomic workflow, features are learned by a DL model using the available data for model training. Most often, a DL model consists of two components: a deep feature extractor followed by a classifier. The deep feature extractor component most often is a convolutional neural network (CNN), and the classifier component of the model is a shallow fully connected neural network. DL-based classification models often conduct a whole-image analysis and bypass the need for tumor segmentation. However, manually, automatically, or semi-automatically acquired segmentations could also be used in a DL-based workflow.

Figure 2.

Current applications of radiomics in brain metastases (BM) management. Radiomics has emerged as a powerful tool in the personalized management of brain metastases (BM). This includes discriminating BM from primary central nervous system (CNS) tumors, identifying the site of primary cancer of origin, discriminating radiation necrosis from recurrence, predicting tumor mutation status, and predicting patient response to stereotactic radiosurgery (SRS).