Radiomics in breast cancer: Current advances and future directions

Abstract

Breast cancer is a common disease that causes great health concerns to women worldwide. During the diagnosis and treatment of breast cancer, medical imaging plays an essential role, but its interpretation relies on radiologists or clinical doctors. Radiomics can extract high-throughput quantitative imaging features from images of various modalities via traditional machine learning or deep learning methods following a series of standard processes. Hopefully, radiomic models may aid various processes in clinical practice. In this review, we summarize the current utilization of radiomics for predicting clinicopathological indices and clinical outcomes. We also focus on radio-multi-omics studies that bridge the gap between phenotypic and microscopic scale information. Acknowledging the deficiencies that currently hinder the clinical adoption of radiomic models, we discuss the underlying causes of this situation and propose future directions for advancing radiomics in breast cancer research.

Graphical abstract

Figure 1

Overview of breast cancer radiomics The workflow of breast cancer radiomics includes four continuous parts: (1) Image acquisition and medical images from various modalities, including MG, US, MRI, and PET/CT, are acquired for further analysis. (2) Tumor segmentation, with manual or semiautomatic segmentation methods. (3) Feature extraction, hand-crafted, or DL features are extracted globally or locally from ROIs. (4) Model building and analysis include feature selection, modeling, and model validation. Abbreviations: MG, mammography; DBT, digital breast tomosynthesis; CEM, contrast-enhanced mammography; CESM, contrast-enhanced spectral mammography; US, ultrasound; ABUS, automated breast ultrasound; SWE, shear wave elastography; MRI, magnetic resonance imaging; ADC, apparent diffusion coefficient; DWI, diffusion-weighted images; DCE-MEI, dynamic contrast-enhanced magnetic resonance imaging; PET/CT, positron emission tomography-computed tomography; SUV, standardized uptake value; MTV, metabolic tumor volume; TLG, total lesion glycolysis; GLCM, gray-level co-occurrence matrix; GLRLM, gray-level run length matrix; NGTDM, neighborhood gray-tone difference matrix; CNN, convolutional neural network; SVM, support vector machine; LASSO, least absolute shrinkage and selection operator; LR, logistic regression; RF, random forest; ROIs, regions of interest; DL, deep learning; Grad-CAM, gradient-weighted class activation mapping.

Figure 2

Future directions for radiomics in breast cancer We propose the establishment of an all-in-one AI-aided radiology-clinic system, as well as optimizing validation methods and enhancing interpretability of radiomic models in breast cancer. For the all-in-one AI-aided radiology-clinic system, users input the unique identifier of a patient and select the needed function. For patients with screening demand, the DL section of the system utilizes well-established CAD systems to mark suspicious lesions for radiologists. Next, if the patient is confirmed malignant, the system automatically inputs the full-scale information, including clinicopathological indices, medical images, as well as multi-omics information. Then, the system automatically predicts the classification of lesion, molecular subtypes, lymph node status, as well as the prognosis and treatment response, and outputs all the results to the user in an integrated report. For the optimization of validation methods, we expect more studies to undergo external multicenter validation to assess the performance of the model, and we also recommend the clinical application of established systems to prospectively evaluate whether these systems will benefit clinical decisions. For enhancing interpretability, we anticipate greater and more transparent use of DL radiomics, and we look forward to broadening the understanding of imaging features through comprehensive radio-multi-omics analysis. Abbreviations: AI, artificial intelligence; CAD, computer aided detection/diagnosis.