Radiomics Science, A Horizon for Early Diagnosis of Brain Metastasis Caused by Non-small Cell Lung Cancer: A Review of the Literature


1 Tabriz University of Medical Sciences*

2 Department of Neuro Science and Addiction Studies School of Advanced Technologies in Medicine Tehran University of Medical Sciences, Tehran, Iran

3 Radiology Department, ParaMedical Faculty, Tabriz University of Medical Sciences



Non-small cell lung cancer (NSCLC) is considered the second most commonly diagnosed cancer, accounting for almost 30% of adult deaths. Brain is one of the most frequent regions for NSCLC metastasis. Magnetic resonance imaging (MRI) is a common imaging modality for diagnosing NSCLC. In addition, computed tomography (CT) and positron emission tomography-CT (18-FDG-PET-CT) have complementary aid. Nevertheless, these methods are highly invasive and fail to reduce the risk of brain metastasis (BM) in NSCLC patients. Introducing noninvasive methods for predicting and monitoring NSCLC patients with BM seems to be helpful. Radiomics is the science of extracting quantitative data from medical images using mathematical algorithms and finding correlations with biological or clinical outcomes via machine learning. This study aimed to investigate whether radiomics is a valuable and predictive method for clinically managing NSCLC patients with BM.                                                                                                                                                    Methods:
The keywords of “Radiomics”, “NSCLC”, “Brain metastasis”, “MRI”, “CT”, “18-FDG-PET- CT”, and “Machine learning” were entered into scientific databases of Google scholar, Scopus, PubMed, and Elsevier. About ten fully relevant papers were extracted and reviewed.
CT was the most used modality for the analysis of NSCLC patients with BM followed by MRI and PET. All papers indicated that textural- based radiomics features (especially gray level co-occurrence matrix group) were highly predictive of BM. Also, age and tumor location were the two important clinical factors for the prediction of BM in NSCLCs. Machine learning- based models showed an area under the ROC curve (AUC) of about [.71-0.81], [0.62-0.83], and [0.62-0.91] for clinical, radiomics, and combined (clinical and radiomics) models, respectively.
It seems that radiomics-based quantitative analysis in combination with clinical factors can significantly help in the prediction of BM and better management of NSCLC patients.
Keywords: “NSCLC”, “Brain metastasis”, “Radiomics”