| Popis: |
Background and Objective: Specific treatment for each patient based on their clinical data is one of the medical prospects of the future. Using data mining and machine learning techniques based on computer science in extracting the quantitative features of an image to improve the process of diagnosis, prognosis, prediction and response to cancer treatment is known as radiomics. This article examines the workflow, findings, challenges ahead, and the role of radiomics in precision medicine and individual therapy. Methods: In this review article, we searched well-known indexes such as ISC, web of science, Google Scholar, Scopus, PubMed without time limit and based on the keywords radiomics, radiotherapy, cancer and quantitative imaging and relevant articles were collected. Findings: Radiomics is a combination of everyday computer-aided diagnosis, machine learning methods, deep learning and human skills that can be used for quantitative description of the phenotypes of cancerous tumors. Image collection and processing, tumor segmentation, extraction of features, processing and modeling are some of the basic steps of the process of radiomics. Computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET) and ultrasound (US) methods are among the used images. Conclusion: According to the results of this study, the prerequisite for the clinical implementation of radiomics is the elimination of deficiencies such as the dependence of the features on the imaging parameters, and the unrepeatability of the features. Therefore, a comprehensive approach should be adopted, stable and reproducible patterns should be developed to accept radiomics as a clinical prognostic tool. |
