Magnetic resonance imaging for the differential diagnosis of primary extra-axial brain tumors: a review of radiomic studies

Abstract

BACKGROUND: The analysis of magnetic resonance imaging data is considered the main method for the preoperative differential diagnosis of primary extra-axial tumors. However, the exact distinction of different primary extra-axial tumors based only on visual rating can be challenging. Radiomics is a quantitative method of analyzing medical image data, which allows us to understand and observe the connection between visual data and phenotypic and genotypic features of tumors. Earlier, several publications presented generalized results of research aimed at the differential diagnosis of primary extra-axial tumors based on the principles of radiomics. Fast accumulation of new clinical cases and increasing of the amounts of research on these cases demonstrate the need for their further analysis and systematization, which has led to this review. AIM: To conduct a systematic analysis of existing data on radiomics potential for the differential diagnosis of primary extra-axial tumors. MATERIALS AND METHODS: The search for publications over the past 5 years in Russian and English was conducted in PubMed/Medline, Google Scholar, and еLibrary databases. The final analysis included 19 papers on the differential diagnosis of extra-axial tumors. The included publications provided radiomic features used for the differential diagnosis of neoplasms. RESULTS: All studies demonstrated the existence of a connection between radiomic parameters (textural and histogram) and tumor type. The effectiveness of tumor differential diagnostics with radiomic models exceeded the neoplasm classification made by radiologists. The most frequently used algorithms for creating mathematical models of tumor classification based on radiomic parameters were the reference vector method, logistic regression, and random forest. CONCLUSION: The use of the radiomic concept shows promising results in the differential diagnosis of primary extra-axial tumors. Further development in this area demands the standardization of both the segmentation method and the set of features and an effective method of mathematics modeling.