Radiomics: the process and applications in tumor research

Abstract

Radiomics enables extraction of innumerable quantitative features from medical images with high-throughput computing for diagnosis and prediction. The practice of radiomics involves image acquisition, identifying and segmenting the volumes of interest, extracting and analyzing of quantitative features, and classification or prediction model development. Compared with traditional visual interpretation of medical images, the deep mining of medical images by computer technology from radiomics makes feature uptake more efficient, relatively objective and rich in feature types. Whereas, radiomic analysis requires high image quality and consistent scan parameters. The features extracted are confined to the segmented area. Radiomics is promising in tumor screening, early diagnosis, accurate grading and staging, treatment and prognosis, molecular characteristics and so on. Combined with traditional visual interpretation of medical images, radiomics is helpful in tumor diagnosis and prediction.