Medical imaging, such as X-ray, ultrasound or magnetic resonance imaging, provides information from inside the human body and helps detect disease for diagnosis and treatment. But there is still relevant information in these images that is imperceptible to the human eye, and a new science, radiomics, comes into play.
Cover of the report on radiomics from the Roche Institute Foundation. Photo provided
Radiomics is a science that extracts, by means of computational algorithms, quantitative parameters in medical images to detect and measure those characteristics that cannot be observed by direct observation, called “radiomic characteristics”, with the aim of associating them with specific physiological states.
While the development and study of radioomics is still in its early stages, “existing evidence suggests great potential for the future application of this omics science in both research and clinical practice.”
This is highlighted in the “Anticipando Report: Radiomics” prepared by Observatory of Trends in the Medicine of the Future of the Roche Institute Foundation and coordinated by Luis Martí-Bonmatí, director of the Medical Imaging Clinical Area of the La Fe University and Polytechnic Hospital of Valencia.
“When we study a pancreatic tumor with a CT scan, in addition to the radiologist reporting its size and resectability, the radiomics studies on these images will tell us more precisely whether or not the patient is amenable to salvage surgery, if he is going to develop a short-term relapse or metastasis in the next three months”, he explains in a statement.
The applications of radiomics
Radiomics offers multiple applications in areas such as oncology, rheumatological or neurodegenerative diseases.
According to Martí-Bonmatí, “it also allows us to analyze the heterogeneity of the injuries. When we look at lesions, radiologists are very good at knowing the size, shape and structure, but we don’t recognize that inside there are groups of cells that have very different aggressiveness characteristics.”
It is also effective when treating the patient by providing information to see if the effect of the drug is as expected and can provide more reliable information than the biopsy “since it may not be sufficient when selecting the best candidates for the administration of a targeted or very specific drug because it shows only part of the tumor or because there are metastatic tumors where the metastasis no longer has the same expression as the primary tumor”, points out the expert.
Radiomics is also useful in the optimization of clinical research since it can be used, for example, in the reanalysis of images from clinical trials to detect methodological biases, such as incorrect selection of patients; or be used as a predictive tool for clinical events such as the subsequent appearance of metastasis.
Recommendations for the future
In the report, and in view of the progress in the implementation of radiomics, a series of
• Harmonize images to eliminate variability. Before comparing radiomic features in large series of patients, it is important to modify the images so that the derived radiomic data are vendor- and protocol-independent.
• Agree, standardize and protocol the selection of characteristics to analyze.
• Incorporate radiomics into the observational research to extend and improve model generation.
• Promote radiomics research and its integration with other omic sciences, which allow the study of a large number of molecules involved in the functioning of an organism.
• Encourage the use of models extraction of clinical information and computational digital models.
• Do clinical trials that allow the predictive models generated by radiomics to be prospectively validated.
• Encourage the collaboration between professionals from different areas of knowledge.
• To set alliances and collaborations between centres, institutions and companies.