The simultaneous advances of deep learning and radiomics may soon yield a single unified framework for clinical decision support that has the potential to “completely revolutionize the field of precision medicine.”
That’s the conclusion of two scholars at Johns Hopkins whose analysis of the technologies is running in the journal Expert Review of Precision Medicine and Drug Development.
PhD candidate Vishwa Parekh and radiological science professor Michael Jacobs, PhD, searched the literature for radiomics as well as deep learning and other AI-related terms, along with medical-imaging terminology such as multiparametric MRI, to examine how these research interests are being studied in the context of precision medicine.
Radiomics are quantitative measures of textures in radiological medical images at the micro-level, where aspects like interpixel relationships and spectral properties can be teased out mathematically.
Precision medicine, aka “personalized” medicine, is an approach to disease treatment and prevention that incorporates data on genetics, environment and lifestyle at the level of the individual patient.
The authors’ key finding: Deep Learning and radiomics are “creating a paradigm shift in radiology and precision medicine by developing a new area of research” into precision medicine.
“In the next five years, we will witness deep learning and radiomic methods transform medical imaging and its application to personalized medicine,” Parekh and Jacobs comment in their discussion section. “These techniques will evolve to hybrid systems based on combinations of the different networks and advanced radiomic methods for a more complete diagnosis.”
As the two technologies mature, they will be folded into clinical decision support systems and be used, the authors add, to “rapidly mine patient data spaces and radiological imaging biomarkers to move medicine towards the goal of precision medicine for patients.”
The authors parse out the details of the developments in the body of their paper, which the journal has posted in full for free.