Computational Molecular Magnetic Resonance Fingerprinting in Digital Health

In this study, we present mathematical and computational concepts (generally applicable to the analysis of biological and non-biological systems) at the molecular level with special application to digital health. Digital health design can be likened to complex systems and are often dominated by large numbers of processes. For example, when deviations occur in normal human system, human disease conditions are produced. Understanding these processes is important not just in unravelling the causes of diseases, but also the manner of disease propagation and the best plan for treatment. The inadequate understanding of the molecular dynamics of diseases is one reason why many diseases remain incurable and become life-threatening. Computational molecular magnetic resonance imaging now provides new ways of visualizing molecular dynamics and the management of human diseases. Currently, the available imaging equipment no longer matches the increasing number of patients requiring healthcare. The few available imaging machines are costly to maintain while financial difficulties are making acquisition of new ones near impossible. Obviously, experimental methods alone are no longer enough for efficient diagnosis, therapy and prognosis. These challenges may now require the development of appropriate mathematical tools and sophisticated computer simulations based on the Bloch NMR flow equations to complement laboratory and clinical observations. Such mathematical techniques have the potential to provide insight into the imaging of molecular interactions through the analysis of relaxation processes as observed in computational molecular magnetic resonance. The aim of this study is to develop computational magnetic resonance fingerprinting (MRF) methods based on Bloch NMR flow equations and artificial intelligence for differentiating intra-axial brain tumors. We calculate simultaneous measurement of multiple tissue properties in terms of T1 and T2 relaxation times for differentiating intra-axial brain tumors. Machine learning models for contralateral white matter, peritumoral white matter and solid tumors dataset are developed with graphic user interface (GUI) and shiny package for model implementation.