Nuclear magnetic resonance-biochemical correlation toward deep learning of theranosis and precision medicine

Abstract

Efforts have been made to employ the nuclear magnetic resonance (NMR)-biochemical correlation concept or a combination of MR imaging (MRI) and MR spectroscopy (MRS) as an established diagnostic tool for medical practice in clinical settings. Recent reviews and meta-analyses indicate the great possibility of using integrated multimodal multiparametric MRI and MRS for deep learning (DL) of soft-tissue pathophysiology, enabling improved decision-making and disease progression monitoring in precision medicine. Recent guidelines and clinical trials suggest the need for DL of the biophysical and biochemical nature of the brain, breast, prostate, liver, and heart tissue from digital spectromics analysis, along with other molecular imaging modalities. The current opinions, based on recent recommendations, available literature on evidence-based MR spectromics, clinical trials, and meta-analyses on high-resolution MRI and MRS suggest that utilizing MRI and MRS signals as theranostic biomarkers for various soft tissues can demonstrate NMR-biochemical correlation and employ MRI with MRS as adjunct real-time tools, generating robust, and fast tissue digital images with metabolic screening. The integration of DL features can aid in evaluating patient disease diagnosis and therapy within a clinical setting, considering the available medical practices and their limitations.