Nuclear Magnetic Resonance Spectroscopy (MRS) offers unique chemical data, complementary to the anatomic information provided by the Magnetic Resonance Imaging (MRI), that adds a new dimension to the in vivo diagnosis of neurological disease. MRS is now emerging from the laboratory and exploratory clinical studies that have shown a wide range of possible clinical applications, ranging from critically ill neonates to dementia in the elderly. The widespread clinical application of this important new diagnostic will require development of fast and reliable automated tools for interpretation of the spectra in terms of metabolite concentrations and medical diagnosis. ORINCON proposes to develop MRS interpretation software based on a neural network model employing a database of spectra of single metabolites. In Phase I of this project, ORINCON will design and implement prototype software to estimate, from simulated, one-dimensional proton spectra of the brain, the concentrations of compounds that are important for diagnosis of neurological disorders. The ultimate goal, to be realized in Phase II of this program, is to implement the MRS analysis software that will, without intervention of the operator, interpret the brain MRS data in terms of the metabolite concentrations and probability of relevant diseases. The interpretation software will be compatible with presently used spectroscopy packages and will also be portable to a personal computer (PC) platform.
Software for automated analysis of NMR spectra of the brain will be a commercially viable product that can be distributed as a part of existing MRS packages. We anticipate that the interpretation software will be implemented within SageIDL - the MRS software package developed by GE Medical Systems for nuclear magnetic resonance image (MRI) scanners. The methods developed here may also find commercial application in medical diagnostic tools based on MRS of urine, cerebrospinal fluid, plasma, bile, semen, and amniotic fluid, as well as in atomic and molecular spectroscopy.
