Meeting Banner
Abstract #3981

Statistical Strategy to Overcome Estimation Bias in CRLB Threshold Approach for LCModel Analysis of MRS

Ping-Chang Lin1, Phil Lee1, 2, Wen-Tung Wang1, William Brooks1, 3, In-Young Choi1, 2

1Hoglund Brain Imaging Center, University of Kansas Medical Center, Kansas City, KS, United States; 2Department of Molecular & Integrative Physiology, University of Kansas Medical Center, Kansas City, KS, United States; 3Department of Neurology, University of Kansas Medical Center, Kansas City, KS, United States

in vivo 1H MRS, particularly acquired at high magnetic fields, is beneficial to noninvasive study of changes in neurochemicals such as characterization of the disease progression and diagnosis of disease in early stages. LCModel analysis has been subject of intense activity in analyzing 1H MRS; however, there is no agreement in valid exhibition of metabolite concentrations in statistics. In this work we utilize weighted analysis of rat brain 1H MRS to demonstrate alternative exhibition of the concentration averages for the neurochemical profiles, in addition to investigation of the effect of threshold values on the averaged metabolite concentrations.

Keywords

accessed activity added addition additionally affected alter alternative although analyses applied approaches arithmetic basis beneficial bias black bound brain brains broadening brooks capability characterization choosing city closer collected concentration concentrations containing contrast cortex cram crucial curves dashed defined derived described desired detecting determined deviations diagnosis dimension disease disturbance disturbed domain early empirical error essential estimation evaluate even example exclude exhibit exhibited exhibits experimental experimentally expressing fields fitting fittings frequency generated graph graphs impact implement implementing in vivo included inconclusive increment indicate induced injection inputs integrative intense investigation localized measures medical metabolite metabolites molecular much neurology noise none noninvasive notable observations obvious outcome outcomes overcome panel particularly passing peaks physiology ping precisely previous prior procedure profiles progression random rats reliability represent representing respectively resultant selected selection several similarly simulated simulation smaller spectra spectral spectrum stability stages statistical statistics still stimulated strategy subject subtle summary threshold thresholds trace tung type uncertainties upon valid varying young