Probabilistic Averaging: An Instinctive Method of Averaging According to Data Confidence Applied to Cardiac 31P MRS

Lowri E. Cochlin¹, Kieran Clarke¹

This work demonstrates an intuitive approach to weighting data according to precision achieved during quantification. Cardiac 31P-MRS is used as an example of data whose accurate fitting is often challenged by low SNR. The 1:1:1 ratio of ::-ATP peaks (excited uniformly with optimized RF-pulse, NAD and T1 corrected) constitutes three measures of the same molecule. Probability density functions generated for each integrated peak therefore combine as a weighted average representing a maximum likelihood estimate of the true value. Data from 11 healthy volunteers processed probabilistically demonstrated reduced variance in every PCr/ATP, and reduced inter-subject PCr/ATP spread, compared to standard weighting.