A Low Channel Number Sensing Approach for an Ethnic Specific Labour Immanency Prediction using Bio-Electromagnetism

摘要

Magnetomyography (MMG) has shown promise in being able to non-invasively acquire magnetic field signals associated with uterine contractions, which in turn can be used to forecast labour immanency and carries clinical appeal. Despite these results, MMG has yet to be adopted widely in clinical settings, largely due to the cost associated with the instrumentation. As part of strides towards scaling down the potential cost of the instrumentation, a heuristic-based electrode selection scheme was used to construct a parsimonious electrode channel representation which minimizes the number of channels whilst maximizing classification accuracy. This exercise was conducted for subject data from Black and Caucasian ethnicities where the results showed that an optimal representation of electrodes could be scaled down from 148 to either 10 or 20, depending on the ethnic group. Further work in this area would now involve the mapping of the optimal electrode channels to various anatomical regions in the uterus in order to further investigate the signal propagation dynamics within the uterus.