Skip to main content
. 2016 May 3;11:909–917. doi: 10.1007/s11548-016-1376-5

Fig. 3.

Fig. 3

Effect of unequal test and training noise. a Low SNR for training, varying testing noise. b High SNR for training, varying testing noise. Depicted is the median absolute error in oxygenation estimation and corresponding interquartile ranges as a function of the amount of noise for baseline and the proposed random forest-based method with 15,000 training samples