The problem of blood oxygen calculation is calculation when the patient has some motion. The methods for increase the accuracy in calculation are proposed by many researchers. In this paper, we proposed the experiment to consider the efficiency between the method we study. In this paper, we consider two methods, first is the adaptive filter and second is minimum correlation discrete saturation transform (MCDST). From the adaptive filter, we consider two different methods for find the motion artifact, Exponentially Weighted Least Square (EWLS), Least Mean Square (LMS). The result from the experiment can summarize that the adaptive filter algorithm has the higher efficiency than the minimum correlation discrete saturation transform (MCDST). The result from the experiment also summarize that Exponentially Weighted Least Square (EWLS) has the better property for reduce motion artifact from pulse oximeter signal than Least Mean Square (LMS).