2.4.2. Entropy Calculation

In order to evaluate entropy based on the measured optical density data, we adopted sample entropy, an algorithm developed to robustly estimate entropy on short and noisy data [63]. The value of embedding dimension(m) was 2 and the value was known as statistically useful for systems with slow dynamics [64,65]. The value of tolerance(r), according to [65], was 0.2× SD, with SD denoting the standard deviation of the data set. A function published in the Mathworks File Exchange site was used to calculate sample entropy [66]. When we estimated a sample entropy, we used a time series of optical density data for 30 s during either the rest period or task period. As a result, the calculated total number of entropy is 22 (the number of subjects) × 3 (the number of optical density signals) × 17 (the number of channels) × 18 (the number of periods) = 20,196.