Skip to main content
. 2017 Mar 13;12(3):e0173707. doi: 10.1371/journal.pone.0173707

Fig 3. Trial-by-trial image analyses for button press and OKN.

Fig 3

A) Preprocessing for image analyses. We first sorted the order of trials according to the button-press consistency, C(t) (see Methods), over each 2 s trial. Next, we stretched the trial x time image along the trial dimension so that we can average the image across subjects who have different numbers of valid trials. Then, we upsampled the image by 1000 points along the trial dimension. Next, we smoothed the image by a boxcar kernel of 31 points along the trial dimension and averaged across subjects. This process was performed for both button press consistency and OKN. B-I) The normalized trial (stretched and upsampled) x time images for button-press consistency (B, C, F and G) and OKN (D, E, H and I), across controls (B, D, F, H) and PD patients (C, E, G, I). Note that button press and OKN velocity was flipped for the left trials (as was done for Fig 2). Button press consistency and OKN speed are denoted according to the color scale key on the right.