2.3. Data collection

Demographic data were collected and managed using REDCap electronic data capture tools hosted at Washington University in St. Louis (Harris et al., 2009). Each subject was seen twice, first for clinical information and to perform the “out of scanner” blink and tic suppression tasks, and then they returned for a session in the MRI suite where they performed the blink suppression and other tasks in the scanner. This report will focus exclusively on the ratings reported during the out-of-scanner tasks. In these, they were observed for a two-minute baseline period to measure the baseline blink rate, and then performed two blink suppression tasks. In each of these tasks, the subject performed 5 trials seated in front of a computer monitor, each characterized by 60 seconds of blink suppression (the monitor read “DON’T BLINK”), followed by 30 seconds during which they were allowed to blink freely (“OK TO BLINK”). During one task (“discomfort trials”), each subject continuously rated discomfort using the vertical position of a mouse-controlled pointer, with continuous feedback that interpreted the position on a 0–9 scale (Subjective Units of Distress) (Benjamin et al., 2010; Specht et al., 2013). The other task (“effort trials”) was identical except that, rather than rating discomfort, subjects were instructed to rate the effort being used to keep the eyes open. Order of tasks (discomfort, effort) was balanced across subjects. The baseline blinking condition was included to ensure that subjects exhibited relative suppression during the “don’t blink” task. Eye closures were recorded using a video camera with 30 or 60 Hz frame rates (we switched halfway through the study to an infrared camera).

We had hypothesized that effort or ability to suppress blinks might vary independently of discomfort: One person might have relatively high discomfort while holding the eyes open, but due to greater effort or greater inhibitory ability might blink no more than another person who experienced relatively little discomfort. Thus we chose to separate urge to blink into two hypothesized components, discomfort and effort, and tracked each separately. However, in our initial pilot data, mean self-rated discomfort and effort curves were fairly similar (Claudio Torres, Black, Richards, & Black, 2014), so we focused our effort on the discomfort ratings (see also section 3.5).

Tic subjects were also observed by video recording of the upper body on the screening day for 5 minutes of baseline (“OK to tic”) and 5 minutes of tic suppression.

Blinks from the first two seconds of each trial were ignored to allow for the participant to adjust to the start of each condition. Because many subjects used only a subset of the full range from 0–9, the discomfort scores were z-standardized for each subject to reduce variance across subjects, as suggested by (Brandt et al., 2016): z=(xi−x¯)/SD. The SD was computed over all ratings recorded by the subject in all blocks. A custom program written in C++ was modified from code kindly provided by Dr. Aleksandra Królak to identify whether on each frame of the video the subject’s eyes were closed. The new program compares the intensity histogram of a region-of-interest (ROI) centered on each iris in each video frame to a baseline frame in which the eyes were open. The pupils and iris widths had been marked manually on the baseline frame. The ROIs of the subsequent frames were centered automatically using a motion detection algorithm to follow head movement. The root mean squared error (RMSE) of the ROI intensity histogram comparison was computed for each frame, and frames in which the RMSE value exceeded a specified threshold were interpreted as eye closures. The results were then visually checked to see if the program happened to miss a blink due to head movement and timing was adjusted as needed. A Python script was used to eliminate suprathreshold deviations lasting less than 100 ms and to create a binary time series indicating eye closure at each video frame. The eye closure time series and discomfort ratings, which were recorded at 4 Hz, were synchronized and resampled to identical 0.25 s blocks. For each quarter-second time bin, the eye closure fraction f_i was defined as the number of eyes-closed frames within a given bin, divided by the total number of frames in that bin. Blink rate appeared similar over repeated trials, so means across trials were used in subsequent analyses. Several subjects were outliers, as defined by a blink rate more than 1.5 interquartile ranges below the first quartile or above the third quartile, so blink rates were compared between groups using the nonparametric Mann-Whitney test.

To analyze tic suppression, the videos were clipped and renamed so that the rater was blind to the condition. Each subject was recorded while sitting alone in a quiet room and tic detection included the upper body. These tic suppression trials were done separately from the blink suppression trials and the subjects did not continuously self-report their discomfort. Author KU, a movement disorders neurologist, reviewed these recordings and indicated the presence of a tic with a button press using a slightly modified version of our TicTimer software (Black, Koller, & Black, 2017). Tic frequency and the number of 10-second tic free intervals per minute were calculated from the TicTimer output and tic suppression was calculated using the difference between the two conditions as a fraction of the baseline.