Cognitive tasks, anatomical MRI, and functional MRI data evaluating the construct of self-regulation

We describe the following shared data from N=103 healthy adults who completed a broad set cognitive tasks, surveys, and neuroimaging measurements to examine the construct of self-regulation. The neuroimaging acquisition involved task-based fMRI, resting fMRI, and structural MRI. Each subject completed the following ten tasks in the scanner across two 90-minute scanning sessions: attention network test (ANT), cued task switching, Columbia card task, dot pattern expectancy (DPX), delay discounting, simple and motor selective stop signal, Stroop, a towers task, and a set of survey questions. Subjects also completed resting state scans. The dataset is shared openly through the OpenNeuro project, and the dataset is formatted according to the Brain Imaging Data Structure (BIDS) standard.


24
We describe the following shared data from N=103 healthy adults who completed a 25 broad set cognitive tasks, surveys, and neuroimaging measurements to examine the construct 26 of self-regulation. The neuroimaging acquisition involved task-based fMRI, resting fMRI, and 27 structural MRI. Each subject completed the following ten tasks in the scanner across two 90-28 minute scanning sessions: attention network test (ANT), cued task switching, Columbia card 29 task, dot pattern expectancy (DPX), delay discounting, simple and motor selective stop signal, 30 Stroop, a towers task, and a set of survey questions. Subjects also completed resting state 31 scans. The dataset is shared openly through the OpenNeuro project, and the dataset is 32 formatted according to the Brain Imaging Data Structure (BIDS) standard. 33 34

Background & Summary
We report a dataset acquired as part of an effort to understand the construct of self 37 regulation, which refers to the processes or abilities that are used to serve long-term goals. Self 38 regulation has been shown to relate to a variety of real-world outcomes, including economic 39 choices, health outcomes, and academic achievement [1][2][3] . We operationalized self-regulation as 40 a large heterogeneity of constituent processes that may be interrelated, including attention, set 41 shifting, decision making, temporal discounting, response inhibition, and planning 4,5 . As a first 42 step in this project, we created a cognitive ontology of self-regulation, or an explicit specification 43 of the entities and relationships composing a domain, which is detailed in previous work 4,5 (data 44 available at http://dx.doi.org/10.17605/OSF.IO/ZK6W9). 45 The dataset detailed in the present data descriptor was acquired to build on the cognitive 46 ontology to measure the neural underpinnings of self regulation. We measured self regulation 47 with a suite of converging modalities including structural MRI and functional MRI (fMRI) during 48 tasks, rest and survey responses. The sample included 103 subjects who each completed two 49 90 minutes scans with nine putatively self-regulatory tasks during fMRI: attention network test 6 50 (ANT), cued task switching 7,8 , Columbia card task 9 , dot pattern expectancy 10 (DPX), delay 51 discounting 11 , simple stop signal 12 , motor selective stop signal 13 , Stroop 14 , and a towers task 52 15 . Additionally, subjects completed resting state fMRI and a set of 40 survey responses within 53 the scanner taken from the Brief Self-Control Scale 16 , Grit Scale 17 , Carstensen Future Time 54 Perspective 18 , UPPS+P 19 , and the Impulsiveness-Venturesomeness scale 20 . Finally, subjects 55 also completed anatomical scans. 56 This work was funded by the National Institute of Health Science of Behavior Change 57 Commons fund (UH2DA041713). To date, this neuroimaging dataset has been used in 1 58 manuscript 21 . 59 This data descriptor provides a description of this neuroimaging dataset, which is openly 60 shared on OpenNeuro (https://openneuro.org/datasets/ds004636/). The shared data are 61 organized according to the Brain Imaging Data Structure (BIDS), which is a data organization 62 structure designed to encourage FAIR sharing and reuse of neuroimaging data 22 . 63 64

66
Participants 67 Prospective participants for the study were recruited from the Stanford campus and 68 surrounding San Francisco Bay Area using several methods including paper flyers, the Stanford 69 Sona Cardiac and respiratory recordings 117 Cardiac and respiratory data was collected with a pulse oximeter and respiration belt. 118 We have included these raw files in the data directory within the subject and session matched 119 with each run. 120 121 Participant procedure 122 The study consisted of 3 parts which included 2 MRI sessions and an at-home self 123 paced online battery hosted on the Experiment Factory 23 (https://www.expfactory.org), which is 124 a framework for deploying experiments in the browser. Each MRI session consisted of a 1 hour 125 practice and setup period prior to a 1.5 hours long scanning session for a total of 2.5 hours. All 126 MRI data was obtained at Stanford's CNI. 127 Before the first MRI session, participants provided informed consent and completed a 128 demographic survey that asked for age, sex, race, and ethnicity. This information is shared in 129 the participants.tsv within the dataset shared. 130 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted September 29, 2023. ; https://doi.org/10.1101/2023.09.27.559869 doi: bioRxiv preprint During both scan sessions, subjects practiced the five tasks that they will do in that 131 session in an observation room on a laptop prior to being scanned. During this practice session, 132 a researcher read the instructions to the subject and the subject had the opportunity to ask any 133 clarifying questions. Sessions were broken up into two groups. Group 1 included the following 134 tasks: Stop Signal task, cued task switching task (two by two), Columbia Card Task (CCTHot), 135 Attention Network Task (ANT), and Ward and Allport Tower task (WATT). Group 2 included the 136 following tasks: Motor Selective Stop Signal task, Dot Pattern Expectancy task (DPX), Survey 137 Medley, Delayed-Discounting task (DDT), and the Stroop task. Tasks were counterbalanced 138 across subjects by employing 4 different task orderings. Separately for the two stop signal 139 tasks, stop signal delay (SSD) was tracked during practice using a staircase algorithm and the 140 final SSD was subsequently used as the starting SSD during the in-scanner version of the 141 tasks. For the two stop signal tasks, response mapping for each shape were randomly decided 142 for each participant during practice and were carried over into the in-scanner version. For the 143 DPX task the valid cue and probe pairing was randomly chosen and used within the practice 144 then carried over to the testing phase. For the task switching task, the color choice order and 145 magnitude choice order were randomly chosen during practice and carried over to the testing 146 phase. All other tasks did not counterbalance response mapping across subjects. 147 Scan sessions consisted of running the following sequences in order: localizer, shim, a 148 single band reference (sbref), rest fMRI, two task fMRI runs, 2nd sbref, three task fMRI runs, 149 and a fieldmap scan. T1-weighted scans were acquired during the first session, and T2-150 weighted scans were acquired during the second session. If time permitted, additional T1-151 weighted and T2-weighted scans were acquired during each session. All subjects were 152 instructed to refrain from movement during scans. Task instructions were presented again 153 during the fMRI sessions and subjects were also given instructions to respond as quickly and 154 accurately as possible. 155 Participants were compensated $20/hr for their participation in the MRI sessions and 156 $10/hr for participation in the practice section prior to scanning. Participants were also 157 compensated $100 for their completion of a ~10 hour online behavioral battery after the scans. 158 This post-scan battery is briefly described below, but these post-scan data are not shared in this 159 data release. 160 161 fMRI tasks 162 The following tasks were selected from a larger set of 36 tasks that were administered in 163 Eisenberg et al. 4,5 . We used a genetic algorithm to select a subset of tasks that best 164 reconstructed the entire behavioral results across subjects in an independent dataset (see 4 for 165 additional details). 166 All tasks made use of variable inter-trial intervals (ITIs, though see ANT) and fMRI 167 optimized trial orderings generated through the Neurodesign python package 24 . Neurodesign 168 offers a four-objective optimization that allows the user to weigh four different measures of 169 efficiency: estimation efficiency, detection power, confound efficiency, and frequency efficiency. 170 Estimation efficiency is focused on modeling the time course of the signal, specifically using a 171 finite impulse response function model, which was not our planned modeling approach. 172 Detection power focuses on the power of a model using canonical hemodynamic response 173 function-convolved regressors, which was the planned modeling approach. The other two 174 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted September 29, 2023. ; https://doi.org/10.1101/2023.09.27.559869 doi: bioRxiv preprint efficiencies focus on the predictability of trials due to trial order (confound efficiency) and 175 whether the desired trial type frequency is achieved (frequency efficiency). We chose values of 176 0, .1, .5 and .4 for estimation efficiency, detection power, confound efficiency and frequency 177 efficiency, respectively, given our primary goals were to minimize predictability while retaining 178 the desired trial frequencies with a secondary goal of maximizing detection power. Four sets of 179 ITI's and trial orderings were then generated by rotating both the ITI's and trial ordering of the 180 originally chosen Neurodesign set, each starting at a different time point within the chosen 181 design of each task to match the four different task orders used across subjects for 182 counterbalancing purposes. Each task used Neurodesign to power for the specific contrasts 183 outlined below. 184 185 Attention network test (ANT) 186 The ANT 6 is designed to engage networks involved in three putative attentional The task consisted of 128 total trials separated into 2 blocks. Each trial consisted of a 198 cue (100ms), fixation cross (400ms), probe (1700ms), and ITI (400ms) where a fixation is 199 shown on the screen. The onset timing of the probe is specified in the event files included in the 200 dataset. Responses were accepted during the probe and ITI period for each trial. Due to a 201 coding error, this was the one task that did not include a variable inter-trial interval. 202 Conditions included 16 possible combinations of cue (double, spatial) * probe location 203 (up, down) * probe direction (left, right) * condition (congruent, incongruent), each equally likely 204 to occur. Neurodesign 24 was used to generate trial orders equally powering for cue, probe 205 location, probe direction, and condition. 206 207 Cued task-switching task (two by two) 208 We used a modified cued task switching task that paired 2 different tasks with 2 different 209 cues 7,8 . Each trial begins with a fixation followed by a central cue instructing the subject on 210 whether to make a magnitude or a color judgment to the subsequent probe number. The cue is 211 either "high-low", "magnitude", "orange-blue", or "color", with the former two cues indicating one 212 task (judge whether the subsequent probe number is greater than or less than 5), and the latter 213 two cues indicating the other task (to judge whether the subsequent probe number is orange or 214 blue). The cue was presented 100ms or 900ms before the target onset (with equal probabilities) 215 and stayed on the screen throughout the probe phase. The probe was a single colored letter 216 between 1-9 excluding 5. Responses are made with the index and middle finger of the 217 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted September 29, 2023. ; https://doi.org/10.1101/2023.09.27.559869 doi: bioRxiv preprint participant's right hand and the magnitude and color choice key mappings were randomly 218 chosen during each subject's practice session and used in the scanner. 219 The task consisted of 240 trials comprising 3 blocks of 80 trials. Each trial consisted of a 220 fixation (500ms), cue (100, 900ms), probe (1000ms), and a blank-screen ITI (1000ms + variable  221 ITI, see below for details). The onset timing of the probe is specified in the event files included in 222 the dataset. Responses were accepted in the probe and ITI periods. 223 Conditions included cue switch (25% of trials), cue stay (25% of trials), and task switch 224 (50% of trials). Cue switch trials are when the current trial has a different cue than the 225 immediately preceding trial, but both indicate the same task (e.g., current trial cue is 226 "magnitude" and previous cue was "high-low"). Cue stay trials are when the current trial has the 227 exact same cue as the immediately preceding trial (e..g, current trial cue is "magnitude" and the 228 immediately preceding trial cue was also "magnitude"). Task switch trials are when the current 229 trial has a cue that indicates a different task than the task that from the immediately preceding 230 trial (e.g., current trial cue is "magnitude" and the previous cue was "color"). Neurodesign 24 was 231 used to generate both trial orders and the variable portion of the ITI sampling from an 232 exponential distribution using a min of 0, mean of .25, and a max of 6 seconds, powering for 233 task switch -(cue switch + cue stay) and cue switch -cue stay contrasts. 234 235 Columbia card task (CCTHot) 236 The CCTHot task 9 has subjects play a card game with the goal of collecting as many 237 points by flipping over gain cards and avoiding flipping over loss cards. Each round presents a 238 set of 6, 8, 9, 10, 12, 15, or 16 cards face down with the following information about the cards: 239 The number of loss cards (1, 2, 3, or 5 possible), the amount they will lose if they turn over a 240 loss card (ranging from -100 to -5), and the amount they earn with a gain card (ranging from 1 to 241 30). The subject can choose whether to flip over a card at random by pressing the index finger 242 key or "cash out" and move to the next trial by pressing the middle finger key. A subject may flip 243 over zero or as many cards as they choose, but the trial ends when they flip over a loss card. 244 After each choice is made, the subject is shown which card they flipped over. For each round, 245 they will be awarded the net number of points from the gain and loss cards they flipped over. 246 The task consists of 87 rounds split into 2 blocks of 44 and 43 rounds. Each trial is 247 subject paced. After each round there is a feedback phase of 2250ms with an added blank-248 screen variable ITI (minimum of 0, mean of .2s, max of 1s). The scan was stopped after 12 249 minutes, so given the variable trial duration this resulted in a different number of trials per 250 subject (M of completed trials across participants = 81.02, SD = 12.75, range = 32 -87). 251 Neurodesign was used to generate both trial orders and ITI's 24 . 252 253 Dot pattern expectancy task (DPX) 254 In the DPX task 10 , subjects see a series of cue-probe stimulus pairs and make a 255 speeded response after the probe. Each stimulus is made with a pattern of dots. There are 6 256 possible cues, and 6 possible probes. One cue is the "target cue" (i.e. 'A') and one probe is the 257 "target probe" (i.e. 'X'). Subjects must press one key if the target cue is followed by the target 258 probe (i.e. 'AX'), and another key for any other cue-probe pairing. The target cue and target 259 probe was randomly chosen during practice and input for the test phase. Responses are made 260 with the right hand index finger for AX pairs and middle finger for any other pair. 261 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made cue preceding a non-target probe, BX trials are a non-target cue preceding a target probe, and 268 BY trials are a non-target cue preceding a non-target probe. Neurodesign 24 was used to 269 generate both trial orders and the variable portion of the ITI's. ITI were sampled from a 270 truncated exponential distribution with a minimum of 0, a mean of .4s, and a max of 6s. We 271 powered for AY-BY and BX-BY contrasts. 272 273 Delay-discounting task (DDT) 274 In the delayed-discounting task 11 , subjects are presented with a dollar amount and a 275 number of days. They are asked to choose between receiving $20 today or the larger amount of 276 money at the presented delay in days. The amount of money ($22 -$85) and number of days 277 (19 days -180 days) varies across trials. Subjects were told that their performance will impact 278 the bonus they receive. Subjects made a right hand middle finger button press for the "smaller 279 sooner" reward of $20 and an index finger button press for the "larger later" presented reward. 280 The task consists of 2 blocks of 60 trials for a total of 120 trials. Each Trial consists of a probe 281 (4000ms) and a blank screen ITI (500ms + variable ITI). Responses were accepted to both the 282 probe and the ITI. 283 Neurodesign 24 was used to generate both trial orders and ITI's (minimum of 0, mean of 284 .5s, max of 2.5s). 285

Stop-signal task 286
In the stop-signal task 12 , subjects are presented with one of 4 "go" shapes (moon, oval, 287 trapezoid, rectangle) and are instructed to respond to two shapes with one key press and the 288 other two shapes with another key press. On a subset of trials, a star appears around the shape 289 after a delay (the stop-signal delay, SSD). This indicates to the participant that they should try 290 not to make any response on that trial. If a subject is unable to withhold their response on a 291 stop-signal trial this is categorized as a stop failure, and if the subject makes no response then 292 this is categorized as a stop success. During both the practice and main task, the SSD was 293 tracked using a "1 up 1 down" staircase algorithm 25 in which SSD increased by 50ms after each 294 successful stop signal trial and decreased by 50ms after each failed stop signal trial. The final 295 SSD in practice was used as the starting SSD during the in-scanner version of the tasks. 296 There are a total of 125 trials, grouped into 3 blocks (41 trials in block 1 and 42 trials in 297 each of blocks 2 and 3), with 60% of the trials being go trials and 40% being stop trials. The 298 stop signal duration was 500ms and the initial SSD during practice was 250ms. Each trial 299 consisted of a go stim (850ms) and an inter-trial interval that included a central fixation cross 300 (1400ms + variable ITI). Neurodesign 24 was used to generate both trial orders and the variable 301 portion of the ITI, which was sampled from an exponential distribution using a min of 0, mean of 302 .225, and a max of 6 seconds, powering for stop -go contrast. Go responses were made with 303 the index and middle finger of the participant's right hand and the choice key mapping was 304 randomly chosen during each subject's practice session and used in the scanner. 305 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made Motor selective stop-signal task 307 The motor selective stop-signal task 13 is identical to the previously described stop-signal 308 task, with the following exceptions. Instead of instructing subjects to stop whenever a stop 309 signal occurs, subjects are instructed to only stop if a stop signal occurs and they were going to 310 make one of their two responses (the "critical" response), but not if a stop signal occurs and 311 they were going to make the other response (the "non-critical" response). SSD was adjusted 312 only on critical stop trials, and the SSD on noncritical stop trials was yoked to the SSD on the 313 critical stop trials. Responses are made with the index and middle finger of the participant's right 314 hand and the choice key mapping is randomly chosen during each subject's practice session 315 and used in the scanner. Critical stop trial stimuli are also randomly chosen during the subject's 316 practice session and used in the scanner. 317 The task consists of 5 blocks of 50 trials for a total of 250 trials. 60% of trials were go 318 trials, 20% critical stop trials, and 20% noncritical stop trials. We powered for the following 319 contrasts: Noncritical stop -noncritical no-signal and critical stop -noncritical stop. 320 321 Stroop task 322 In the Stroop task 14 , subjects are presented with a color word (e.g., "red") written in ink 323 that either matches the word (congruent, e.g., "red" in red) or does not (incongruent, e.g., "red" 324 in blue). Subjects are instructed to quickly and accurately respond via keypress to the ink color 325 of the word is. The color of the word and the written word could both be red, blue or green. The 326 participants were asked to respond with index, middle, and ring finger for each of these colors. 327 The task consists of 2 blocks of 48 trials for a total of 96 trials. Each trial consists of a 328 probe (1500ms) and an inter-trial interval that includes a central fixation cross (500ms + variable 329 ITI). There were 48 congruent and 48 incongruent trials. Neurodesign 24 was used to generate 330 both trial orders and the variable portion of the ITI sampling from an exponential distribution 331 using a min of 0, mean of .2, and a max of 6 seconds, powering for incongruent -congruent. 332 333 Ward and Allport tower task (WATT) 334 In the WATT 15 subjects are presented with stacked balls in a specific initial configuration 335 on three pegs, and they are asked to move the stacked balls to a different target configuration 336 using the least amount of moves. 337 The task consists of 3 blocks of 16 rounds for a total of 48 rounds. These moves are not 338 timed and the round ends when the subject moves the ball to match the target configuration 339 shown on the screen. All trials were designed to be completed in 3 steps, where a step is 340 defined by taking a ball off the board then placing it on a different peg. After the participant 341 correctly solves each round there is a feedback phase that consists of feedback shown for 342 1000ms and a blank screen (1000ms + variable ITI). Some subjects did not complete the full 48 343 rounds, as there was a max scan time of 10 minutes and the task was ended after 10 minutes where the target configuration had one ball on one peg and two balls on another peg (Partially 348 Ambiguous). Trials included conditions in which a ball had to be moved out of the way to reach 349 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made Neurodesign 24 was used to generate both trial orders and the variable portion of the ITI 352 sampling from an exponential distribution using a min of 0, mean of .9, and a max of 6 seconds, 353 powering for conditions shown above. 354 355 Survey medley 356 In addition to the above 9 traditional cognitive tasks, subjects also completed 40 survey 357 questions that were answered within the scanner. We scanned the following three complete 358 surveys: the Grit Scale 17 (8 questions), Brief Self-Control Scale 16 (13 questions), and the 359 Carstensen Future Time Perspective 18 (10 questions, Carstensen). We also selected a subset 360 of items from the Impulsiveness-Venturesomeness 20 (3 questions) and UPPS+P impulsivity 19 (6 361 questions) questionnaires. 362 Each trial presented the survey question for 8500ms and was followed by a blank screen 363 for 500ms plus a variable ITI (minimum of 0, mean of 1s, max of 5s). Neurodesign (Durnez et  364 al., 2018) was used to generate both trial orders and ITI's. 365 366 Post-scan battery 367 Upon completion of the scanning sessions participants were also given an at-home 368 behavioral battery to do on their own time that consists of 63 tasks and surveys, matching the 369 full battery from Eisenberg et al. 5 . These data are not included within this data release and are 370 not described in detail in this data descriptor. 371 372 The participants.tsv file contains age and sex of each participant, and quality_control.csv 391 contains notes on suggested exclusions based on behavior and neuroimaging data quality 392 assurance (described below). 393 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made All collected data are included in the current data release. We have completed quality 422 assurance of the data that is described in detail in the immediately following sections, and we 423 recommend that certain subjects and scans be excluded, but we recognize that exclusion 424 decisions can depend on the goals of each project, so we have decided to include all available 425 data in the data release so that data users can make exclusion decisions that align with their 426 use cases. 427 428

Data Records
Behavioral Quality Control 429 We completed the following behavioral quality control, and we have provided 430 annotations of these quality control recommendations in suggested_exclusions.csv on 431 OpenNeuro. 432 Falling outside the normal range on any of the following criteria was sufficient for that 433 subject's data to be recommended for exclusion. In the stop-signal tasks, stop success rate 434 being below .25 or above .75, in line with consensus recommendations 29 . In all tasks, having an 435 omission rate of over .5. Subjects missing more than half the tasks had their entire dataset (all 436 tasks) recommended for exclusion. We ran a binomial test on noncritical signal trials in motor 437 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted September 29, 2023. ; https://doi.org/10.1101/2023.09.27.559869 doi: bioRxiv preprint selective stopping to ensure that subjects were making responses on more than half of trials. If 438 they failed this binomial test, they were recommended for exclusion as it indicated they were 439 treating the task as a simple stop signal task. Last, we looked to see if any subjects omitted a 440 majority of their responses at the beginning or end of any scans, but this did not identify any 441 additional subjects. 442 We also evaluated descriptive statistics for each subject on each task, including their 443 accuracy, RT, and other key dependent variables (e.g., range of SSDs for the stop-signal task), 444 for aberrant behavior. This resulted in the recommended exclusion of 11 tasks across 6 445 subjects, mostly for accuracy that was at or below chance in at least one condition. Details of 446 each of these 11 decisions can be found in the subjective_rating_descriptions.json within the 447 BIDS directory. 448 As a result of this quality assurance effort, we recommend that 12 subjects be 449 completely excluded because performance in at least 5 of their 9 speeded tasks (i.e., all but the 450 survey medley) were deemed unsatisfactory. Additionally, we identified 22 subjects who had 451 satisfactory behavior on at least 5 of their 9 tasks but we recommend excluding their data for 1 452 to 4 of their tasks based upon performance.  Figure 1 for T1w quality assurance metrics and 471 Table 1 for an explanation of the quality control metrics. 472 473 Functional BOLD scans 474 The following quality control metrics were computed using MRIQC for each fMRI scan. 475 MRIQC reports were visually inspected for visible artifacts such as aliasing, distortion, and 476 ghosting. Specific scans with MRIQC reports that did not pass inspection are noted in 477 quality_control.csv within the dataset. See Figure 2 for distributions of quality control measures 478 across all functional scans (task and rest) and Table 1 for an explanation of the quality control 479 metrics. Quality control measures were highly similar across functional scans, so we include all 480 task and rest data aggregated in Figure 2 (gsr_y (M = 0.038, SD = 0.001, range = 0.037 -481 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted September 29, 2023. ; https://doi.org/10.1101/2023.09.27.559869 doi: bioRxiv preprint   Tables   539  540  Table 1. Description of QC measures of T1-weighted anatomical data 541

Image Quality Metric (Label) Description
Artifact detection (qi_1) Detects artifacts from motion, blurring or ghosting 30 . Lower values are better.
Contrast-to-noise ratio (cnr) Measure of contrast between white and gray matter 31 .
Entropy focus criterion (efc) Proxy for ghosting or blurriness caused by head motion 32 . Lower values are better.
14 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted September 29, 2023. ; https://doi.org/10.1101/2023.09.27.559869 doi: bioRxiv preprint