,Shaw et al. 10.1073/pnas.0707741104. |
Movie 1. Dorsal view of the age of attaining peak cortical thickness for the ADHD and typically developing control groups. The darker colors (dark blue in the ADHD and dark purple in the typically developing control images) indicate regions where a quadratic model was not appropriate, and thus a peak age could not be calculated, or that the peak age was estimated to lie outside the age range covered. The movie is stopped at age 13 because there was no further peaks detectable in this view.
Movie 2. Right lateral view of the age of attaining peak cortical thickness. Beyond the end of the age range is shown that the anterior temporal regions continued slowly to attain peak thickness at a similar rate in both groups.
Fig. 5. Regions where the ADHD group had delayed cortical maturation, as indicated by an older age of attaining peak cortical thickness. The regression equation included the degree of motion artifact. As can be seen through comparison with Fig. 2, there was little change in the pattern of cortical delay.
Fig. 6. Regions where the ADHD group had earlier cortical maturation, as indicated by an older age of attaining peak cortical thickness. The regression equation included the degree of motion artifact. As be seen through comparison with Fig. 4, there was little change in the pattern of earlier cortical maturation in the
SI Text
Quality control of the scans was implemented at several stages. At acquisition, foam padding was placed around the head to steady the head and minimize disturbance from scanner noise, and subjects were scanned late in the evening to promote natural sleep. Upon completion of the MRI sequence, the image was immediately rendered and quality of the image assessed (by J.G.). If excessive motion artifact was detected, the subjects were given the option of staying in the scanner for another attempt. This occurred in »20% of all children (under 12 years) and 15% of adolescents (13-20 years). The rates were similar in healthy controls and children with ADHD, although detailed records of exact numbers were not kept. About half of the time, the repeat scan was successful, for an overall successful scan yield of »90%.
After processing, each scan was assigned a motion artifact rating of none, mild, moderate, or severe by trained raters as described in detail in (1). Scans rated as "none" had little or no detectable motion artifact, scans rated as "mild" had enough detectable motion to results in subtle concentric bands to appear in the automated classification; scans classified as "moderate" had significant banding, and scans rated as "severe" had a degree of distortion so great that segmentation of the tissue in gray and white matter was difficult. The ratings were performed by two raters (J.B. and L.C.) with high intrarater reliability (a= 0.96) and interrater reliability (a= 0.95). There are no fully automated techniques that can adequately estimate the degree of motion artifact in our T1-weigthed scans.
For the current study, scans rated as having either moderate or severe motion artifact were excluded. The overall rate of this degree of motion artifact was 7.7% (34 of 442 scans) in the ADHD group. The rate of severe/moderate motion artifact in the entire typically developing children (from which age-, sex-, and IQ-matched subjects were drawn) was 4.4% (47 of 949 scans). Rates for each age group are given in SI Table 3.
In a binomial logistic regression with presence or absence of severe/moderate motion artifact as the dependent variable, there was a significant effect of age (b = 0.11, Wald = 10.1, P = 0.002), but not diagnosis (b = -0.45, Wald = 0.43, P = 0.51). There was also no significant interaction between age and diagnosis (b = 0.008, Wald = 0.015, P = 0.9); thus younger children, regardless of diagnosis, were more likely to have scans that were excluded because of severe or moderate motion artifact.
Examining the scans retained for final analysis, the overall rate of mild motion artifact was higher in those with ADHD (39 of 408 scans, 9.6%) compared with typically developing controls (16 of 416 scans, 3.8%). As can be seen from SI Table 4, there was greater motion artifact in younger subjects. In a logistic regression, with the presence/absence of mild motion artifact as the dependent variable, there was a significant effect of age (b = -0.31, Wald = 12.9, P < 0.001), but not diagnosis (b = 1.1, Wald = 1.1, P = 0.29), and there was no significant interaction between age and diagnosis (b = -0.004, Wald = 0.001, P = 0.97).
The degree of motion artifact was entered into the regression equation, as follows: The kth cortical thickness of the ith individual in the jth family was modeled as
,
where dij are nested random effects modeling within-person and within family dependence; the intercept and b terms are fixed effects, eijk represents the residual error, and motioni is the degree of motion artifact (mild or none) in the ith individual. As can be seen from SI Figs. 5 and 6 , this made little change in the central finding of delay in our index of cortical maturation in the ADHD group. There were some minor differences, reflecting a change in the degrees of freedom and the uneven age distribution of the presence of motion artifact.
1. Blumenthal JD, Zijdenbos A, Molloy E, Giedd JN (2002) NeuroImage 16:89-92.
Table 3. The number of scans at each age that were of acceptable quality (i.e. had no or mild motion artifact) and those that were excluded because of severe or moderate motion artifact
Age | ADHD | Typically developing | ||
Passed | Failed | Passed | Failed | |
<5 | 13 | 2 | 24 | 4 |
6 | 18 | 0 | 24 | 9 |
7 | 37 | 8 | 28 | 6 |
8 | 44 | 4 | 24 | 4 |
9 | 45 | 5 | 31 | 1 |
10 | 40 | 2 | 39 | 1 |
11 | 33 | 4 | 38 | 3 |
12 | 33 | 3 | 38 | 4 |
13 | 24 | 3 | 33 | 3 |
14 | 29 | 1 | 30 | 2 |
15 | 22 | 0 | 30 | 1 |
16 | 17 | 0 | 25 | 2 |
17 | 19 | 1 | 15 | 3 |
18 | 13 | 0 | 17 | 2 |
19 | 8 | 0 | 7 | 1 |
20 and over | 13 | 1 | 13 | 1 |
Data for the entire typically developing cohort is given; the subjects in the current study were a subset of this larger cohort, matched with the ADHD group on the basis of age, gender and IQ.
Table 4. The numbers of scans at each age that had either no detectable motion artifact or mild motion artifact
Age | ADHD | Typically developing | ||
No motion artifact | Mild | No motion artifact | Mild | |
<5 | 13 | 0 | 21 | 3 |
6 | 13 | 5 | 23 | 1 |
7 | 25 | 12 | 25 | 3 |
8 | 40 | 4 | 21 | 3 |
9 | 36 | 9 | 30 | 1 |
10 | 37 | 3 | 36 | 3 |
11 | 29 | 4 | 36 | 2 |
12 | 33 | 0 | 38 | 0 |
13 | 22 | 2 | 33 | 0 |
14 | 29 | 0 | 30 | 0 |
15 | 22 | 0 | 30 | 0 |
16 | 17 | 0 | 25 | 0 |
17 | 19 | 0 | 15 | 0 |
18 | 13 | 0 | 17 | 0 |
19 | 8 | 0 | 7 | 0 |
20 and over | 13 | 0 | 13 | 0 |