Table 3.
SVM-identified brain structure measures which can together distinguish among patients based on their diagnosis (i.e., ASD vs. TD, Welch's T test) or on the interaction between sex and diagnosis (two-way ANOVA).
| Diagnosis | Diagnosis × Sex | ||||||
|---|---|---|---|---|---|---|---|
| Anatomic structure | Hemisphere | Measure | t | d. f. | p | F(3, 189) | p |
| FRONTAL LOBE | |||||||
| Medial orbital sulcus | Right | Thickness | 2.388 | 189.649 | 0.018* | 3.113 | 0.027* |
| Straight gyrus | Right | CD | 2.124 | 187.787 | 0.035* | 2.351 | 0.074* |
| Left | Thickness | 2.150 | 183.505 | 0.033* | 3.409 | 0.019* | |
| Left | Volume | 2.190 | 186.758 | 0.030* | 4.618 | 0.004* | |
| Inferior frontal gyrus, orbital part | Left | CD | −2.138 | 170.856 | 0.034* | 2.655 | 0.050* |
| TEMPORAL LOBE | |||||||
| Temporal pole | Left | Curvature | 2.909 | 190.090 | 0.004* | 3.929 | 0.009* |
| Right | Thickness | −1.971 | 183.857 | 0.049* | 2.883 | 0.037* | |
| Parahippocampal gyrus | Right | Volume | −2.026 | 166.292 | 0.044* | 3.394 | 0.019* |
| Left | Volume | −2.304 | 190.792 | 0.022* | 2.254 | 0.083* | |
| Superior temporal gyrus | Right | Curvature | 2.080 | 182.553 | 0.039* | 3.052 | 0.030* |
| LIMBIC LOBE | |||||||
| Isthmus of the cingulate gyrus | Right | Volume | 2.276 | 188.168 | 0.024* | 3.606 | 0.014* |
| Left | Area | 2.543 | 186.679 | 0.012* | 3.002 | 0.032* | |
| Pericallosal sulcus | Left | Area | 2.318 | 178.371 | 0.022* | 2.320 | 0.077* |
| Right | Area | 1.962 | 189.338 | 0.049* | 1.271 | 0.285* | |
| OCCIPITAL LOBE | |||||||
| Cuneus | Right | Area | 2.223 | 186.596 | 0.027* | 3.676 | 0.013* |
| Left | Area | 1.957 | 185.521 | 0.049* | 1.976 | 0.119* | |
| Superior and transverse occipital sulci | Left | CD | 2.069 | 181.599 | 0.040* | 1.837 | 0.142* |
| Occipital pole | Right | Area | 2.454 | 190.459 | 0.015* | 4.570 | 0.004* |
| Left | Area | 2.443 | 180.449 | 0.016* | 3.536 | 0.016* | |
In the first case (diagnosis), a posteriori statistical analysis indicates that all identified measures exhibit statistically significant differences between the study (ASD) group and the control (TD) group. Welch's T-test with a significance level of α < 0.05 was used; t statistics, the associated d. f. and p-values are reported. A positive t statistic indicates that a corresponding measure's mean over the study (ASD) group is significantly larger than its mean over the control (TD) group. A negative t statistic indicates the reverse. To estimate the d. f., Welch's T-test relies on the Welch-Satterthwaite approximation, which involves the variances of the two samples. Thus, unlike in the case of Student's T-test, the d. f. of Welch's T test can differ whenever sample variances also differ. In the second case (diagnosis × sex interaction), a two-way ANOVA (factors: ASD diagnosis, sex) was used to identify the brain features which, in addition to being able to distinguish ASD from TD subjects, can also distinguish these subjects based on their sex. In this second analysis, the F statistic has 3 and 189 d. f. In both analyses, the null hypothesis is rejected at a significance level α < 0.05 subject to a multiple comparison correction, and statistically significant p-values are marked with an asterisk.