Table 2. Comparison of manual parcellations across optimized 3T and 7T MRI.
| Structure | Dice scores | Conjunct Volumes | Anatomical Distance | QSM CNRs | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| One-tailed | Two-tailed | Two-tailed | One-tailed | |||||||||
| o3T | 7T | BF | o3T | 7T | BF | o3T | 7T | BF | o3T | 7T | BF | |
| GPe | 0.77 (0.04) | 0.81 (0.05) | M1 BF10 = 4.11 (Mc BF10 = 34.26) | 1047 (127.46) | 999.52 (136.57) | BF01 = 0.58 | 37.35 (1.67) | 36.72 (1.22) | BF01 = 0.86 | 1.22 (0.16) | 1.09 (0.14) | M2 BF10 = 5.43 (Mc BF10 = 46.65) |
| GPi | 0.71 (0.06) | 0.76 (0.05) | M1 BF10 = 4.22 (Mc BF10 = 35.13) | 420.25 (85.71) | 415.09 (95.59) | BF01 = 0.31 | 31.35 (2.00) | 31.95 (1.40) | BF01 = 0.89 | 0.81 (0.16) | 0.75 (0.15) | M2 BF01 = 0.66 (Mc BF10 = 3.62) |
| RN | 0.80 (0.06) | 0.85 (0.03) | M1 BF10 = 7.20 (Mc BF10 = 63.73) | 233.40 (45.44) | 227.44 (32.59) | BF01 = 0.47 | 8.73 (0.41) | 8.71 (0.43) | BF01 = 0.32 | 1.92 (0.51) | 2.05 (0.42) | M1 BF10 = 1.64 (Mc BF10 = 11.76) |
| SN | 0.78 (0.07) | 0.81 (0.03) | M1 BF10 = 1.06 (Mc BF10 = 6.89 | 434.45 (75.49) | 401.39 (91.94) | BF01 = 0.35 | 16.49 (0.83) | 16.45 (0.48) | BF01 = 0.31 | 1.62 (0.17) | 1.78 (0.32) | M1 BF10 = 1.20 (Mc BF10 = 7.87) |
| STN | 0.69 (0.06) | 0.74 (0.06) | M1 BF10 = 1.85 (Mc BF10 = 13.59) | 88.90 (15.12) | 87.35 (23.20) | BF01 = 0.51 | 17.93 (1.34) | 17.71 (1.47) | BF01 = 0.34 | 1.04 (0.26) | 1.33 (0.32) | M1 BF10 = 61.75 (Mc BF10 = 629.61) |
Dice scores, conjunct volumes, Anatomical distances, and QSM CNRs are averaged across hemisphere and presented as mean values and standard deviations for o3T and 7T MRI contrasts. BF10 indicates evidence for the alternative, and BF01 refers to evidence for the null hypothesis. Dice scores and QSM CNRs were compared using a Bayesian one-tailed paired samples t-test, where BF10 assumes that in both cases 7T is higher than o3T (model 1 (M1)), and BF01 assumes either no difference or a decrease in 7T compared to o3T (model 2 (M2)). For one-tailed paired samples t-tests, only the BF for the winning model is noted, and the likelihood ratio is calculated between the winning and losing models and is noted by ‘Mc’ (standing for model comparisons). Conjunct volumes and Anatomical distances were compared between 3T and 7T with two-tailed paired-samples t-tests. o3T = optimized 3T, QSM = quantitative susceptibility mapping, CNR = contrast to noise ratio, GPe = globus pallidus externa, GPi = globus pallidus interna, RN = red nucleus, SN = substantia nigra, STN = subthalamic nucleus.