Table 1 Model Performance Bias Assessment (Dice Coefficient).
From: A probabilistic deep learning approach for choroid plexus segmentation in autism spectrum disorder
Model | Group | Mean [IQR] | T-Test |
|---|---|---|---|
FreeSurfer | ASD Female | 0.26, [0.17, 0.32] | T(11, 23) = −1.48, p = 0.42 |
ASD Male | 0.3, [0.25, 0.35] | ||
CON Female | 0.25, [0.23, 0.29] | T(9, 18) = −1.71, p = 0.42 | |
CON Male | 0.29, [0.24, 0.33] | ||
All ASD | 0.29, [0.22, 0.34] | T(35, 28) = 0.4, p = 0.69 | |
All CON | 0.28, [0.23, 0.32] | ||
ASCHOPLEX | ASD Female | 0.41, [0.23, 0.57] | T(9, 19) = −0.28, p = 0.89 |
ASD Male | 0.43, [0.33, 0.61] | ||
CON Female | 0.56, [0.5, 0.67] | T(7, 14) = −1.59, p = 0.42 | |
CON Male | 0.64, [0.59, 0.7] | ||
All ASD | 0.42, [0.31, 0.59] | T(29, 22) = −4.04, p < 0.05 | |
All CON | 0.61, [0.57, 0.7] | ||
ASCHOPLEX Finetuned, Deterministic | ASD Female | 0.77, [0.71, 0.84] | T(9, 19) = −0.52, p = 0.89 |
ASD Male | 0.79, [0.76, 0.84] | ||
CON Female | 0.8, [0.75, 0.84] | T(7, 14) = −0.09, p = 0.93 | |
CON Male | 0.8, [0.78, 0.84] | ||
All ASD | 0.78, [0.75, 0.84] | T(29, 22) = −1.39, p = 0.34 | |
All CON | 0.8, [0.77, 0.84] | ||
ASCHOPLEX Finetuned, Probabilistic | ASD Female | 0.77, [0.72, 0.83] | T(9, 19) = −0.59, p = 0.89 |
ASD Male | 0.79, [0.76, 0.83] | ||
CON Female | 0.77, [0.75, 0.81] | T(7, 14) = 0.4, p = 0.89 | |
CON Male | 0.77, [0.74, 0.8] | ||
All ASD | 0.78, [0.75, 0.83] | T(29, 22) = 0.72, p = 0.63 | |
All CON | 0.77, [0.74, 0.81] |
