Development and evaluation of a deep learning framework for pelvic and sacral tumor segmentation from multi-sequence MRI: a retrospective study

Table 4 Dice and IoU of all-sequence and T2-fusion model on T2

Model performance	Dice score^c	IoU^d
All-sequence model^a	0.819	0.707
T2-fusion model^b	0.833	0.719

Note:
a, All-sequence model represents the fusions of T1-w & CET1-w & T2-w & DWI
b, T2-fusion model was constructed by the T2-w & CET1-w preprocessed image patches and took the T2-w sequence as input
c, Dice score is 2×the area of overlap divided by the total number of pixels in both images
d, IoU is the area of overlap divided by the area of union of both images

ISSN: 1470-7330