SGPNet: A Three-Dimensional Multitask Residual Framework for Segmentation and IDH Genotype Prediction of Gliomas

Figure 3

Illustration of the performance of segmentation task of the SGPNet: (a–d) the 3D T2-Flair, T1Gd, T1, and T2 volumes of a glioma patient in the axial, coronal, and sagittal slices after data preprocessing; (e) the ground-truth segmentation labels of the lesion; (f) the segmentation results are predicted by the SGPNet.

(a)

(b)

(c)

(d)

(e)

(f)