Table 2.

Summary of the four methods and their parameters used for evaluating Ischemic Stroke lesion segmentation

Method	2D Multimodal U-Net	2D U-Net (DWI)	3D Multimodal U-Net	3D U-Net (DWI)
Input Image Size	512 × 512 × 2	512 × 512 × 1	512 × 512 × 32 × 2	512 × 512 × 32 × 1
Starting Filter Number	64	64	32	32
Kernel Size	5 × 5	5 × 5	5 × 5 × 5	5 × 5 × 5
CNN Activation Type	ReLU	ReLU	ReLU	ReLU
Activation Layer	Sigmoid	Sigmoid	Sigmoid	Sigmoid
Layer Depth	2 Down & 2 Up	2 Down & 2 Up	2 Down & 2 Up	2 Down & 2 Up
Optimizer / learning rate	Adam / 1e-5	Adam / 1e-5	Adam / 8e-4	Adam / 8e-4
Loss Function Type	Binary Crossentropy	Binary Crossentropy	Binary Crossentropy	Binary Crossentropy
Multi-GPU	4	4	4	4
Training Batch Size	50	50	8	8
Epochs	200	200	200	200
Testing Method	6-Fold Cross Validation	6-Fold Cross Validation	6-Fold Cross Validation	6-Fold Cross Validation