Table 1.

The architecture of the proposed 3D-CNN model.

Model	Type	Filter Size	Number of Filters	Stride
Layer 1	Conv1 + Maximum Pooling	3 × 3 × 3	16	(1, 1, 1)
Layer 2	Conv2 + Maximum Pooling	3 × 3 × 3	32	(2, 2, 2)
Layer 3	Conv3	3 × 3 × 3	64	(1, 1, 1)
Layer 4	Conv4 + Maximum Pooling	3 × 3 × 3	64	(2, 2, 2)
Layer 5	Conv6	3 × 3 × 3	96	(1, 1, 1)
Layer 6	Conv6 + Maximum Pooling	3 × 3 × 3	96	(2, 2, 2)
Layer 7	Conv7	3 × 3 × 3	128	(1, 1, 1)
Layer 8	Conv8 + Maximum Pooling	3 × 3 × 3	128	(2, 2, 2)
Layer 9	FC1	-	-	-
Layer 10	FC2	-	-	-
Layer 11	FC3 (SoftMax)	-	-	-

Conv—convolutional layer; FC—fully connected layer.