Segmenting Brain Tissues from Chinese Visible Human Dataset by Deep-Learned Features with Stacked Autoencoder

Table 1

Details of the SAE architectures with different input patch size and layer depth in this study.

Patch size		Layer 1	Layer 2	Layer 3	Layer 4
	Layer type	AE	Softmax	—	—
	Input size	50	25	—	—
	Hidden size	25	—	—	—
	Layer type	AE	AE	Softmax	—
	Input size	162	81	36	—
	Hidden size	81	36	—	—
	Layer type	AE	AE	Softmax	—
	Input size	338	150	64	—
	Hidden size	150	64	—	—
	Layer type	AE	AE	AE	Softmax
	Input size	578	400	200	100
	Hidden size	400	200	100	—
	Layer type	AE	AE	AE	Softmax
	Input size	882	400	200	100
	Hidden size	400	200	100	—