. Author manuscript; available in PMC: 2020 Jul 10.

Published in final edited form as: IEEE Access. 2020 May 29;8:101550–101568. doi: 10.1109/access.2020.2998537

TABLE IV.

Quantitative comparison of DCN-Net with Self-Training strategies in dentate and interposed nuclei segmentation.

Target	Dentate				Interposed

Metric	CMD (mm)	MSD (mm²)	DC	Volume (mm³)	CMD (mm)	MSD (mm²)	DC	Volume (mm³)
Metric	Manual labels	0.514±0.35	DC	0.380±0.11	0.873±0.05	736±165 (668±159)	DC	1.085±0.92	0.514±0.35	0.682±0.16	53±18 (53±24)
Self-training strategy #1 (pre-training)	0.613±0.46	0.373±0.12	0.872±0.05	713±155 (668±159)	1.029±0.76	0.486±0.22	0.676±0.15	52±16 (53±24)
Self-training strategy #1 (fine-tune)	0.529±0.42	0.381±0.12	0.874±0.05	731±165 (668±159)	1.003±0.78	0.424±0.24	0.686±0.14	67±24 (53±24)
Self-training strategy #2 (model distillation)	0.552±0.44	0.390±0.13	0.868±0.05	778±185 (668±159)	1.005±0.94	0.468±0.43	0.701±0.16	60±23 (53±24)

Bold indicates p<0.05 for paired t-tests with DCN-Net (trained on manual labels). () is ground truth volume.