notesum.ai

Published at December 4

Is Foreground Prototype Sufficient? Few-Shot Medical Image Segmentation with Background-Fused Prototype

cs.CV

Released Date: December 4, 2024

Authors: Song Tang¹, Chunxiao Zu, Wenxin Su, Yuan Dong, Mao Ye, Yan Gan, Xiatian Zhu

Aff.: ¹University of Shanghai for Science and Technology

Arxiv: http://arxiv.org/pdf/2412.02983v1

Refer to caption

Setting	Method	ABD-MRI					ABD-CT
Setting	Method	Liver	R.kidney	L.kidney	Spleen	Mean	Liver	R.kidney	L.kidney	Spleen	Mean
Setting-1	PANet [29]	47.37	30.41	34.96	27.73	35.11	60.86	50.42	56.52	55.72	57.88
	ADNet [9]	76.79	84.21	62.97	49.74	68.42	77.26	32.86	31.27	41.17	45.67
	RPTNet [33]	60.32	86.83	65.58	73.72	71.61	64.51	60.52	84.37	68.48	69.47
	Q-Net [24]	72.47	86.40	72.13	76.24	76.81	68.65	55.63	69.39	56.82	62.63
	CAT-Net [14]	70.59	83.00	75.30	70.54	74.86	66.24	47.83	69.09	66.98	62.54
	GMRD [4]	73.65	89.95	75.97	65.44	76.25	63.06	62.27	79.92	56.48	65.43
	SSL-ALPNet [21]	74.32	84.88	79.61	67.78	76.65	67.29	72.62	76.35	70.11	71.59
	SSL-ALPNet+Bro	74.30	87.06	83.49	68.13	78.25	71.01	79.07	77.91	71.71	74.93
Setting-2	PANet [29]	69.37	66.94	63.17	61.25	65.68	61.71	34.69	37.58	43.73	44.42
	ADNet [9]	77.03	59.64	56.68	59.44	63.19	70.63	48.41	40.52	50.97	52.63
	RPTNet [33]	67.45	60.11	66.27	75.15	67.25	54.24	53.84	82.28	60.12	62.62
	Q-Net [24]	71.52	74.71	64.15	74.71	71.27	64.44	41.75	66.21	37.87	52.57
	CAT-Net [14]	77.45	60.23	78.57	60.23	69.12	52.53	46.87	65.01	46.73	52.79
	GMRD [4]	74.85	70.25	69.37	73.80	72.07	60.88	55.35	72.46	64.16	63.21
	SSL-ALPNet [21]	68.38	76.38	73.24	55.35	68.34	69.14	59.05	64.18	61.97	63.59
	SSL-ALPNet+Bro	69.55	81.94	81.40	61.31	73.55	60.81	66.81	65.49	67.87	65.24