notesum.ai

Published at November 20

GraphCL: Graph-based Clustering for Semi-Supervised Medical Image Segmentation

cs.CV

cs.AI

Released Date: November 20, 2024

Authors: Mengzhu Wang¹, Jiao Li², Houcheng Su, Nan Yin³, Shen Li

Aff.: ¹Hebei University of Technology; ²University of Electronic Science and Technology of China; ³Zayed University of Artificial Intelligence, United Arab Emirates

Arxiv: http://arxiv.org/abs/2411.13147v1

Method	Scans used		Metrics
Method	Labeled	Unlabeled	Dice $\uparrow$	Jaccard $\uparrow$	95HD $\downarrow$	ASD $\downarrow$
V-Net [27]	4(5%)	0	52.55	39.69	47.05	9.87
V-Net [27]	8(10%)	0	82.74	71.72	13.35	3.26
V-Net [27]	80(All)	0	91.47	84.36	5.48	1.51
UA-MT [51]			82.26	70.98	13.71	3.82
SASSNet [18]			81.60	69.63	16.16	3.58
DTC [22]			81.25	69.33	14.90	3.99
URPC [23]	4(5%)	76(95%)	82.48	71.35	14.65	3.65
MC-Net [45]	4(5%)	76(95%)	83.59	72.36	14.07	2.70
SS-Net [47]			86.33	76.15	9.97	2.31
BCP [2]			87.07	77.42	8.83	2.15
GraphCL			88.80 $\uparrow^{\textbf{1.73}}$	80.00 $\uparrow^{\textbf{2.58}}$	7.16 $\downarrow^{\textbf{1.67}}$	2.10 $\downarrow^{\textbf{0.05}}$
UA-MT [51]			87.79	78.39	8.68	2.12
SASSNet [18]			87.54	78.05	9.84	2.59
DTC [22]			87.51	78.17	8.23	2.36
URPC [23]	8(10%)	72(90%)	86.92	77.03	11.13	2.28
MC-Net [45]	8(10%)	72(90%)	87.62	78.25	10.03	1.82
SS-Net [47]			88.55	79.62	7.49	1.90
BCP [2]			89.39	80.92	7.26	1.76
GraphCL			90.24 $\uparrow^{\textbf{0.85}}$	82.31 $\uparrow^{\textbf{1.39}}$	6.42 $\downarrow^{\textbf{0.84}}$	1.71 $\downarrow^{\textbf{0.05}}$