Brain Tumor Segmentation Based on Hybrid Clustering and Morphological Operations

Table 3

Comparison of four clustering algorithms and the proposed algorithm.

Clustering methods	Evaluations	Patient 1	Patient 2	Patient 3
K-means	Dice	0.9001	0.9316	0.8298
	Sensitivity	0.9630	0.9064	0.9424
	Specificity	0.9952	0.9984	0.9780
	Recall	0.8449	0.9583	0.7412
FCM	Dice	0.9137	0.9341	0.9004
	Sensitivity	0.9263	0.8905	0.9336
	Specificity	0.9971	0.9994	0.9880
	Recall	0.9015	0.9823	0.8694
sFCM	Dice	0.8169	0.9258	0.9144
	Sensitivity	0.7112	0.8645	0.9290
	Specificity	0.9968	0.9994	0.9878
	Recall	0.9597	0.9965	0.9002
csFCM	Dice	0.8069	0.9179	0.9166
	Sensitivity	0.6960	0.8540	0.9290
	Specificity	0.9971	0.9996	0.9880
	Recall	0.9597	0.9922	0.9045
Proposed	Dice	0.9261	0.9400	0.8978
	Sensitivity	0.9622	0.9184	0.9380
	Specificity	0.9971	0.9996	0.9881
	Recall	0.8926	0.9625	0.8608