Multimodal MRI Brain Tumor Image Segmentation Using Sparse Subspace Clustering Algorithm

Table 5

Comparison of multimodal image segmentation results with 5% noise.

Experimental sample		Index
		Dice	Jaccard	Precision	Recall
Malignant tumor	1	0.8123	0.7758	0.9036	0.8080
	2	0.8016	0.7469	0.9229	0.7568
	3	0.8001	0.7154	0.9086	0.7735
	4	0.8462	0.8369	0.9321	0.8858
	5	0.8528	0.7427	0.9650	0.7804
	6	0.8134	0.7528	0.9487	0.7940
	7	0.8347	0.8274	0.8936	0.9166
	8	0.8006	0.7144	0.9376	0.7130
	9	0.8104	0.7423	0.9721	0.7668
	10	0.8110	0.7417	0.9325	0.7878
	11	0.8234	0.7326	0.9639	0.7626
	12	0.6841	0.5146	0.9688	0.5229
	13	0.7695	0.6155	0.9129	0.7007
	14	0.7996	0.6639	0.9639	0.7013
	15	0.8005	0.6582	0.9470	0.6982
Bright tumor	1	0.7486	0.6301	0.9003	0.6663
	2	0.7952	0.8204	0.9575	0.8452
	3	0.8985	0.6471	0.9588	0.6620
	4	0.8625	0.6598	0.7176	0.9031
	5	0.8563	0.8446	0.9425	0.8206
	6	0.9012	0.8396	0.9393	0.8759
	7	0.6852	0.5329	0.9579	0.5414
	8	0.7410	0.6012	0.9493	0.6225
	9	0.8863	0.8071	0.9389	0.8223
	10	0.8401	0.7540	0.9522	0.7639
Mean		0.8110	0.7167	0.9315	0.7557