An unsupervised strategy for biomedical image segmentation

Figures & data

Figure 1 (A) Original image and (B) unsupervised segmentation using our algorithm. The arrows mark the isolated lesions.

Figure 2 (A) Original image and (B) segmentation using our unsupervised strategy. The arrows indicate isolated lesions. The split arrow indicates a zone which is not a lesion.

Figure 3 (A) Original image. (B) Segmentation by using our unsupervised strategy. The arrows indicate the isolated lesions. Note the quality of segmentation of the Type lesion in (B).

Figure 4 (A) Original image and (B) segmentation using our unsupervised strategy. The arrows indicate the isolated lesions. The split arrows indicate zones which are not lesions.

Figure 5 (A) Original image and (B) segmentation using our unsupervised strategy. The arrow indicates the isolated lesion.

Table 1 Numerical results of the validation

Images	Vp	fp	fn	FN	FP
Figures 1a and 1b	2	1	0	0%	0%
Figures 2a and 2b	8	1	0	0%	12.5%
Figures 3a and 3b	2	0	0	0%	0%
Figures 4a and 4b	10	2	0	0%	20%
Figures 5a and 5b	1	0	0	0%	0%

Abbreviations: FN, false negatives; FP, false positives.