Breast Cancer Characterization Based on Image Classification of Tissue Sections Visualized under Low Magnification

Table 1

Partial and overall classification accuracies of individual classifiers and leave-one-out method.

	Accuracies (%)				Best features; combination
	Grade I	Grade II	Grade III	Overall
kNN	100	90	100	96.9	SRE^a, GLNU^a, RLNU^a, and RLNU^r
PNN	100	95	92	95.4	SRE^a, GLNU^a, and RLNU^a
SVM	100	95	96	96.9	SRE^a, RLNU^a

SRE: short run emphasis; GLNU: gray level nonuniformity; RLNU: run length nonuniformity; a: average; r: range.