Molecular Image Analysis: Quantitative Description and Classification of the Nuclear Lamina in Human Mesenchymal Stem Cells

Table 1

Confusion matrix for classification of the two-cell populations using two-feature classifications and three-feature classifications. The two-feature classifier uses normalized curvature and average normalized intensity (Figure 7(a)) with a linear classifier and a quadratic classifier. The three-feature classifier uses normalized curvature, average normalized intensity and skewness with a linear classifier and a quadratic classifier. The percentage classification is given together with the standard error of its estimate (𝜇±𝜎).

		Biological sample
		Passage 4 cells	FKC8 cells
		𝑁=28, 100%	𝑁=21, 100%
Statistical classification (2 features)	Linear classifier
	passage 4	27,𝟗𝟔%±4%	3,14%±8%
	FKC8	1,4%±4%	18,𝟖𝟔%±8%
	Quadratic classifier
	passage 4	26,𝟗𝟑%±5%	3,14%±8%
	FKC8	2,7%±5%	18,𝟖𝟔%±8%
Statistical classification (3 features)	Linear classifier
	passage 4	27,𝟗𝟔%±4%	5,24%±9%
	FKC8	1,4%±4%	16,𝟕𝟔%±9%
	Quadratic classifier
	passage 4	25,𝟖𝟗%±6%	1,5%±5%
	FKC8	3,11%±6%	20,𝟗𝟓%±5%