Detection of Pulmonary Nodules in CT Images Based on Fuzzy Integrated Active Contour Model and Hybrid Parametric Mixture Model

Table 1

Features calculation of every potential nodule object.

	Features	Definition
Intensity	Mean value 𝐺𝑚	𝐺𝑚=∑^𝑀𝑥=1∑^𝑁𝑦=1𝐼(𝑥,𝑦)/(𝑀×𝑁), in 2D space
	Standard deviation 𝐺𝑢	𝐺𝑢=∑^𝑀𝑥=1∑^𝑁𝑦=1(𝐼(𝑥,𝑦)−𝐺𝑚)^2, in 2D space
	Mean 𝐺𝑚3D of intensity	In 3D space
	Standard deviation 𝐺𝑢3D of intensity	In 3D space
Position	Centroid (𝑖0,𝑗0)	𝑖0=𝑚10/𝑚00, 𝑗0=𝑚01/𝑚00
Shape	Area 𝐴	𝐴=∑(𝑓(𝑥,𝑦)=1), in 2D space
	Perimeter 𝐿	𝐿=𝑛𝑜+√2𝑛𝑒, in 2D space
	Diameter 𝑎	Long axis in 2D space,; 𝑎=2×[2(𝜇20+𝜇02−(𝜇20−𝜇02)^2+4𝜇^211)]/𝜇00^1/2
	Ellipticity 𝑒	𝑒=𝑎/𝑏, 𝑎,𝑏 are the long axis and short axis, respectively, in 2D space
	Circularity 𝐶	𝐶=𝑟𝑖/𝑟𝑐, in 2D space
	Slenderness 𝑆	𝑆=min(𝑊,𝐻)/max(𝑊,𝐻), in 2D space
	Rectangle degree 𝑅	𝑅=𝐴/(𝑊×𝐻), in 2D space
	Compactness 𝐹	𝐹=4𝜋𝐴/𝑃^2, 𝑃 is the perimeter of region contour. 𝐴 is the area in 2D space; Compactness ratio reflects degree closing to circle and smoothness of region. And 𝐹 can be used to detect the rough feature of edge
	Concavity ratio 𝐸	𝐸=𝑆𝑒/𝑆, 𝑆 is area of concave region, 𝑆𝑒 is difference of convex hull and original region, and 𝐸 reflects cupped degree of boundary
	Volume	In 3D space
	Volumetric quotient	In 3D space
	The long axis of the circumsphere	In 3D space
Texture	Energy, contrast, entropy, and adverse moment	In 2D space