A Novel Approach to Segment and Classify Regional Lymph Nodes on Computed Tomography Images

Figure 1

Experiments of the revised nonlinear diffusion on noise removal and edge preservation. The green arrows point to lymph nodes. (a) A real CT image, (b)–(f) smoothing results after the Perona-Malik method, TV, median filtering, bilateral filtering, and our method, respectively. The color map is used for visual comparison. (g)–(l) Edges obtained by applying the Canny operator on (a)–(f), respectively. We note that the classical methods leave “undesirable clusters” in (b)–(e), reducing the accuracy of edge detection. The proposed method produces a clean result while preserving important edge information; for example, node 6 is well preserved without any cluster around it in (f) and (l) after applying the proposed method.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)

(k)

(l)