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

Algorithm 2

Forward selection.

for each sample {𝑥_𝑖,𝑦_𝑖}∈𝑅^𝑀×{−1,1},𝑖=1,…,𝑛𝐝𝐨

  Feature subset construction 𝑆={𝑥_𝑘,𝑙∣𝑙≠𝑖,𝑖−1,…,1} by including the first 𝑖th feature

Train 𝑖th SVM using 𝑆;

Compute the lost function of Loss(𝑖)−Loss(𝑖−1) using (9);

𝐞𝐧𝐝𝐟𝐨𝐫

Maximize the energy function of 𝑝=max_𝑖(Loss(𝑖)−Loss(𝑖−1)) to find a feature subset 𝑆^∗.